Skip to main content

Full text of "MODIS: Moderate-resolution imaging spectrometer. Earth observing system, volume 2B"

See other formats




EARTH OBSERVING SYSTEM 
Volume Db 




INSTRUMENT PANEL REPORT 



IVIASA 

National Aeronautics and 
Space Admini<;?ration 



1986 



® 



EARTH C»SERVING SYSTEM REPORTS 



Vi>lumc I Science and Misskm Requirements Working (irtnip Report 

Volume II From Pattern to Pri»ccss: The Strategy *>f the Earth Observing System 

Science Steering C\)mmittee Report 



[^' 



\ 



Volume I la 



Volume Ilh 



Data and Information System 
I>ata Panel Rep4>rt 

MODIS 

MtHlerate-RcMilution Imaging Spectrometer 
Instrunwnt Panel Report 



Volume lie HIRIS & SISEX 

High-Resolution Imaging Spectrometry: Science Opportunities Un the IW(K 
Instrument Panel Report 

Vi>lume lid LAS A 

Lidar Atmospheric Sounder and Altimeter 
Instrument Panel Report 

Volume lie HMMR 

lligh-RcM>lution Multifrequency Microwave Rudi<*meter 
Instrument Panel Report 

Volume iir SAR 

Synthetic Aperture Radar 
Instrument Panel Report 

Volume Ilg LAWS 

Laser Atmospheric Wind Stiunder 
Instrument Panel Report 



Volume Ilh 



Altimelric System 
Panel Rept>rt 




MODERATE.RESOLUTION IMAGING SPECTROMETER INSTRUMENT PANEL 
FOR THE EARTH OBSERVING SYSTEM 



Wayne Esaias, Chairman 

William Barnes. Executive Secretary 

Mark AblH>tt 

Steve Cox 

Ri>bert Evans 

Ri>bert Eraser 

Alexander Cj<H:tz 

Christopher Justice 

E:. Paul McClain 

Marvin Maxwell 

Robert Murphy 

Joseph Pn>spero 

Barrett R<Kk 

Steven Running 

Raymond Smith 

Jerry Soli>mon 

Jivl Susskind 



Howard Ciordon. Exofficio Member 
Michael Spanner, Exofficio Member 



in 



k> 



f*4>-f^ 



COVER PHOIOGRAPH: 



GRiairiAL PAQE IS 



'^1 




The figure abiwe and on the Ciwer sh<iws th*? average plant abundance and distribution for a fifth of the Earth's surface 
in May. Different scales are used for land and water. For land areas, the Vegetation Index is derived from the NOAA-6 
Advanced Very High Resolution Radiometer (AVHRR) for May 1982 (dark green is the highest density of green 
vegetation - right color bar). For (Keans and lakes the amount of phytoplankton in terms of chlorophyll pigment density 
(red is highest - left color bar) is derived from the C oastal Zone Color Scanner (CZCS) >n NASA's Nimbus-7 for May 
1979. 

Ihc Sahara IX»scrt. tropical rainforests, and spring giccning oi temperate forests and fields are evident on land. The 
corn and wheat bell south of the (ireat lakes is less green bccuuse crops are just beginning lo grow. In the tKcans. 
productive upwelling areas along the coast (especially off N.W Africa) and the "spring bloom* in the North Sea and 
northern Ntirth Atlantic arc very evident. C'/C'S data do not distinguish ixtwccn high sediment and pigments, and values 
in coastal and lake waters can be ambiguous. 

this composite includes all daylight data collected by the two sensors during the periods in this region ( MfS to Hfl^N. 
and !(>' F lo MMT W). For the land, this is but a part of a multiyear gU»bal data set. For the iKcans. this is the first view of 
biological activity on a basin scale. Black areas in the tKcan indicate no observations, white indicates that clouds or sea 
ice was present during everv obsiTvation. Striping also inoicates undersampling of tKcanic variability with the limited 
duty C/CS. 

This image illustrates current sensor capability, aid the need for M()I)!S. The AVHRR is a weather sensiu not 
optimized Un vegetation sensing, and the C/CS collected limited data as a proof -i>f-conccpt mission. MODIS will provide 
comprehensive and simultaneous 4>bservatit>ns of variable land, ocean, and atmosphere properties on a global basis. 
Belter spectral rcs4>lulion and lcmp*»ral coverage wi'l pr<»vide less an'biguous and more accurate data on variations of 
plant abundance, and many additional properties essential for understanding and quantifying global change. 

(i Fcldman and C. Uuker. (JSFC 



IV 



» 



EXECUTIVE SUMMARY 



Over the last few years there has developed 
within the scientific community the conviction that 
the Earth must be treated as a single interacting 
ecosystem if major advances in our knowledge of 
the Earth and of man's impact are to be achieved. 
This has resulted in numerous integrated research 
programs having as a common theme the need for 
long-term global data bases to establish the Earth's 
present state and to delineate trends. These data 
bases are vital for the initialization and testing of r 
variety of models currently under development. The 
natural variations are such that the Earth's surface 
must be viewed every few days in order that dynamic 
events can be observed, and a minimum l()-year 
data base is required if we arc to separate long-term 
trends from short-term natural variations. 

In the spring of 1983 the Earth Observing Sys- 
tem (Eos) Science and Mission Requirements 
Working Group was formed by NASA (National 
Aeronautics and Space Administration) with rep- 
resentatives from the various disciplines of Earth 
science to define critical questions for the I99()s and 
to delineate low Earth orbit observables that would 
materially address these questions. Eos has since 
become the anticipated payload(s) for the Space Sta- 
tion polar platforms. The results of this group's de- 
liberations included requirements for a multispec- 
tral radiometer capable of frequent global surveys 
at a 1 kilometer spatial resolution for terrestrial, 
cKcanic, and atmospheric properties. This system 
was designated the Moderate-Resolution Imaging 
Spectrometer (MODIS). The MODIS Instrument 
Panel was formed in mid-19H4 to further define the 
scientific requirements and generate a set of sensor 
parameters that would ensure achievement of the 
scientific goals. The Panel was composed of repre- 
sentatives from the land, cKcan, and atmospheric 
scientific communities, including representatives 
from NOAA (National Oceanic and Atmospheric 
Administration) and NASA. This dwumcnt is a 
compilation of the recommendations of the MODIS 
Instrument Panel. 

Terrestrial studies amenable to being addressed 
via low-resolution imaging radiometry include rate 
of tropical deforestation and type and rale of re- 
growth; arcal distribution and effect of acid rain on 
the boreal forests of Europe and North America; 
rate and extent of desertification at the edge of the 
world's deserts; update of global vegetation maps; 
extent of freeze or drought damage in croplands and 
natural communities; land cover change and its ef- 
fect on terrestrial biophysical systems; continental 
changes in snow cover with associated changes in 
albedo; and derived products including standing 
green biomass. intercepted photosynthetically ac- 
tive radiation, and net primary productivity. Com- 
puter simulation models that will depend on data 



from a sensor of this type include global surface 
climate models, carbon cycle models, hydrologic 
cycle/energy models, and comprehensive biogeo- 
chemical cycling models. 

Ocearo. rrnhers will utilize MODIS-visible data 
to charac the global distribution of phyto- 

plankton bu xss and its temporal and spatial vari- 
ability. Using in situ data together with regional 
scale models, it is possible to convert biomass to 
primary productivity. Thus oceanic and global meas- 
urements of primary productivity and their tem- 
poral variation will be possible, iliermal infrared 
(TIR) data will be used to study variability in sea 
surface temperature (SST) as related to physical 
processes on the climatological and physical dy- 
namic scales. In addition, mesoscale ocean circula- 
tion features, such as warm and cold core rings and 
jets, can be observed and their development fol- 
lowed for months or years using ocean color and 
thermal infrared data. The effects of riverine fluxes 
and other forms of sediment transport can be ob- 
served and quantified, demonstrating their impor- 
tance to <x:ean prtxluctivity and global biogeochem- 
ical cycles. Important phytoplankton subgroups, 
such as cyanobacteria and coccolithophores, will be 
identifiable. 

Atmospheric constituents that will be moni- 
tored directly by MODIS include global distribu- 
tions of clouds and aerosols, both of which have a 
direct impact on climate as well as on the geochem- 
ical and hydrological cycles. 

Sensor requirements generated by the terres- 
trial studies subgroup include a spatial resolution of 
5(K) meters based on a need for adequate discrimi- 
nation of agricultural and forestry features, experi- 
ence with National Oceanic and Atmospheric 
Administration/Advanced Verv High Resolution 
Radiometer (NOAA/AVHRR)' 1 kilometer data, 
and practical data volumes. Required spectral chan- 
nels are essentially the same as those for Thematic 
Mapper with the possible addition of one or more 
channels in both the near infrared and shortwave 
infrared. Since thermal data is potentially quite use- 
ful in studies of soil moisture and evapotranspira- 
tion, a total of seven bands in the 3.5 to 4.0 and 8.5 
to 12.0 micrometer window regions is also recom- 
mended. There is also a requirement for viewing 
fore and aft of nadir at several angles up to (if in 
order to study the bidirectional reflectance distri- 
bution function (BRDF) of plant canopies. It was 
determined that an equatorial crossing lime be- 
tween 1 m and 2:(KI p.m. was acceptable and that a 
revisit timo of two or three days was sufficient to 
measure vegetation dynamics. 

Requirements generated by the tKcanographer 
members of the MODIS Panel are based on expe- 
rience with the Nimbus-7 Coastal Zone Color 



,0 




■mx.m' 



Scanner (CZCS) and NOAA/AVHRR data. This 
has resulted in a need for at least 17 spectral bands 
in the wavelength region from 0.4 to 1.0 micrometer 
with the visible bands having bandwidths less th.an 
20 nanometers and signal-to-noise in excess of 
6(K):1. The near-infrared bandwidth and signal-to- 
noise requirements are less restrictive since those 
channels are principally used for aerosol correc- 
tions. The ocean color requirement having the 
greatest impact on the system is the requirement to 
point up to 20° fore or aft of nadir to avoid specular 
reflection (sunglint) from the ocean surface. An- 
other ocean color requirement is the need for peri- 
odic solar-referenced calibrations. This is unique to 
cKcans but will benefit all other disciplines. Addi- 
tional ocean color requirements include one or more 
polarized channels in the visible to enhance atmos- 
pheric corrections; an equatorial crossing time as 
near mxjn as possible; spatial resolution of I kilo- 
meter in coastal regions with acceptable averaging 
to 4 kilometers in open oceans; minimal sensitivity 
to incoming polarized radiance at all viewing angles 
(with the exception of the pi>larized channels); a 
continuous, 10-ycar global data base; and two-day 
revisit. 

Ocean thermal infrared requirements are aimed 
at obtaining SST retrievals accurate to within 
±0.5 K. This results in a need for "split" window 
channels in the 3.5 to 4.0 and 10.5 to 12.0 microm- 
eter spectral regions. 

Atmospheric science sensor requirements for 
MODIS are driven to a large extent by the terrestrial 
and (Kcanic requirements. The most demanoing is 
a set of narrow spectral channels liKated in the ox- 
ygen A-band near 0.76 micrometer to measure 
cloud altitude. Other channels required for atmos- 
pheric and ice observations arc readily accommo- 
dated within the constraints established by the ter- 
restrial and oceanic requirements. 

The MODIS requirements for (a) a view of 2{f 
to Mf fore and aft of nadir for iKcan observations 
and land bidirectional reflectance studies, (b) un- 
interrupted long-term global cKcan surveys, and (c) 
minimum atmospheric path radiance fot routine ter- 
restrial sensing (i.e., nadir view) are incompatible 
\vith a single sensor package. Therefore, it was pro- 
pi>scd that the optical component of the MODIS 
system be divided into two packages to be desig- 
nated MODIS-T (tilt) and MODIS-N (nadir); the 
former containing the visible and near-infrared 
channels requiring fore or aft of nadir viewing, and 
the latter containing those channels with no require- 
ment for off-nadir pointing, including all of the in- 
frared (IR) channels requiring cooled detectors. 
Conceptual systems for the two sensors are dis- 
cussed below. 

The MODIS-T component will address those 
requirements that call for viewing the surface at pre- 
determined angles fore and aft the subsatellite point 
(nadir). These include: (a) minimizing the amount 



of specular reflectance from the surface, (b) exam- 
ining the BRDF from large homogeneous targets, 
and (c) performing atmospheric studies by examin- 
ing the spectral signal as a function of optical depth. 
The MODIS-T requirements can be satisfied by any 
of several types of imaging radiometers. A practical 
system in terms of size, complexity, available tech- 
nology, and overall utility is that of the imaging 
spectrometer. The version discussed in this report 
includes a crosstrack scan motor, collecting optics, 
spectrometer, and a 64 x 64 element silicon detec- 
tor array. The optical aperture is on the order of 5 
centimeters. This results in a very compact system 
capable of ±60° rotation about the optical axis to 
give the required fore-aft tilt. The 1,500 kilometer 
(90°) swath is scanned in 9.5 seconds, the time re- 
quired for the subsatellite point to advance 64 kil- 
ometers. The image of the spectrometer slit on the 
surface consists of 64 pixels along-track with each 
pixel being dispersed within the sensor into 64 per- 
fectly registered contiguous spectral bands of ap- 
proximately 10 nanometers width covering the range 
from 400 to I, (KM) nanometers. 

The MODIS-N (nadir) component will address 
those scientific tasks that do not require off-nadir 
pointing. The requirements cover the spectral range 
from 0.4 to 12.0 micrometers and include a require- 
ment for 5(K)-meter resolution in 12 channels in the 
region from 0.4 to 2.5 micrometers. Spectral width 
requirements vary from 1.2 to 500 nanometers. 
These requirements result in a system with at least 
35 spectral bands. Due to the range of spectral 
widths, a requirement to measure polarization, and 
cooling requirements, it is impractical to use the 
imaging spectrometer concept of MODIS-T. In- 
stead, a more conventional imaging radiometer con- 
cept is considered. This system consists of a cross- 
track scan mirror and collecting optics and a 
dichroic filter that divides the incoming energy onto 
a cryogenically cooled focal plane ( 18 channels, 1 to 
12 micrometers) and an ambient temperature focal 
plane (18 channels. 0,4 to 1.0 micrometer). The fo- 
cal planes each have 36 detector elements. The re- 
quired optics aperture diameter is 40 centimeters. 

Data from both optical components will be of 
interest to many, if not all, users and should be con- 
sidered as a single data set. The data rates from the 
instrument are about 8.8 megabits per second during 
the daylight and 1.2 megabits per second during night. 
Some redundancy of channels in the 400 to 1,(XM) 
nanometer range will enable complete spatial cover- 
ape during fore-aft tilt operations, which is especially 
useful along coastal lines for both terrestrial and 
(Kcanic applications and will aid in the required inter- 
calibration of the T and N components. Accurate 
(< ± 1 percent relative to the solar spectrum) cali- 
brations are essential, and several scenarios utilizing 
Shuttle servicing capabilities are envisioned to moni- 
tor changes in instrument spectral sensitivity iwer the 
mission life of more than ten years. 



VI 



^ -rr**^ 



S) 



i . 



In summary, MODIS, as presently conceived, is 
a system of two imaging spectroradiometer compo- 
nents designed for the widest possible applicability 
to research tasks that require long-term (S to 10 
years), low-resolution (O.S to 1.0 kilometer), global, 
multispectral (52 channels between 0.4 and 12.0 



micrometers) data sets. The system described is pre- 
liminary and subject to scientific and technological 
review and modification, and it is anticipated that 
both will occur prior to selection of a final system 
configuration; however, the basic concept outlined 
above is likely to remain unchanged. 



1 



i 

»; 



i 



vti 










CONTENTS 

Plage 

EXECUTIVE SUMMARY 

LIST OF TABLES 

LIST OF FIGURES 

ACRONYMS AND ABBREVIATIONS - 

I. INTRODUCTION j 

II. MODIS OBJECTIVES , 

Eos Mission Objectives j 

MODIS Complements Operational Systems •■.. i '!! ^ ""!!!."!!.!.!...! ! 7 

III. SCIENCE OBJECTIVES AND INSTRUMENT REQUIREMENTS 4 

Terrestrial Studies , 

Oceanographic Studies 17 

Atmosphere Studies .............'. -54 

Snow and Ice Research 14 

Operational Needs \< 

IV THE MODIS SENSOR SYSTEM ,. 

Background \, 

MODIS-T Z 

MODis-N "^^'^!!!'!!'!!!!."^!. "!!!!!!! !!;;.' '^7 

Calibration \^. 

Data Rates 4/ 1 

V MISSION OPERATIONS REQUIREMENTS ai 

Tilts .'.■.■.■.■.■.■.■:.■.■.'.■.■:.■.'.■.■.■.'.■.■.■:.■.:.■.■.■ A 

Gains ' 

Onboard Processing 4^ 

Calibration ,'^ 

Operations /^ 

VI. GROUND SYSTEM PROCESSING AND ARCHIVING REQUIREMENTS 44 

Overview , . 

Levels of Data Processing .[[...., 44 

MODIS Archival and Distribution Requirements 44 

VIL MODIS/HIRIS SYNERGISM 4^ 

Dynamic Phenomena ***.**. 47 

Context and Pixel Structure [^.\..., ....,..] 47 

Signature Extension and Spatial Extrapolation 47 

Atmosphere ^^ 

APPENDIX A: ATMOSPHERIC CORRECTIONS OVER LAND 4y 

APPENDIX B: ATMOSPHERIC CORRECTIONS OVER OCEANS 51 

APPENDIX C: MODIS INSTRUMENT PANEL STATEMENT OF WORK 54 

REFERENCES "^^ 



^1 



IX 



.® 






7 . 



trt 



LIST OF TABLES 
larae 

Page 

1 Characteristic arj Status of the NOAA/AVHRR Systems 

2 Proposed AVHRR Characteristics on NOAA-K.L.M ^ 

3 Spectral Regions for Remote Sensing of Vegetation ^ 

^ RlfrSvl^.'.'".^''!"'^^'''''^'*^"^ *^ 

5 .Proposed MODIS Spectra! Bands and Priority for the Oceans ^^ 

in the Visibly and Near-Infrared Regions .... . . _ 

* t^i^T^M^f'^*''^''™^"^'*^' Sea Surface Temperature F^ '^ 

the World Climate Research Program ,., ... 

7 AerosolTypes ^^ 

8 VariablesRequircdft>rMODIS Atmospheric Applications. .'.'.".".".'.' J^ 

9 Spectral Channels for Detailed Observations of Clouds 

10 Spectral Channels for Editing Cloud or Aerosol Pixels 

11 Specification of Cloud Climatology Requirements ... ^^ 

12 Summary of Passive Techniques to Determine Cloud Physical Parameters " .' ." 3! 

13 MODIS-T Parameters for Sensitivity Calculations 

14 MODIS-T Twenty-Five of Sixty-Four Spectral Bands ".' f 

15 MODIS-N Example of Performance Calculations (Channel 25) ,, 

16 MODIS-N Visible/Near IR Channels (Preliminary) 

17 MODIS-N Thermal Channel S/N Calculation (Channel 35) f 

18 MODIS-N Thermal Channels (Preliminary) 

19 Definition of MODIS Data Products Levels '^^ 

20 Compositing Scales '^^ 

21 MODIS Data Requirements-Expected Requests for Data ^^ 

B.l Reflectance for One CZCS Digital Count 

B.2 Typical Values of p„ p., and p ^ ' 

52 



i 

#* 



t.v<.* . 



i)i 




I_^l21!l^ 




LIST OF FIGURES 

Figure Page 

1 Global map of the Earth's mean monthly surface skin 

temperature for January 1979 6 

2 Organizational diagram of a proposed model of net primary 

production for a coniferous forest 7 

3 AVHRR global vegetation index for July, 1982 9 

4 Normalized difference vegetation index derived from AVHRR 

GAC data resampled to 8 kilometers 10 

5 Comparison of AVHRR local area coverage with MSS 

false color composites 11 

6 Reflectance vs. wavelength for a variety of soils 15 

7 Aitken nuclei concentrations in 10' cm ' 27 

8 Average haze frequency for June, July, August time period 28 

9 MODIS-T scan geometry and conceptual system layout 37 

10 Conceptual optical system for MODIS-T 38 

1 1 MODIS-N focal plan layout 40 



XI 



2); 




ACRONYMS AND ABBREVIATIONS 



AIS 

AMSU 

AMTS 

AO 

AVHRR 

BPI 

BRDF 

CCT 

CWR 

CZCS 

DMSP 

DOMSAT 

DOM 

EBB 

Eos 

ERS-1 

FOV 

GAC 

gC 

GIMMS 

GOES 

GOFS 

GSFC 

GVI 

HIRIS 

HIRS-2 

HMMR 

IFOV 

IPAR 

IR 

ISLSCP 

JPiyPODS 

LAC 

LAI 

LASA 

LED 

LFMR 

Mbs 

MIZ 

MLA 

MODIS 

MODIS-N 



Airborne Imaging Spectrometer 

Advanced Microwave Sounding Unit 

Advanced Moisture and Temperature Sounder 

Announcement of Opportunity 

Advanced Very High Resolution Radiometer 

Bits Per Inch 

Bidirectional Reflectance Distribution Function 

Computer Compatible Tapes 

Clear Water Reflectance 

Coastal Zone Color Scanner 

Defense Meteorological Satellite Program 

Domestic Communications Satellite 

Dissolved Organic Material 

Equivalent Black Body 

Earth Observing System 

European Space Agency Remote Sensing Satellite- 1 

Field-of-View 

Global Area Coverage 

Grams Carbon 

Global Inventory, Monitoring, and Modeling Studies 

Geostationary Operational Environmental Satellite 

Global Ocean Flux Study 

Goddard Space Flight Center 

Global Vegetation Index 

High-Resolution Imaging Spectrometer 

High-Resolution Infrared Radiometric Sounder Model-2 

High-Resolution Multifrequency Microwave Radiometer 

Instantaneous Ficld-of-View 

Intcrcepied Photosynthetically Active Radiation 

Infrared 

International Satellite Land Surface Climatology Project 

Jet Propulsion Laboratory/Pilot Ocean Data System 

Local Area Coverage 

Leaf Area Index 

Lidar Atmospheric Sounder and Altimeter 

Light Emitting Diode 

Low Frequency Microwave Radiometer 

Megabits Per Second 

Marginal Ice Zone 

Multispectral Linear Array 

Moderate-Resolution Imaging Spectrometer 

MODIS Nadir 



xi< 



41 

4i 










^ \.v^-sii 



ACRONYMS AND ABBREVIATIONS (continued) 



^ i 



5^.1 * 



MODIS-T MODIS Tilt 

MOS/LOS Marine Observation Satellite/Land Observation Satellite (Japan) 

MSS Multispectral Scanner 

MSU Microwave Sounding Unit 

N Nadir 

NASA National Aeronautics and Space Administration 

NESDIS National Environmental Satellite, Data and Information Service 

nm Nanometer = 10 " meter 

NOAA National Oceanic and Atmospheric Administration 

NOSS National Oceanic Satellite System 

NPP Net Primary Productivity 

NROSS Navy Remote Ocean Sensing System 

OCI Ocean Color Imager 

PBL Planetary Boundary Layer 

PEC Particulate Elemental Carbon 

RMS Root Mean Square 

ROS Research Optical Sensor 

SAR Synthetic Aperture Radar 

SISEX Shuttle Imaging Spectrometer Experiment 

R Scanning Multifrequency Microwave Radiometer 

vWG Science and Mission Requirements Working Group 

S/N Signal-to-Moise 

SPOT Systeme Probatoire dObservation dc la Terre 

SSM/I Special Sensor Microwave/Imager 

SST Sea Surface Temperature 

SSU Stratospheric Sounding Unit 

STS Space Transportation System 

SW Shortwave 

T Tilt 

TCSM Tropospheric Chemistry Systems Model 

TIMS Thermal Infrared Multispectral Scanner 

TIR Thermal Infrared 

TIROS TV Infrared Operational Satellite 

TM Thematic Mapper 

TOGA Tropical Oceans and Cjlobal Atmosphere 

TOMS Total Ozone Mapping Spectrometer 

|jLm Micrometer = 10 " meter 

VIS Visible 

VIS/NIR Visible and Near Infrared 

WCRP World CMifnate Research Program 

WOCE World Ocean Circulation Experiment 






xni 



^ 



JS 



mmmmmmmmmm 






I. INTRODUCTION 



In the spring of 1983. the Earth Observing Sys- 
tem (Eos) Science and Mission Requirements 
Workmg Group was formed by NASA with repre- 
sentatives from the various disciphnes of Earth sci- 
ence to define major questions for the 1990s and to 
delmeate low Earth orbit observables that would 
materially address these questions. Eos has since 
become the anticipated payload(s) for the polar 
platform portion of the Space Station. The results 
of this groups deliberations (Butler et al 1984) 
included requirements for a multispectral radiome- 
ter capable of frequent global surveys at a 1 km 
spatial resolution. This system was designated the 

mnm^fxr^'l'lJi?!! '™^«'"8 Spectrometer 
(MODIS). The MODIS Instrument Panel was 
formed in mid- 1984 to further define the scientific 
goals and observational requirements and generate 
a set of sensor parameters that would ensure 
achievement of these scientific goals. The MODIS 
statement of work is given in Appendix C. 

w^H'J*.'^*'*^"""-*"' '* "^ '■*^P"" ^'f ihe findings of the 
MODIS Instrument Panel. It includes a set of sci- 
entific objectives; ocean, atmosphere, terrestrial, 
and snow and ice research tasks requiring MODIS 
data; and a set of sensor requirements for each of 
these disciplines. These requirements have been 



combined and a preliminary sensor system that 
addresses most of the requirements has been gen- 
erated. Owing to the diversity of these require- 
ments. It has been necessary to divide MODIS into 
L'l'Xr^e'^,'' f«'='"'ges. Which have been designated 
MODIS-N (nadir) and MODIS-T (tilt), and to in- 
clude within the former several channels having S(X) 
m resolution. A description of the system, including 
calculated performance parameters, is given in 
Chapter IV. ^ 

It is assumed thai the Eos paytoad will include 
ttie High-Resoiution Imaging Spectrometer (HIRIS) 
which will consist of 192 spectral channels between 
0.4 and 2.5 jim with .K) in resolution over a 48 km 
'"'I'L'^lc "If' "'R'S will be on the same platform 
as MODIS. A discussion of the synergism that will 
ensue from this scenario is given in Chapter VII 

Additional topics covered include mission op- 
erations requirements (Chapter V). ground svslem 
r.'^^i'i"^ '"^"^ archiving requirements (Chapter 
VI). MODIS HIRIS unique opportunities for syn- 
ergism (Chapter VII). and algorithms (Appendices 

Mnm^ M '^i'"!:';!l^^'"'" "^*^ '''^^'^'^-A 'Studies of 
MODIS N and MODIS-T that were completed dur- 
mg the preparation of this document are not 
included. 



mmmmmm 



^^ 



T 



11. MODIS OBJECTIVES 



Eos MISSION OBJECTIVES 

The primary objective of the MODIS instru- 
ment is to provide a comprehensive series of global 
observations of the Eanh (land, oceans, and atmos- 
phere) in the visible and infrared regions, at suffi- 
cient spacial resolution to permit complete global 
coverage within a few days. The word 'comprehen- 
sive" has several implications. First, it refers to the 
spectra! as well as the continuous temporal coverage 
required to resolve the major frequencies of ob- 
served variability ranging from the synoptic-scale 
storm event (three to five days) to the climatic-scale 
event (a month to a decade or longer). Second, it 
refers to the unified nature of the observations, 
which is necessary for multidisciplinary studies of 
land, ocean, and atmospheric processes and their 
interactions and exchanges. The observations, made 
with an optimized set of sensors, will be nearly si- 
multaneous, and thus will eliminate many of the 
higher-frequency geophysical variabilities and 
biases that must be removed from observations 
taken at slightly different times of day (or even 
within a few days) prior to various scientific anal- 
yses. Many of these biases have sources and varia- 
bilities that are themselves the subject of investiga- 
tion as factors affecting large-scale biogeochemical 
fluxes, for example. Third, the word "comprehen- 
sive" refers to the fact that the observations encom- 
pass bands that have been measured with past and 
contemporary visible and infrared imagers and 
scanners, so as to provide for a longer-term record 
while permitting improved spectral sensitivity and 
hence better information content. 

The heritage o\ the MODIS instrument includes 
the Landsat Multispectral Scanner (MSS) and The- 
matic Mapper (TM), used for Earth resources; the 
Advanced Very High Resolution Radiometer 
(AVHRR). used for meteorology, sea surface tem- 
perature, sea ice, and vegetation indices: the 
(Oastal Zone C Olor Scanner (C ZC S) used for 
oceanic hiomass measurements and ocean circula- 
tion patterns; and a variety of experimental instru- 
mentation including the thermal Infrared Muiti- 
speclral Scanner (TIMS). Shuttle Imaging 
Spectrometer l^xperiment (SISI-X). etc. 

MODIS will provide a continuing series of ob- 
servation^ complementary to those obtained with 
the above instruments. It is included as part of an 
[•OS research obser\atit)n strategy that would aug- 
ment the eapalility of the predecessor instruments 
anp.iderahl\, boih in terms of frequency t)t coverage 
(at reduced resolution, for the MSS ;:nd IM), and 
m terms of comprehensive spectral coverage. The 
.idded capahiln in the spectral domain serves to 
permit advanced algorithm development with a con- 
si >tent data set, which con'ains the full ramie of 



global variability and richness, and will enable such 
advanced algorithms to be easily applied retrospec- 
tively to study linkages between biogeochemical 
components. 

The MODIS instrument requirements are de- 
fined by the large spatial-scale (a=l km) observa- 
tional requirements within the atmospheric, terres- 
trial, and oceanic sciences. These requirements are 
based on present (realized) capabilities and require- 
ments as well as those for which a firm foundation 
has been established (e.g., near infrared for im- 
proved ocean atmospheric correction). The result- 
ing matrices of spectral, temporal, and spatial cov- 
erage requirements, together with specific 
instrumental capabilities and characteristics, have 
been resolved into a unified set of requirements. 

If spatial resolution of 1 km precludes the pos- 
sibility of obtaming narrow enough spectral bands 
in the infrared necessary to have an atmospheric 
sounding capability on MODIS, then it is extremely 
important to have atmospheric temperature-humid- 
ity sounding capability provided alongside MODIS. 
This is needed both to correct MODIS observations 
for atmospheric effects and to permit thorough stud- 
ies of atmospheric phenomena and their interaction 
with surface processes. We are assuming that, at a 
very minimum, a sounding instrument of the quality 
of the High-Resolution Infrared Radiometric 
Sounder 2 (HIRS-2), the current operational in- 
frared sounder, will accompany MODIS. To obtain 
the most from MODIS capabilities, as well as to 
obtain an improved understanding of surface at- 
mospheric interactions, we recommend that an ad- 
vanced high spectral resolution infrared sounder 
With 10 km horizontal spatial resolution be devel- 
oped to accompany MODIS. 

MODIS COMPLEMENTS 
OPERATIONAL SYSTEMS 

It is expected that MODIS will complement pre- 
existing and concurrent capabilities on operational 
satellites, chiefly the slightly modified AVHRR on 
N()AA-K. L. M spacecraft. One of the two opera- 
tional NOAA payloads (one morning, one after- 
noon) will be in an orbit similar to that of MODIS 
or will be onboard the same platform. These two 
payloads together will provide more frequent global 
thermal infrared coverage (twice daily). The five- 
channel AVHRR. however, will be constrained to 
two long\\ave and one shortwave infrared window 
channels at night and two longwave infrared chan- 
nels and three reflected radiation channels (two in 
the near-infrared and ime in the visible spectrum) 
during the day. and will have no ocean color capa- 
bilities. It is not known v/hether there will be an 



g) 



wm 



wm^n^ifmmm^ 



operational Ocean Color imager (OCI) in orbit, es- 
pecially on a U.S. satellite, during the time frame 
in question. 

Thus the U.S. concurrent operational satellites 
(including Defense Meteorological Satellite Pro- 
gram (DMSP) and Geostationary Operational En- 
vironmental Satellite (GOES-Next)) will form an 
incomplete basis for an adequate and integrated 
Earth Observing System, even if complemented by 
somewhat similar foreign systems such as Systeme 
Probatoire d'Observation de la Terre (SPOT), 
Marine Observation Satellite/Land Observation 



Satellite (Japan) (MOS/LOS), and European Space 
Agency Remote Sensing Satellite-1 (ERS-1). 
MODIS can provide the basis, from both an instru- 
ment development and an integrated data base man- 
agement standpoint, for follow-on operational sat- 
ellite data collection, processing, archiving, and 
dissemination. Research with the more advanced 
and more comprehensive (both spectral and tem- 
poral) measurements to be available from the total 
MODIS system should demonstrate conclusively the 
value of providing the more comprehensive data on 
an operational basis. 



i 

I 



■h 



mmm 



^mmmmmmmmmw 



tmmmm 



-^ 



^^wm 



III. SCIENCE OBJECTIVES AND INSTRUMENT 
REQUIREMENTS 



TERRESTRIAL STUDIES 

The anticipated contribution of MODIS to ter- 
restrial studies is to provide regular, high temporal 
frequency coverage potentially of the entire land 
surface. This moderate spatial resolution, global 
coverage sensor will be complemented by the high 
spatial/spectral resolution capability of HIRIS for 
detailed study of limited areas of the Earth's sur- 
face. MODIS is conceived as being an essential com- 
ponent of an integrated multisensor system, and 
from the terrestrial studies point of view, ii is this 
simultaneous coverage by MODIS and HIRIS that 
would provide a unique capability of the proposed 
Eos system. 

The MODIS sensor will be the primary tool for 
global ecological research on the Eos platform. Be- 
cause of critical global issues facmg mankind, such 
as climate change, desertification, resource deple- 
tion, and region-wide pollution, the science com- 
munity is being directed toward the problems of 
global ecological research. Existing data bases to 
support such research on a global scale are at best 
limited and quite often completely unavailable. As 
a consequence, much of the existing global ecolog- 
ical research has been through computer simulation 
modeling based on small and often unrepresentative 
sample data sets. Although these models are cur- 
rently providing insight into the scale and dynamic 
nature of global biogeochemical cycles, and focus 
attention on rate-limiting processes, the models are 
often constructed from fabricated and temporally 
static global data bases in which we have little sta- 
tistical confidence. As part of the Earth Observing 
System, the MODIS instruments proposed in this 
document will provide a long-term data base giving 
a view of the dynamic changes of the Earth's sur- 
face. This will permit phenological changes in ter- 
restrial vegetation to be quantitatively assessed at 
regional, continental, and global scales. 

Our present understanding of the utility of low- 
resolution satellite data for terrestrial studies is, for 
a large part, based on the results of a small number 
of recent vegetation studies using NOAA AVHRR 
data (Mayes. 19S5|. These studies give us an indi- 
cation of the potential usefulness of MODIS data 
for vegetation inventory and monitoring at regional 
and continental scales, but have only just started to 
reveal the full potential of 1 km resolution multitem- 
poral satellite data. 

Recent developments using I km resolution 
mullispeclral sensors have shown that it is possible 
to provide reliable cimtinental- and global-scale data 
sets i>f sufficient precision to supplement and im- 
prove upon existing inputs for global models (Tucker 
(7 al.. IMH5a and b). lixisting small-scale vegetation 



maps are inadequate for providing a comprehensive 
and up-to-date estimate of the distribution and areal 
extent of the major vegetation formations of the 
world. Low-resolation, remotely-sensed data pro- 
vide the means by which global vegetation maps can 
be updated and improved (Tucker et al.. 1985a). 

At both regional and global scales there is a lack 
of baseline information on such environmental is- 
sues as the rate of tropical deforestation and the 
type and rate of regrowth, the areal distribution and 
effect of acid rain on the boreal forests of Europe 
and North America, and the rate and extent of de- 
sertification at the margins of the world's deserts 
and arid regions. These examples illustrate a diver- 
sity of environmental issues ihat will require care- 
ful, sustained measurement through and beyond the 
next decade to understand fully their effect on the 
biosphere. 

Monitoring long-term changes in the bounda- 
ries of selected natural and man-altered ecosystems 
will undoubtedly provide us with direct evidence of 
the effects of climatic change on the biosphere. Most 
perturbations of the biosphere will be evidenced 
first by subtle vegetation responses that may reflect 
both the nature and severity of the perturbation. 
Vegetation response to stress varies with both the 
type and degree of stress. As a general rule, chronic 
agents such as geochemical stress (Goetz et al,. 
1983) affect vegetation in the following ways. Low 
levels of stress generate biochemical changes at the 
cellular and leaf level, which have an influence on 
pigment systems and canopy moisture levels (La- 
bovitz tt ai. 1983: Chang and Collins, 1983). In- 
creasing stress may result in phenological changes 
such as delayed leaf flush and/or premature senes- 
cence (Labovitz et ai. 1983). Alteration of canopy 
structure and the resulting reduced percent cover, 
canopy closure, or biomass may also result from 
increasing stress levels. At higher levels of stress, 
community composition will change, with less tol- 
erant species being replaced by more tolerant spe- 
cies (Rock and Vogelmann, l9K,Sa: Abrams et ai. 
1985). Incipient water stress will have a dominant 
effect on the canopy moisture content, which in turn 
influences the canopy reflectance and emittance. A 
wide variety of vegetation responses to stress as well 
as some causes of stress may be monitored by re- 
mote sensing, and the repetitive coverage by 
MODIS should permit regular measurement of veg- 
etation stress conditions worldwile. For example, 
information on the areal extent of freeze or drought 
damage to vegetation, for both cropland and natural 
communities, would be economically and ecologi- 
cally useful and technically feasible through analysis 
of MODIS data. Air pollution or toxic chemical 
damage may also be measurable. 



..„ ,JP'''!''"> """P^^^hcnsivc and timely information 
n the phenology ot natural vegetation and the sta- 
tus ..I agneultural crops through the growing season 
..re unavailable k,r all but a fcw kKalized aa-as. ve" 
such miormation is essential for m.Hleling global 

pcrcd bv the absence of reliable and quantitative 
data ot^ agricultural and pastoral conditions at a 
regional sea e Such data can be crf^tained through 
analysis « daily low-resolution multispectral satel- 
lite data (Justice ei al. . 1985). 

nh.rP""- '1^.*:"''"""?. "f ''"•^face climate and atmos- 
phe ic conditums will be among the more important 
unc lons ol hos. An example (,f global mean sur- 
frnm mK'"'*'. "i«:'"*"'^^'"^nts currently available 
trom HIRS-2 and Microwave Sounding Unit (MSU» 

oi in.'?.";'"." *? ^T' ' ■'^"^•^*^^' "^""^ variables 
merest to land pnH.esses include temperature, 
radia ion balance, humidity, and precipitation. 
Ihcsc data can be used to drive computer simula- 
tion nKKlels <,f important biogccKhemical prcKcsses 
in order to interpret changes in surface features and 

be used to calculate cn.p phenologv and urowth 
rates stress events, nitrogen and sulllir MuxJs. car- 
b,.n dioxid- exchange, and numerous other process 

«/«/.. 1977). Several studies have shown the utility 
ot m»,n.toring albedo by satellite remote sensing 
Ot.crman. 19X1; Courel. I9.S.S). MOUSS for the 

Npeitral albedo changes simultancuslv with bit,|oc- 
lea lly imp.rtant parameters,.,, a regional and ulc^l 
studies" '"'''^'" **' '"'"="*^-'^'^'a'"' -nvironr^iental 
M()I)IS will als<. provide a number (,f useful 
products lor surface hydrologv. Changes in land 

Xi Z; ^' r'"""'"' ^''^"^ '^'y sipnilicantlv 
affect most ol the maj.ir terrestrial biophysical svs 
terns. Among the most imp..rtant influences are 
changes in runoff, inhltration. and evap<,.ranspira- 
t.<.n rates altect.ng the hydrological system: changes 
■n evapotransp,rati..n rates a fleet inu the dobal en- 

^Tfi'.i »"'!•• ''T^ '" '■•^••^i'^i'i'.v of surface ma- 
tcrals alfecting sediment transport systems anu the 
distribution ol areas of net er,.sion-dep,.sition 

M<mitoring continental chanees in smm cover 
and associated changes of albed.. will alv. be pos- 
sible AVatershed to basin-level snowmelt estimates 
can provide prediction ..f the temp,.ral dynamics of 

e Im .n^'^"^^K '""'u '^ iniP<Tti.nce for calculating 
elemental exchange between land and .Kvans for 
ll«H»d warning, and f,.r irrigation management In 
c-«.n,uriction with precipitation rates sampled by 
her E. OS se„M,rs. large area estimates of soil mois- 
ture *ill provide critical input into assessments of 
vegetation gr.mih and water stress. Regional evap- 
oration and transpiration rates mav be modeled 
Irom surface wetness and surface climate These 



processes have implications as surface feedbacks to 
global climate models and as controls of vegetation 
stress and changes in primary production 
Mn^^r""^ " ^^^^ inventory of global vc-etation. 
MODIS spectral data should be able to prov de mea- 
sures of some important derived products, such as 
standing green biomass. intercepted photosynthct- 
ically active radiation (IPAR). and net primary 
productivity (NPP). dependent on growth in our 
understanding of spectral information in the next 
live to seven years. For example, vegetation leaf 
area index (LAI) is an important structural variable 
that can ne used to caiculatc mass and energy ex- 
change from vegetated surfaces. Global estimates of 
LAI w.uld allow computation of carbon exchange 
rates bav.-d on measured data, instead of the 
guesses that must be used today Estimates of global 
:ZT '".n?^''""''''' P'""' ^"""ass vary from 4(H) to 
■::l\h ^. (gi-ams carbon). Thi; uncertainty 

could be considerably reduced with a MODIS surveV 
ot plant biomass. and this would help significantly 
m understanding the balance of the global carbon 

ed thmugh global geographic information systems 
u,ac.htate implementation of computer simulation 

e.vele models, hydrolog.c cycle energy models, and 
comprehensive biogeochemical cycling models will 
all require this data stream. It is recognized that 
.d r ' K '^'T^^^^' global ecological issues arc 

UKvmh " /""^^'"^-'"•''l processes such as pho- 
osvn hesis will be necessary for a dynamic viel of 
the global ecosystem (Figure 2). MODIS data will 
provide a direct, timely gk)bal measurement of the 
m"Idds' '■""''"'""' ^'' •*"- '''"^ '^"rf^ce for these 
A brief description of the NOAA/AVHRR svs- 

being used Schneider e: al.. mi), will give the 
background k,r the present perception <,f what will 
be P<.ss.ble f.„m improved moderate spatial reso- 
AViTuD '""'"'""" '"'■'' '•"' '^"I^'S The NOAA/ 
wkjI r u"'"' ""'" '•'" '''""^h^d in 1^78 (Kidwell. 
iyS4) and has provided data since that time at a 
ominal spatial resoluti<,n of 1. 1 km at the subsalel- 
lite point ( (able I). The standard NOAA AVHRR 
product, collected worldwide on a daily basis, is .he 
gkibal area coverage (C.AC) data with a .s km x ^ 

ulllirTK'"'" '■''"'"y"' <-'^*--''^''''^- IW2: Gatlin cf al'.. 
• h . IV n"'^"^*-"^' ^y onhoard processing of 

theraw l.lkm x 11 km IcKal area coverage ( LAC 

ui'F-.'^h he- *"'T^^' '""''■ "^"y ^' '^'""'"'"'-'J 
to I arth by special request. Since April 1982 CiAC 

P^Hlucts have been used to generate a third pr,Kl- 

(N(>AA. I98A; larplev «7 «/. . 1984). The (iVI prod- 
uct IS resan.pled from the CiAC data to give a polar 



L2 



® 



mmmm 



m 







® 



Ms^»>lftL^ 



HH^ 



w 



m 



M 

i^. 



'^i'^ 



SATELLITE 
DATA 

Meteorological 

Incoming Short-wave 
Radiation 

Surface Temperature 

Surface Moisture 

Invento ry 

Biome Type & Area 
Topography 

Striicture 

Leaf Area Index 

Biomass 

I 



PROCESS 
MODELS 



Gross 

jPhotosynthesis 

Controls 

Carbon Dioxide 
Radiation 



ECOSYSTEM 
MODELS 



GLOBAL CARBON 
MODEL 



Net 
Photo- 
synthesis 




^ 



Total 1 

I Biomass !" I j 

Z ^\ Atmosphere] 

I T , \ t t V 

J Net Primary J^ ^ i^- , 

1 Production | Nw , , I Oceans 

«v I Uve Terrestrial I 

A ♦ I Vegetation 

Detritus ^hlT* ~ ~ ^ 



Detritus 
Soil Carbon 



3 



Rgjire 2. OrganizatkMMl diegram of a proposed 
J«n.bles ar* derived from satllliie data-pSl 



Ekli^tor.KT'^J!!"*'''*""" '"' » «"•»««»«« forest. All drivine 
linkages to a global carbon model are shown by dashed lines ( Ru!!Sl 



4\ 



Table 1. Characteristics and Status of the NOAA/AVHRR Systems 

I '1 1 ■ rt,rtl* .r« >-l /^ —J. _l. an^,^ _ _ -. 



Nmi ft l! ?!?.^**'*'" '''^«- NASA funded protofligi,, 
SoA aI'.'^""*'''*''' ■'""^ '^7^- NOAA funded ^ 

SoA a'« '^""<^^*=d June 1981. NOAA funded 

Nm a"2* ^""'^5 ^^'"^ ^*^^^' NOAA funded 
NOAA-9 launched December 1984. NOAA funded 
Orbit inclination: 98.8° 
Orbital height: 82()-87() km 
Orbital period: - 102 min 
Scan angle range ±55.4° 
Ground swath coveraee- "> 7(M) km 
IFOV 1.39 to 1.51 mrfd " 

Ground resolution 1.1 km (nadir): 2.4 km (max. scan-angle alon. tr.ckr 
6.9 km vmax. scan-angle crosstrack) ^ ^ ^''^'^ '• 

Quantization 10 bit 
Equatorial crossing Descending 

07:Mi 

l4:Mi 



Spectral channel 
Spectral range (nm): 



1 
0..58-0.68* 



Ascend inu 
iy:3(»(NOAA-6anJNOAA-X) 
<»::.^(l(NOAA-7andNOAA-y) 



NOAA-6 
NOAA-7 
NOAA-8 
NOAA-9 



2 3 

<».72.5-|.| 3.5-3.93 

Status as oflVlay 1985 



4 
I(»..VI|,3 



11.5-1; 



Taken out of operational service 5 March 1981 R.Mno.., .mi 
Taken out of operational service MarJ!, 198^ ^""'''"'^ " ^""^• 
lakcn out of operational service 12 June 1984 
(.operational. 



1984. 



"Channel I range on TIROS-N (> ^s t„ n go 
•N<ii on TIROS-N. NOAA-h. or NCJAA-N 



s> 



stereographic projection with 15 km resolution at 
the equator. 

Examples of the NOAA GVI, GAC, and LAC 
data produced by the Global Inventory, Monitoring, 
and Modeling Studies (GIMMS) group at NASA/ 
GSFC are presented as Figures 3, 4, and 5, respec- 
tively. Figure 4 is an example of 8 km resampled 
GAC produced by the GIMMS group for the pur- 
poses of African continental vegetation inventory 
and regional drought monitoring. The range of cur- 
rent activities using AVHRR data includes conti- 
nental land cover mapping (Tucker et aL, 1983; 
Townshend and Tucker, 1984); rangeland monitoring 
and grassland productivity estimation (Tucker et ai , 
1983); tropical forest inventory and deforestation 
monitoring (Tucker et aK, i984a; Justice et aL, 
1985); agricultural crop and drought monitoring 
(Justice et al. , 1984; Tucker et ai , 1984b; Justice et 
ai, 1985); ecological modeling (Norwine and Gree- 
gor, 1983; Goward et ai, 1985); desert locust mon- 
itoring (Tucker et ai, 1985b); and forest fire moni- 
toring (Malingreau et al. , 1985; Matson et ai , 1984). 

The methodologies for analyzing high temporal 
frequency, low spatial resolution data are currently 
being developed and can be expected to continue to 
substantially improve over the next few years. Of 
particular interest is the establishment of long-term 
data bases showing trends in vegetation response 
over a number of years. These can be used to ex- 
amine the effects of long-term climatic change and 
altered land use practices. 

The NOAA/AVHRR systems arc assumed to be 
funded through the 1990s with slight modifications 
planned for the NOAA-K. L, and M satellites start- 
ing in 1989 (McElroy and Schneider, 1984 (see Table 
2)). There is thus the exciting possibility of contem- 
poraneous orbiting of the AVHRR and MODIS 



systems. With the sensor system characteristics out- 
lined in Chapter IV, MODIS will represent a sub- 
stantial improvement over the existing and planned 
AVHRR systems, especially in terms of the number 
of spectral bands designed to derive land and ocean 
parameters. The AVHRR system was designed as 
an operational meteorological sensor and has sev- 
eral characteristics dictated by this mission objective 
(Table 1) that are inconsistent with land and ocean 
science objectives. The present 7:30 a.m. and 2:30 
p.m. overpass times result in problems of lov^ light 
levels at high latitudes and cloud cover over equa- 
torial zones. The amount of 1 km data (LAC) cov- 
erage possible has been and will continue to be se- 
verely limited by the capacity of the onboard tape 
recording system. The locational accuracy require- 
ments of meteorological applications are consider- 
ably less severe than those for surface mapping and 
measurement. Similarly, the pre-launch and on- 
board sensor calibration priKcdures currently as- 
stKJated with the A\ HRR leave much to be desired. 
MODIS-N is being designed with terrestrial 
monitoring as a major mission objective, with Eos 
orbital characteristics, sensor calibration and config- 
uration, liKaiional accuracy, and data requirements 
specified accordingly (see Chapter IV). With the 
projected improvements in present data processing 
and ai chiving technology, global coverage at 500 m 
every two days will give an enhanced monitoring 
capability over the present ^nd planned AVHRR 
systems. This improved spatial resolution will per- 
mit the detection of loud transformations undetect- 
able using the current daily 4 km AVHRR data. 
Beyond that capability, a unique contribution of 
MODIS for terrestrial observations lies in the Eos 
concept of an integrated, multilevel sensing system. 
The capability for monitoring surface conditions will 



Table 2. Proposed AVHRR Characteristics on NOAA-K, L, M 



Channel 



Spectral Band 



S/N 



r 

2* 

3a (day)* 

3b (night) 

4** 

5** 



0.55 - 

0,84 - 

1.58- 

3.63- 

10.3 - 

11.5 - 



0.65 \im 

0.87 ^tm 

1.64 \kxr\ 

3.9 \x.m 

11.3 ^im 

12.5 M-rn 



9:1 at 0.5% reflectance 
20:1 at 0.5% reflectance 



Ml is alv> proposed that cither the digitization he increased from U) hit to 12 hit or that sampling of the dynamic range be changed to: 

I 



Rdlectivc Range {^) 

()-25 
2fKl()0 

26-100 

0-12.5 
12.6-100 



Counts Raitite 

(»-5o<» 

5lN)-t.4NNI 

t>-5(M) 
5IN»-I.(NM1 

t^-5tlO 
500-!.<t(H» 



* 'It has Ken prop*ised that the maximum brightnesN temperature in channels 4 and 5 he mcreased frt>m 320 K (*> 340 K. 



® 



«■■ t- 



fiEIKXWQUAUn 




10 



ORIGINAL PMH 
COLOR PHOTOQRAFH 




FigHre 4. Nomalizcd difference vegetation index derived from AVHRR GAC data resampled to 8 iiiiometers. 



10 



I 



, ^ I.*!.-/. 



® 



^ORIGINAL FAdt 
COLOR PHOTOSWH 




4 



lis 
'if 

Z 



1 






Sic 

2 




ll 

^^ A ^ 

«l^ 



z 



.7 

a 
I 

c 

k 
3 






z 



II 



® 






increase considerably when ii becomes possible to 
simultaneously and selectively sample complete 5iK) 
m resolution coverage with a high resolution (30 m) 
limited swath. This is the conceptual basis for an 
integrated MODIS/HIRIS sensor configuration as 
proposed tor Los. The possible results from such a 
system would include cloud identification, atmos- 
pheric correction, and interred rainfall in support of 
terrestrial studies. The practical problems of inte- 
grating high- and low-resolution data from different 
satellite systems in icrms of data acquisition, tem- 
poral and spatial registration, and calibration pro- 
hibit this approach with existing satellites. The 
MODIS/HIRIS combination is one of the unique 
i)ppjrtunities of the Eos polar platform approach. 

Instrument Requirements for Terrestrial 
Studies 

The instrument requirements for MODIS are 
determined by the applications for which the system 
is intended. The prime function of MODIS for ter- 
restrial studies is the monitoring of vegetation dy- 
namics and land transformations for a wide variety 
of research studies. One example of such studies 
that is receiving considerable attention at present is 
global CO> modeling (Hansen ei «/., \9[\\: Fung et 
ai , I9S6). The MODIS contribution to this research 
would include measures of the contribution from the 
vegetation component of the biosphere to the CO, 
cycle. Such measures would require a reliable map 
1)1 the distribution of the major vegetation forma 
tions. the timing of their major phenological events, 
and the associated variations in photosynthetic ac- 
tivity and biomass production. Such objectives 
would necessitate a sDfficiently high temporal fre- 
quency monitoring to detect phenological events 
such as green-up and senescence and to monitor 
variations in the timing of such changes over a num- 
ber of years. To undertake such a measurement pro- 
gram will require consistent and reliable data sets 
over a number of years, which implies the need for 
a continuous coverage policy and a dependable op- 
erating status. In support of such a studv. the HIRIS 
system would provide greater precision in our un- 
derstanding of the relative contribution of the dif- 
lerent components of the vegetation canopv to the 
measured reflectance, to the observed gieen-up, 
and to iO exchange. The above example is just 
one ol several potential uses of data from a MODIS 
system designed for measuring vegetation dvnamics. 
In the following subsections we examine some of 
the required sensor characteristics in more detail. 

Spatial Resolution 

I he proposed MODIS resolution (or terrestrial 
studies is 5(MI m, an improvement bv a factor of 2 
over the AVHRR (I km) spatialVesolution for 
regional monitoring. A 5(MI m resolution at the edge 
of the swath will afford improved location of ground 



features for detailed calibration and complex scene 
radiation studies while permitting resampling for 
those continental studies requiring a lower spatial 
resolution (i.e., >500 m). Similarly, a resolution of 
^m m will facilitate integration with the HIRIS 30 
m resolution data. Agriculture and forestry prac- 
tices frequently subdivide land on 50 to 100 acre 
blocks, or 0.2 km-. The 5(K) m resolution should 
allow adequate discrimination of these features, 
which would be lost at lower resolutions. 

High-frequency monitoring of large areas in- 
volves a trade-off between spatial resolution and the 
quantity of data that can be practically handled. The 
data stream from MODIS will be over 3 x 10^ times 
less than HIRIS per unit of land area. Even so, the 
predicted data flow from the MODIS system (see 
Chapter IV, Data Rates) will require considerable 
improvements in data handling and storage over the 
existing Landsat and AVHRR systems. 

Revisit Time (Temporal Resolution) 

The required revisit time of the svstcm and, 
indirectly, the orbital altitude are determined by the 
highest temporal frequency changes in the surface 
phenomenon that is to be measured. It is often the 
periods of rapid change in the vegetation that trigger 
other important processes within the ecosystem. 
The temporal resolution of the system should be 
sufficient to resolve such events! Vegetation re- 
search by remote sensing does not normally require 
daily data, as the development and senescence of 
plant communities is a relatively continuous pro- 
cess. However, it does require reliable and usable 
coverage within a week to 10 days, which, owing to 
cloud cover problems, necessitates a higher fre- 
quency of acquisition. Occasional threshold events, 
such as catastrophic subtropical freezes, are instan- 
taneous: however, the resulting damage to the veg- 
etation is best evaluated days after the event. Many 
vegetation stress responses, such as water or pollu- 
tion stress damage, accumulate over periods of 
weeks to even years (Running, 1984). In contrast. 
semi-arid grassland species can germinate and fruit 
within a two- to three-week period. At present, the 
highest possible temporal coverage for the AVHRR 
LAC (within ± 3(P off-nadir viewing limits) of three 
passes every nine days is required to produce ac- 
ceptable cloud-free coverage of semi-arid areas, 
where such vegetation flushes are of major ecologi- 
cal importance. Even so. for areas of persistent cloud 
cover and haze, cloud-free coverage will remain 
L obtainable and, for such areas, emphasis must be 
pla... J on the Eos active and passive microwave sen- 
sors. Symhetic Aperture Radar (SAR) and High- 
Resolution Multifrequency Microwave Radiometer 
(HMMR). respectively Some land surface hvdro- 
logical events are sufficiently transient in nature that 
daily coverage is required, e.g. . flooding and surface 
wetness. Seasonal snow accumulation and melt 
could be monitored at less frequent time intervals. 



12 



vt) 



One of the most compelling reasons for a high 
temporal frequency for MODIS coverage is to pro- 
vide estimates of surface climate conditions. The 
most important surface climate variables include air 
temperature, incoming shortwave radiation, humid- 
ity, and precipitation. MODIS, accompanied by a 
sounder, will be able to provide estimates of tem- 
perature and radiation, with inferences of humidity 
and precipitation through estimation of the dew 
pomt, cloud monitoring, and measurement of sur- 
face wetness. Important vegetation processes such 
as photosynthesis and transpiration respond to daily 
changes in surface climate. Remote sensing cannot 
always directly monitor such continuous vegetation 
processes; however, they can often be inferred by 
computer simulation of plant responses to other 
measurable environmental parameters (Running, 
1984). For example, integration of vegetation struc- 
tural features, such as leaf area index, with daily 
surface climate parameters derived from the j:atel'- 
lite, may provide the best means of estimating CO. 
and H,0 flux from vegetated surfaces. 

Based on understanding of the measurement 
studies that will be performed using MODIS data, 
the terrestrial studies group recommends as close to 



daily coverage as possible within a 30° sensor view- 
ing angle limit. 



Spectral Resolution 

Present requirements for vegetation remote 
sensing from MODIS could be met by an approxi- 
mation of the spectral channels used on the The- 
matic Mapper (TM) that were selected for this pur- 
pose. There has been little change in understanding 
of spectral requirements for vegetation monitoring 
smce the launch of TM; however, refinements re- 
sulting from current TM studies (Barker, 1985) 
could easily be incorporated within the MODIS de- 
sign and construction time frame. 

Table 3 lists the major spectral regions of value 
for vegetation analysis (Cox, 1983). It cites specific 
wavelength regions of value for vegetation stress de- 
tection, species discrimination and mapping, and 
biomass estimation. Without sufficient scientific jus- 
tification for high spectral resolution at the pro- 
posed moderate spatial resolution of MODIS, 
broader spectral bands have been kept than those 
recommended for HIRIS. For terrestrial studies in 
general, the MODIS channels should not encomi ass 



Table 3. Spectral Regions for Remote Sensing of Vegetation^ 



Wavelength (n^m) Ty pe of Feature 

0.440-0.500 



Value 



0.650-0.700 

0.700-0.750 
0.800-0.840 
0.865 
0.940-0.980 

1.060-1.100 

1,140-1.220 

1.250-1.290 

1.630-1.660 

2.I90-2.3(K» 

3.(KK)-5.(MK); 
8.(KK)-I4,(XK) 



Absorptance 

Absorptance 

Reflectance 
Absorptance 
Reflectance 
Absorptance 

Reflectance 

Absorptance 

Reflectance 

Reflectance 

Reflectance 
Emittancc 



Detection of changes in chlorophvll/carotenoid ratios rrelatec to 
stress). 

Detection of chlorophyll states as well as tarnin and a.ithocyanin 
content. Initial stress detection. 

Senescence detection. Detection of dead or dormant vegetation. 
Possibly related to leaf anatomy and/or state of hydration. 
Height of feature may be useful in species discrimination. 
Shifts in this minor water absorption band mav be useful in 
species discrimination and determination of hydration state. 
Shifts in peaks may be related to leaf anatomy and/or 
morphology May be useful for species discrimination. 
Shifts in this minor water absorption band mav be useful in 
species discrimination and determination of hydration state . 
Height of this feature very useful for species discrimination of 
senescent rorest species. A ratio of this feature with the one at 
1-645 offers a good indication of moisture content and thus stress. 
An indication of moisture content of leaf. Mav also be an 
indicator of variation in leaf anatomv. May be' useful for species 
discrimination. An indicator of leaf moisture content when used 
as a ratio with the 1.270 data above. 

An indicator of moisture content. Mav also be of value in species 
discrimination. 

Little is known concerning the optimal thermal wavelengths lor 
studying different vegetation parameters. This is an area that 
needs further study. 



«1 



* Cox. I9S3 



13 



® 



a mospheric water absorption bands. Where possi- 
ble, sufficiently narrow bands should be selected to 
minimize atmospheric effects, allowing a trade-off 
between bandwidth and radiometric sensitivity. The 
following paragraphs highlight some of the more 
critical wavebands for vegetation studies. 

For detection of changes in chlorophyll/caroten- 
oid ratios which provide a measure of plant stress 
a channel in the 440 to 5()() nm region is required' 
A spectral channel near TM channel 2 (520 to 60() 
nm) is necessary because of the increased re.lec- 
tance of green vegetation in this portion of the spec- 
trum. In addition. Richardson ei at. (1983) found 
that reflectance in the near-infrared region (0.76 to 
U.90 Jim) was closely related to canopy nitrogen an 
reSrch"' parameter in biogeochemical cycling 

(.m r^AcS^Tilf ^^''"^at'""- i^Pectral channels at 
630 to 690. 660 to 680. and 760 to 9(K) nm are re- 
quired. A channel close to TM3 (630 to 690 nm) is 
extremely important owing to its sensitivity to the 
chlorophyll concentration in vegetation, as well as 
tannin and anthocyanin content. A channel at 700 
to 750 nm would provide useful information relating 
to senescence of vegetation, which would allow th§ 
monitoring of translocation of nutrients from the 
fohage. The 760 to 900 nm region, where infrared 
reflectance is at a maximum, is essential for vege- 
tation monitoring and is dircctiv sensitive to foliage 
biomass. A ratio of near infrared (760 to 9(K) nm) 
and red (630 to 690 nm) has been shown to be well 
correlated with foliage biomass for crops, range- 
land, and forests (Tucker. 1980: Spanner et al 1984- 
Running et al. . 1985). 

oi/*''^ ''P'^'^'ral wavelength chanp :! from I ^S to 
1.-9 Jim has been shown to be usciul for discrimi- 
nation of forest species (Rock. 198:> 1985b) A 
channel comparable to TM5 ( 1.53 to 1.73 am) spe- 
cihcally 1.63 to 1.66 ^m. is sensitive to the moi^iirc 
content of leaves and may be indicative of variations 
in leaf anatomy. This channel, when ratioed with 
tnc 1.25 to 1.29 Jim channel, is an indicator of leaf 
moisture content. Two channels within the "> to 
-_4 nm region will be useful for examining canopv 
chemical composition. A channel centered at -> 18 
Jim IS useful tor observing a protein absorption fea- 
ture and may allow for determination of the nitrogen 
content of tbiiage ( Peterson ei al. . 1985 ). 

Relatively little is known regarding the value of 

hermal data m the study *)f vegetation as compared 

lo the visible and near-infrared portions of the spec- 

irum. However, such studies as presented bv Gurnev 

Muvn/u'^^"^\- "^'""''"'■g '-' "/• (iyx2); and Soer 
( N(l) have demonstrated the potential and utilitv 
ol thermal data for regional soil moisture and evapo- 
transpiration studies. While it is not possible at pre- 

f'Tllrvi?'!'-'' '^'" "f"™"'" "^^"™al band selection 
K r MODIS. future results from inalvsis of Thematic 
Mapper (channel 6). /V'HRP (channels 3. 4 S) and 



the airborne TIMS data will permit an informed 
band selection in the future. 

Infrared atmospheric windows, that is, spectral 
regions in which the atmosphere is relatively trans- 
parent to infrared radiation, are found primarily in 
the regions between 8 and 12 jim (longwave win- 
dow) and 3.5 to 4.2 fim (shortwave window) The 
regions each have advantages and disadvantages for 
monitoring surface temperature. The shortwave 
window observations are much less sensitive to 
water vapor absorption, especially water vapor con- 
tinuum absorption, which becomes very large at 
longer wavelengths in humid atmospheres The 
shortwave observations are also roughly three times 
more sensitive to variations in surface temperature 
and one-third as sensitive to variations in surface 
cmissivity. On the other hand, shortwave observa- 
tions are strongly affected by reflected solar radia- 
lon during the day and the surface emissivity over 
and is more variable than at longer wavelengths 
(see Figure 6). * 

AVHRR currently has two longwave window 
channels and one shortwave window channel (see 
latJle .) At night, observations in all these channels 
are used to help correct the temperature observa- 
tions for atmospheric effects. During the day. only 
t^ne two longwave window channels are used. HIRS- 
-. in the other hand, has one longwave (10 9 to 11 3 
^m) and two shortwave (3.95 to 4.0 jtm and 3.70 to 
3.83 jtm) channels. The shortwave windows allow 
tor the simultaneous determination of daytime sur- 
face temperature and reflected solar radiation (by 
assuming the surface reflectivity is the same in both 
shortwave channels). 

MODIS technology allows for spect.al resolu- 
lon comparable to that of HIRS-2 while keeping 
the I km spatial re olution of AVHRR. The miin 
improvement from .is will come from the ability to 
maKe split windows in the shortwave region allow- 
ing for the daytime use of those channels that are 
less sensitive to water vapor absorption. Several op- 
tions exist and one of these is outlined below. 

Table 4 shows the central waveicncth (X) and 
channel widths (A\) for the thermal infr^ircd (TIR) 
considered for MODIS. Also shown in the Table are- 
he transmittance to the surface, t.: the transmit- 
tance excluding water vapor continuum absorption 
T. , „; and the difference between the computed 
brightness temperature T,, and the sea surface tem- 
perature I,, for both July (()°N) and April (MfN) 
chmatological profiles. This difference shows the ef- 
fect of atmospheric absorption and the correction 
needed to obtain the surface temperature. The com- 
putations assume that the surface temperature is 
Kienticai to the surface air temperature and that the 
surface cmissivity is unity Effects of clouds or re- 
flected solar radiation are not included. 

This^ leads to a recommendation for four chan- 
nels in the 8 to 12 jim region and three channels in 



i 

4»' 



14 



® 



^s 



Pi (til 



m^fs^^mm 



i mmmmnmim^i «-■ 



T^ 



100.0 




— Beacn Sand, Daytona Beach, Florida 
— ^ Soil, VereenJng, Africa 
►-— Colts Neck Loam, New Jersey 
^- Chilean Nitrate Soil, Oficina, Victoria 



.\-. 



Cj-h-:/ T^Hr-.t^.-t^*^- 



5.0 10.0 

WAVELENGTH (Azm) 
Figure 6. Reflectance vs. wavelength for a variet> of soUs (Hblfe and Zissis, 1978). 



15.0 



the 3.6 to 4.1 ^im region. The channels at 11 and 12 
and 3.75 \im are the most similar to those on 
AVHRR. In addition, there are three longwave 
channels that may help in identification of aerosols, 
such as those put into the atmosphert bv the eI 
Chichon eruption, and two shortwave channels that 
will help account for reflected solar radiation and 
that are quite insensitive to atmospheric water vapor 
variation, N, being the absorbing gas. 

The channels at 8.55 ^m and 10.15 ^im were 
selected to help identify sulfuric acid aerosols, which 
have a maximum absorption at 8.55 |im and a min- 
imum near 10.45 ^tm. Water aerosols have nearly 
the same transmissivity at these two wavelengths. 
Thus, for an atmosphere where only aerosol type 
was different, the difference in equivalent black 
body (EBB) temperature between the two channels 
should be almost as constant as aerosol optical 
thickness changes for water particles, but should 
change noticeably with changes in optical thickness 
of sulfuric acid particles. As seen in Table 4, by 
comparing atmospheric absorption for the two pro- 
files, all shortwave channels are much less sensitive 
to water r apor absorption than the longwave chan- 
nels, for which absorption is so great in tropical 
atirosphcrcs that they become relativelv insensitive 
to the surface temperature. ObservationN at the 
longer wavelengths are less affected bv absorption 
by fixed gases (sec April 4(rN in Table 4) than arc 
the observations at shorter wavelengths. As a result 
of these relative advantages, observations in the two 
wavelength regions really complement each other. 
In the higher latitudes, where water vapor amounts 
are less and where atmospheric temperature is var- 
iable, the long wavelength channels perform well 
because of the greater sensitivity to water vapor and 
the lesser sensitivity to ^'ases other than water vapor. 



In the tropics, where moisture content is high and 
atmospheric temperat^i^es are nearly constant, the 
shorter wavelengths [perform better because of the 
greater contribution of the surface, and because the 
almost constant temperatures allow for compensa- 
tion for the larger absorption by the gases with con- 
stant mixing ratios. 

Table 4 also includes calculations for high spec- 
tral resolution infrared window channels (\/A - 
1,200, 600) to demonstrate what can be achieved if 
narrow-band channels cm be used to sound the sur- 
face at frequencies between absorption lines. It is 
clear that much cleaner windows can be obtained in 
the 3.7 to 4.0 jim region, which give atmospheric 
effects of less than 2° even in tropical atmospheres. 
Moreover, three clean windows in the shortwave re- 
gion will allow for simultaneous determination of 
ground temperature, reflected solar radiation, and 
surface emissivity. The last factor becomes verv im- 
portant when attempting to measure land tempera- 
tures. High spectral resolution does not reduce 
broadband effects such as those due to the N, con- 
tinuum around 4.0 ^m and the water vapor contin- 
uum in the 8 to 12 ^xm region. N. absorption is easily 
accounted for, however, because the N, concentra- 
tion is well known. The high-resolution, longwave 
channels are good windows in dry atmospheres but 
exhibit considerable absorption due to the water va- 
por continuum (but not water lines) in humid at- 
mospheres. This makes these channels particularly 
good for determining low-levol water vapor once the 
ground temperature is determined from the short- 
wave channels. 

The need for a supplemental infrared sounding 
instrument at higher spectral resolution but lower 
spatial resolution than MODIS will be discussed in 
Chapter III, on Complementary Atmospheric 



15 



A^flv 



tmmmmmmvmmmgmi 



ff^ 



(fim) 


AX 

(nm) 


3.750 
3.989* 
4.05* 
S.55 

10.45** 

11.03 

12.02 


'X) 
50 
50 
500 
500 
500 
5(K) 


3.723 


3.2 


3.822 


6.4 

3.2 


3.955* 


6.4 
3.2 


8.866 


6.4 

7.5 


10.914 


15.0 
9.0 


11.994 


18.0 
9.0 




18.0 



""" o^SurfS:*T^ Atmospheric Absoiption 
°^ !>urface Temperature Retrievals 

iuly (CN) 



April (40"N) 




MODIS Windows 



0.763 
0.845 
0.738 
0.375 
0.423 
0.405 
0.221 



0.768 



0.623 
0.868 
0.879 
0.815 



2.87 
2.42 
4.12 
8.09 
5.48 
5.34 
8.08 



H^h-Resoiution Windows 



0.911 

0.905 

0.958 

(J.926 

0.885 

0.879 

(».553 

0.552 

0.460 

0.460 

0.269 

0.269 



0.915 
(J.909 
0.%! 
0.929 



0.933 
0.929 
0.981 
0.981 
0.985 
0.983 



■ N amlinuum abM.rpl.on in s,eni(ic;,ni 
" ahMvption is M^nilicani ' 



Sounding Data. This sensor would provide atmm 
phenc soundmg capability as well asThigh s^c: 
tral resolution w ndows neeH^H t.. ;«. ^ V^ 

'"■<>" OreanographkSliidics '""^"aplcr 

Visiblc-ncar IR 4-8 channels 
Shortwave IR 2-4 channels 
Midwave IR 3 channels 
I «»ngwave IR 4 channels 



a pr'". :..ional list of 
the hands is 
provided in 
Table 16 



Orbital Overpass Time 

The decision concerning the orhii^i „ 
••me for terrestrial studies wf|| inev ab |v k^'^^P"'''' 
pron.se of several conflicting "luiifniL^;.^;-; 



1.08 

1.23 

0.74 

1 07 

1.85 

1.91 

4.39 

4.33 

4.29 

4.29 

6.30 

6.31 



0.874 
0.869 
0.741 
0.725 
0.837 
0.845 
0.749 



0.966 

0.957 

0.976 

0.9oO 

0.894 

0.891 

0.876 

0.878 

0.887 

0.887 

0.825 

0.824 



0.876 



0.782 
0.932 
0.948 
0.906 



0.967 
0.958 
0.977 
0.961 



0.947 
0.949 
0.993 
0.993 
0.9% 
0.995 



1.58 
2.14 
3.92 
3.78 
7.92 
1.44 
2.25 



0.73 

0.86 

0.70 

0.84 

1.77 

1.80 

2.43 

2.34 

1.03 

1.03 

1.33 

1.33 



IHln m AnnZr ^ ^^'P"»'''™ "round 



16 




M^mm m 



^^^^Fmfmmm^mmmm. u 1..J111 1 



nrr 



m 



siTSTss^iSr ;!''■ "^ "■" • -- 

diaraclerislics "^ ' Proposed senso. 



OCEANOGRAPHIC STUDIES 
*<*>8«al Oceanography 

there are Urge difS to .t^**"" "*«"*'• 

over lime of a fci aLS~nS'- ^."^J' "'' « '"™- 
.i»«. H^Thj^l " 7^'' «^e a„d',„m„ve, 
carbon cKle is he caSr/n""" '" "^ «'*»' 

" hasa ™mo«?TiSt"n,'n,r::f ^fr-'- »"• 

plankton a«i itf Sife, 'r^,„T^ 1^ P^V* 

sr^LeiiarBr-"-- 
.on;:i;:!SK?nSS'"'i^>^™ 

emphasize that othL « ^''^'^Wes of interest, we 

solar radiaSn vcSal S'h ' ^''^''''' ""^^ as 
and mixing ZngTy affect ^hv?"?"?'"' '"""P"" 
and trans^,rt (MSaf^rj'' S 

i.y h^ ^1^^^^^^^^^^ "f Pnmary prcHluctiv- 

(McCartX?^ K and re?""" '" ""^ ''"" ''"«<''^ 
Ho^*ever. the e isslll Z?!'*'""'' "''''^ 'herein). 

concernmg th?a .u "ct ZhTZ "' "'"'^"^""-^ 
variety of pr(Kess..« nvT. « ^ measurements. A 

such as inK::Vdra,1on'^STT"T"'^^ 



1975). these have'Sei b sId on? '"^ '""'^ ^^°' 
which may severciv i,n?w!^- ^ measurements 
duction. Kd esHm«. 'r^rl'y"^ P"'"«ry Pro- 

be 'as'lerarS: V KSeSa"^^^^^^ ^ 
estimates 'arger than earlier 

5!SiJiK?h2S?"^S?rS 

space scales S^^ii • ^ characteristic time and 
variabS mi£ ^1^,'^"'"^^ "^ Phytoplankton 

1984) Shinca .• '. *°= Denman and Powell 

the global estima^^f ;S S ItSd"^"' 
lier are based on comoosit^ of I '^ described ear- 
several decades and SIr!f measurements over 
portantly. we wish to det i°'^ l"''^"- "^"^^ *"»- 

fluctuations !n 7 SductSt m"uch""''"*^"'' "'^ 
have an accurate man <lf .1 ™"*^'' ** *« «''sh to 

Pling cannorS^'aTcoCl shed'S^h'^!?^ "'^'' ''^'"- 
ments alone. '''^°'"P'«sned with ship measure- 

the i^m^'SLundlp/ti^'r f' «"«-'-ns in 
plankton are argeiyTheresuUnf'""''""'' °^ P^y'^' 
ical processes. 4 nVed toTnde/stlTr '"i^''^'^- 
coupling between these Sivs^caS k ' "'^ .'""'' 
cesses if we are to unLrSn .k '"«'o«'cal pro- 
fate of biogenic mateSk'^u*'^ Production and 
and space Eales tKI^ ' "'""^ ''"^' °f '™c 

cesse isoft\,'urarT:ro«arerr'e'^^^^^^^^^ 
coastal upuelline teniU to L example. 

ern bounSary'ci;:;^ ^hel'ev^mSV" ''''■ 
uiate the growth nf .-..,Ji events tend to stim- 

may resulfnhi„h raters oJ' ""i^"^ *^'''""^- *''*'^h 
near-surface wafers -hi hfV""''P«" »"' «f 

type of variabfXin; 'TS l^'""^ '^ '"'^ 
large scales (as well -^tL , u '"•^^o^cales and 
understcxS. "'*' ^''*^' '^^'e) "^ed to be 

Hiom^as: fmS^Shte mer ""'"^ Phvtoplanlcton 
has been shown n hi ."''"*'"'' "^«^^^" color 

.ration^(SctnVj''T9«;^"jr^ 'r*^"' ^«"^«^"- 
and Prieur. N??" where thin'' ' ^^•"'' (M"rcl 
fecting lipht -.hv, rmi . ^""'''''y materials af- 



Preliminary work h L ' 7 " generally, 

surface ph t.m .nklon ^"" "" "'^ "'' "^ "<^«r- 
«ater ...^7^1^ " nT"^'"''"''''"^ '"estimate 
Vcnriek. ir-'^fh," rr^"^"^"> fH^'>ward and 

Mderabic scatter in ihk r'l " u ' There is con- 



17 



h 



® 



^p« 



MMP 






mmm 



fP 



sea surface temperature, daylength, and winds will 
improve regional models (Eppley et al. , 1985) How- 
ever, this uncertainty points to the need for contin- 
ued work on the factors and processes affecting 
water column production. Despite the relative in- 
accuracy of satellite estimates of water column pro- 
ductivity, it is likely that the improved sampling 
characteristics of satellite data will help reduce thesS 
errors It is also clear that any program that intends 
to study global biological processes in the ocean will 
need to employ a variety of sampling methods 
(Smiths/ a/., 1982). 

A program to monitor global primary produc- 
tivity would be a key component in any study of 
global biogeochemistry (National Academy of Sci- 
ences 1984; Butler et al., 1984). Such a program 
»«uld consist of both long-term global monitoring 
by satellites of physical and biological processes 
and a range of detailed in situ process studies. The 
pnmary satellite measurement would be ocean color 
as an estimate of near-surfa.re phytoplankton bio- 
mass. Given the long term scales of variability (such 
as El Nmo-Southem Oscillation events), it is essen- 
tial that these measurements be made for at least 
one decade. Additional data could be collected on 
chlorophyll fluorescence and pigment group abun- 
dance by the same satellite sensor. The monitoring 
and mapping studies bv the satellite need to be de- 
veloped in parallel with in situ process studies 
These measurements would be used to complement 
the sampling characteristics of the satellite as well 
as to dtKumcnt subsurface and other processes not 
directly tibser\able by the satellite. These studies 
would consist of the traditional shipboard sampling 
programs as well as long-term moorings and drifters 
equipped with both biological and physical sampling 
gear. A specific task will be to impnwe our under- 
standing ot the relationship of near-surface phyto- 
plankton biomass to water column productivity 
Other studies arc i>bviously important: grazing, ver- 
tical mixing, and sinking rates are examples of such 
prwesses that will depend on in situ measurements. 
The Coastal Zone Color Scanner has been mak- 
ing useful measurements of near-surface phyto- 
plankton pigment concentrations since late 1978 
(Hovis. I9KI). Much work has been directed to- 
wards development of prwessing algorithms (e e 
Gordon and Morel. 1983). and initial uses of CZCS 
in (>cean<igraphic research have been described 
(Smith et al.. I9S2; Abb«.tt and Zion. I98S Brown 
eial.. I9K5). The basic CZCS algorithm (after cali- 
hr:!tn>n anil removal of atmospheric effects) relies 
t»n the basic fact that the reflectance spectrum of 
water carries mlormation on all of the constituents 
that are suspended in the water 

Although the number of bands available on the 
CZC S IS small, it has been shown that information 
on C(H:colith«:ph<.re presence can be extracted from 
the satellite data as well as chK)r«>phvll pigment con- 
centrations (Molligan <■/ a/.. 1983). ' 



The MODIS instrument as described in this re- 
port will be able to recover the entire reflectance 
spectrum. Aside from the fact that the increased 
number of spectral bands will greatly improve at- 
mospheric correction, this instrument will allow a 
number of new and important measurements to be 
inade from space. First, we will be able to resolve 
the chlorophyll fluorescence peak at 685 nm (at least 
for chlorophyll concentrations >4mg/m') The flu- 
orescence/chlorophyll ratio is known to vary for a 
vanety of reasons, including species composition 
light history, nutrient status, and growth rate (Kie- 
fer, 1973; Harris, 1980; Abbott et al., 1982) Basi- 
cally, the amount of light-absorbed solar radiation 
that goes into fluorescence rather than into the pho- 
tosynthetic system depends on the physiological 
state of the phytoplankton. Although fluorescence 
has been used in many studies as a measure of bi- 
omass (e.g., Denman. 1976), we will be able to 
obtain an independent measurement f biomass us- 

u.L?^-r!*"'* '" '^^''" "^^"^^ (Gordon and Morel, 
1983). Thus, we will be able to investigate the pat- 
terns of the fluorescence/chlorophyll ratio. Such in- 
formation may be extremely useful in estimating 
productivity. * 

Second, we may be able to separate the pres- 
ence of other pigments in addition to chlorophyll 
This will be very useful information, as many of the 
major groups of phytoplankton play different roles 
in nutrient cycling (e.g.. nitrogen fixation by cyano- 
bacteria). Although little in siiu work has been done 
on this measurement, there is at least some promise 
that It can be done (e.g., Mitchell and Kiefer, 1984 
Campbell and Esaias. 1985). Finally, more detailed 
information on the reflectance spectrum will allow 
quantitative measurements to be made in turbid 
coastal waters. Thus, we might be not only able to 
measure pigment concentrations in such conditions 
but also to obtain quantitative estimates of sediment 
concentrations. This would be valuable in studies of 
transport of materials from the land to the coastal 
iKcan. 

There are certainly manv other uses of such 
information besides support for multidisciplinary 
Momc "^ *^'»g'-'"chemical cycles. For example. 
MODIS imagery would be valuable for studies of 
mesoscalc processes (10 to 2(K) km. 2 to 20 days) 
which are very difficult to sample adequately using 
ships. Imagery can also be used to track identifiable 
tcatures over several days in order to infer near- 
surface water vekKJtics. Imagery of phytoplankton 
Momass has also been shown to be useful in fisheries 
research (Laurs etal.. 1984). Finally, real time 
broadcast of color imagery is extremely useful for 
ship-based investigations in order to increase .sam- 
pling cjficiency. There are manv other discipline- 
oriented studies in which satellite imagery of phy- 
toplankton biomass. fluorescence, pigment groups 
and sediment concentrations would be valuable 



18 



® 



wmmmmmm9^ 



mmmmmmmfmm 



7* *■ 



.««! ^. *='*"'?<^e"stic length and time scales of phy- 
toplankton biomass distributions are generally in 

ms'T °^ * *°/o?"y* '"^ » •" 10 ^™ (Steeli 
1978 Denman and Powell, 1984). Thus, the soatial 
resolution of MODIS should be on tte iKfl 
km in the coastal region and 4 km in the open 
ocean The temporal resolution (or revisit time) 
should be nearly one day, given the average cloudi- 
ness over the ocean (about 50 peicem). This means 
that on avera^ a particular portion of the ocean 
will be sampled 3 to 4 times per week. However we 
note that this is an optimistic estimate; some regions 
ami seasons are more likely to be ck>udy than oth- 
ers. The result is a space and time series that is 
irregularly sampled temporally and spatially. For ex- 
ample, on the west coast of North America, clear 
periods tend to be associated with equatorward (or 
upwellmg-favorable) winds. These events are epi- 
sodic, lasting for about 4 to 6 days (Huyer 1983) 
Between events, the winds are less or are eiren u^ 
welling unfavorable, and the ocean is obscured by 
clouds. Also, the ocean west of about 130° W tends 
to be much cloudier than the area closer to the 
coast. Thus, the coverage of 130° W is limited and 
the coverage along the coast biased towards upwell- 
ing conditions This is a fundamental limitation of 
visible and infrared remote sensing that must be 
dealt with. There will be other prt^esses that wiS 
result in gaps in the time series: specular reflection 
from the sea surface (which cannot be avoided en- 
tirely) and low solar radiance at high latitudes (es- 
pecially outside of summer) xvill cause gaps How- 
ever, traditional ship sampling is even more 
irregular: or as long as we are careful with our sta- 
tistical estimates, we can cope with the satellite se- 
ries that are basically oversampled in space and un- 
dersampled m time (and irregular as well), and still 
greatly improve our estimates of phytoplankton 
chlorophyll, global ocean productivity, and oth"; 
parameters. 

Spectral Band Requirements 

During the last several years, considerable ef- 
fort, both theoretical and experimental, has gone 
into the mte.pretation of images of the color of the 

!!?T n^'^'J'^ *'•'' '^ Nimbus-7 CZCS. The goal 
ot the CZCS was to provide estimates of the near- 
surface concentration of phytoplankton pigments 
principally chlorophyjl-a. These estimates are based 
on measurements of the amount of solar radiation 
diffusely backscaitered out of the ocean. Phyto- 
plankton pigments arc strongly absorbing in the 
blue and blue-green regions of the spectrum, in con- 
trast to the green-yellow region. Thus, waters low 
in phytoplankton pigments reflect more blue lichl 
than green, while waters high in pigments appear fo 
reflect mtire green light owing to the selective ab- 
wrption of the pigments. The amount of radiation 
backscattered out of the water and reaching MODIS 



19 



also depends on facton other than the pigment con- 
centration: the absorption by the water itself, which 
IS weak in the blue but strongly increases at longer 
wavelengths toward the red; the amount of solar 
radiation incident on the sea surface in the particu- 
lar spectral band under consideration; and the 
amount of radiance backscattered from the atmos- 
phere. In fact, the radiance backscattered from the 
atmosphere can contribute from 80 to 90 percent of 
^e total radiance received by the sensor, and must 
be removed to determine the radiance backscat- 
tered out of the ocean that contains the information 
concerning the near-surface constituents of the 
water The removal procedure is referred to as at- 
mospheric correction. 

The success of the CZCS program led to propo- 
sals for follow-on sensors such as CZCS-il proposed 
for National Oceanic Satellite System (NOSSrthe 
Ocean Color Imager (OCI) proposed for the NOAA 
senes of polar orbiting satellites, and the OCI pro- 
posed to fly on the French space platform SPOT-3 
Much of the following discussion is based on infor- 
mation generated during those studies. 

Choices of the spectral bands for these instru- 

.'IirrV7Q'lll'''^^-.fe^"y""^°f^''P«riencewith 
the CZCS. The CZCS had only four spectral bands 
lor atmospheric correction and the subsequent es- 
timation of the pigment concentration: 443 520 
550. and 670 nm. The bands at 443 and 670 nm are 
in regions of absorption maxima of chlorophyli-a- 
550 nm is in a region where the phytoplankton con- 
tributes little to the optical characteristics of the 
water. At 670 nm the water is also strongly absorb- 
ing, so the radiant signal at this wavelength oriei- 
nates. for the most part, from the atmosphere, and 
hence this band is used for atmospheric corrections. 
Starting from these highly successful CZCS spectral 
bands, fine tuning was carried out to try to achieve 
a better sensitivity in the pigment determinations, 
and new spectral bands were added to overcome 
difhculties encountered in situations with very high 

rh!Tp"A?',"''^'"''°;?' '^' '^'"' P^°P«^^ sensor, 
the SPOT.3 Ocean Color Imager, like the present 
Coastal Zone Color Scanner (CZCS). is an imaging 
radiometer t^hat will view the ocean in six to eight 
co-registcred spectral bands: 443. 490. 520, 565. 
620. 665. 765. and 867 nm. These spectral bands 
arc chosen to optimize the accuracy of estimatinc 
the near-surface concentration of phytoplankton 
pigments on a global scale. For maximum sensitivity 
to the presence of phytoplankton. spectral bands 
centered at the wavelengths of the maximum and 
minimum of phytoplankton absorption are desira- 
ble; however, a Fraunhofer absorption line (G) falls 
near the maximum of pigment absorption (435 nm) 
and the absorption cwfficient of water itself varies 
rapidly with increasing wavelength in the retion of 
minimum pigment absorption (565 to 610 nm) To 
avoid problems asscKiatcd with these spectral re- 
gions. I.e.. a reduction of extraterrestrial solar 



^ 



irradiance because of the Fraunhofer line and diffi- 
culty in interpretation because of the rapid spectral 
variation in the absorption of water, bands of 20 nm 
width centered at 443 nm (the original CZCS blue 
band) and 565 nm have been chosen. The CZCS 
red band has been moved from 670 nm to 665 nm 
to avoid strong overlap with the in vivo sunlight- 
induced fluorescence feature of chlorophyll-a, cen- 
tered at 685 nm. The CZCS band at 520 nm, which 
seems ideally suited for the detection of very low 
concentrations of suspended material in water, re- 
mains unchanged. At high concentrations of sus- 
pended material this band will saturate and so a new 
band centered at 620 nm, where water is more 
strongly absorbing, has been introduced to over- 
come this effect. 

Dissolved organic material (DOM) (or yellow 
substances) in moderate to high concentrations in- 
terferes with the detection of phytoplankton be- 
cause of its increasing absorption toward the blue 
region of the spectrum but is found only in low- 
salinity estuaries, the outflow from principal rivers, 
and low-salinity seas like the Baltic. Separation of 
the effects of phytoplankton pigments and DOM 
requires very accurate measurements of the reflec- 
tion spectrum of the ocean at wavelengths smaller 
than 443 nm. Bands in this spectral region are not 
included in the OCI because design constraints 
would have made the addition very expensive, and 
their absence will not have a serious impact on the 
proposed mission of the project. 

Finally, two difficulties have been encountered 
with the operation of the CZCS: (I) At high con- 
centrations of pigments or surface suspended sedi- 
ments, atmospheric correction is difficuU because 
significant amounts of radiance can emei^ - from the 
ocean in the red band. (2) At high pigment concen- 
trations the radiance backscattered out of the water 
in the blue is very small, requiring use of the green 
and yellow bands (520 to 550 nm). which are rather 
insensitive to phytoplankton pigments, to estimate 
the concentration. To overcome the first problem, 
two new spectral bands in the near infrared (765 
and 867 nm) arc planned for the OCI, whereas the 
CZCS has only one broad 7(K) to 800 nm channel. 
These bands have been positioned in windows be- 
tween strong atmospheric water vapor absorption 
features (the band centered at 765 nm actually con 
sists of two bands. 745 to 759 nm, and 770 to 785 
nm illuminating a single detector). At these infrared 
wavelengths sea water is approximately an order of 
magnitude more absorbing than at 670 nm. hence, 
the ocean can be considered totally absorbing at 
much higher pigment and suspended sediment con- 
centrations. The second problem is overcome by an 
additional spectral band at 490 nm, i.e., on the 
shoulder of the phytoplankton blue absorption fea- 
ture. This band will be available for use at pigment 
concentrations above those for which the 443 nm 
hand tails. 



The choice of these spectral bands was made 
with the assumption that the OCI would be flown 
on a platform with an AVHRR to provide estimates 
of sea surface temperature in the split window 10.5 
to 11.5 fim and 11.5 to 12.5 jim. In the event that 
the OCI platform does not contain an AVHRR, 
these thermal infrared spectra were to be added to 
the OCI. Finally, if data rate considerations limited 
the total number of bands to eight, the OCI bands 
at 520 and 620 nm, being least critical to the primary 
mission of OCI, could be eliminated, and the band 
at 490 nm moved to 500 nm to provide more sensi- 
tivity at high pigment concentrations. Thus, the re- 
sulting spectral band choice would be 443 nm, 500 
nm, 565 nm, 665 nm, 765 nm, 867 nm, 10.5 to 11.5 
^.m, and 11.5 to 12.5 p.m. 

By greatly expanding the number and range of 
the spectral bands available, the MODIS design can 
remove many of the constraints on the design of the 
systems described above, providing an opportunity 
to observe the complete spectrum of the oceans 
from 400 nm to the thermal infrared. This is impor- 
tant because it will no longer be necessary to limit 
the choice of individual spectral bands for a specific 
function, e.g., 443 nm for estimation of pigments at 
the lower concentrations. Now it will be possible to 
address the problem of recovering the constituents 
of surface water analytically through the recognized 
fact that the information regarding the constituents 
is contained in the entire reflectance spectrum-its 
shape and mean reflectance. Through a judicious 
choice of a few bands it is clearly possible to obtain 
limited, although valuable, information, as with 
CZCS; however, the full potential of ocean color 
remote sensing can be realized only when the entire 
reflectance spectrum is measured. The choice of 
spectral bands for MODIS, with this in mind, is 
made based on two criteria: 

1. Spectral bands should be placed at obvious 
absorption features, e.g.. the maximum (435 
nm) and the minimum (565 nm) of phy- 
toplankton absorption or the sunlight in 
vivo fluorescence peak of chlorophyll at 685 
nm. 

2. Additional bands are required to "fill-in" 
the remaining spectrum throughout the vis- 
ible region such that the information content 
of the spectrum is preserved. 

One obvious addition of spectral bands over 
those described for the instruments above 's for the 
separation of phytoplankton pigments and DOM in 
coastal and estuarine waters. This requires bands in 
the 400 to 440 nm spectral region. A spectral band 
centered at 405 nm with a bandwidth of 10 nm would 
suffice for this purpose and would satisfy the criteria 
above; however, it should be pointed out that g(K>d 
atmospheric correction for this band will be critical 
and is subject to research both pre -launch and post- 
launch. As mentioned above » the maximum of the 



I 



20 



^w 



mirw^m 



3^K 



Tm 



phytoplankton absorption in the blue occurs at 435 
nm, but this wavelength is not used in present de- 
signs owing to the presence of a Fraunhofer absorp- 
tion line in the solar spectrum. Since a state-of-the- 
art sensor could be expected to have considerably 
more sensitivity than conventional sensors, there 
would be no reason to avoid Fraunhofer lines, so a 
band should be added at 435 nm with a bandwidth 
of 10 nm. Again, atmospheric correction will be 
somewhat more difficult with this band than with 
that at 443 nm. 

Other spectral bands that should be added for 
specific purposes include 490 nm for estimation of 
pigments at medium concentration; 520 nm for es- 
timation of pigments at high concentrations and de- 
tection of inorganic suspended material at low con- 
centrations; 620 nm for detection of inorganic 
suspended material at moderate concentrations; 680 
nm to observe the chlorophyll-a fluorescence peak 
centered at 685 nm (placing this band at 680 nm 
rather than 685 nm avoids interference from the 
atmospheric water vapor band near 690 nm); and 
665 nm, 765 nm, 867 nm, and 1,060 nm for atmos- 
pheric corrections. To fill-in the remaining spec- 
trum, the following bands are suggested (in order of 
decreasing importance): 460 nm, 535 nm, 590 nm, 
450 nm, 415 nm, and 640 nm. All bands, except 
those in the near infrared, have a total bandwidth 
of 10 nm. With these bands, the largest inter-band 
gap is 20 nm. A complete listing of these bands is 
presented in Table 5. 

Radiometric Sensitivity 

The radiometric sensitivity of these bands 
should be determined based on the minimum de- 
tectable water-leaving radiance desired and the 
maximum total signal expected. The radiometric 
sensitivitv and digitization should be determined by 
the requirement that the sensor noise level be sig- 
nificantly less than the signal produced by the min- 
imum water-leaving radiance of interest, and ♦hat 
the noise correspond to one-ro-two digital counts. 
Or, equivalcntly, the signal-to-noise ratio based on 
the water-leaving radiance divided by the sensor 
noise should be the basis of deriving the radiometric 
sensitivitv It is reasonable to expect that this ratio 
should exceed 50 under typical conditions. For those 
bands for which the water radiance is not insignifi- 
cant under typical conditions (wavelengths less than 
600 nm), this corresponds to a much higher than 
conventional signal-to-noisc ratio; e.g., at 520 nm 
under tvpical conditions the conventional signal-to- 
noise ratio is 3<K) to 500. The saturation radiance 
should correspond to approximately 1.3 times the 
maximum radiance expected in each band under 
tvpical atmospheric and oceanic conditions. These 
sensitivity and saturation criteria can be easily de- 
termined through simulations. 



Table 5. Proposed MODIS Spectral 

Bands and Priority for the Oceans in the 

Visible and Near-Infrared Regions 





Wave- 


Band- 


Priority 




Buid 


Length 

(nm) 


Width 

(Total nm) 






1 


405 


10 


1 




2 


420 


10 


4 




3 


435 


10 


1 




4 


450 


10 


3 




5 


460 


10 


2 




6 


490 


10 


1 




7 


520 


10 


1 




8 


535 


10 


2 




9 


565 


10 


1 




10 


590 


10 


2 




11 


620 


10 


1 




12 


640 


10 


4 




13 


665 


10 


1 




14 


680 


10 


1 


) 


15 


765 


40* 


1 


! 


16 


865 


47 


1 


1 

9 


17 


1.060 


100 


1 





' This band is "notched* to prevent interference from the 
oxygen band between 76(» and 770 r.n. 



Additional Considerations 

Additional considerations for the MODIS 
design: 

1 . Co-register channels 

2. Active calibration (e.g., diffuser plate) 

3. Depolarize incoming visible radiation 

4. Aerosol measurements 

Other Eos instruments will need to measure 
aerosois for use in MODIS corrections. This 
information probably can be stated as con- 
centration versus latitude band. Current vis- 
ible correction calculations use a 'standard ' 
concentration and therefore a nominal ab- 
sorption for gases such as ozone. This con- 
cern is also relevant for infrared work in the 
3,5 to 4 fim band wheie aerosol absorption 
is a consideration. CO. and sulfur aerosols 
such as those from El Chichon are impor- 
tant. Others should be checked. Again, since 
these are effects governing the rms error 
terms, knowledge of their concentration is 
necessary over relatively large areas. 

5. Infrared corrections 

Infrared corrections can benefit from the rel- 
atively large space scales for atmospheric 
variation, and can derive atmospheric tem- 
perature and water vapor profiles over scales 



« 



ipwu^^r 



f^trmmmmni^^mm^ 



wmm 



"fr 



m 



1 



A ■ 



of tens of kilometers for inclusion in infrared 
calculations for sea surface temperature 
(SST) frontal variability where scales from 1 
through 4 km should be resolved. 

Spatial resolution 

MODIS visible resolution should be on the 
order of 1 km for correction, but can be de- 
graded in stored sample space, if onboard 
tape recording is a limitation, to the order 
of 4 km, as is done for the AVHRR 
instruments. 



Sea Surface %n^eratuie 



r^^}^ "^ AVHRR-derived estimates of SST and 
CZCS-denved estimates of pigment concentration 
has been useful in several studies of mesoscale pro- 

m«x' l!^ ' ^^"^^ "' "'- '^^' Abbott and Zion. 
1985). However, as the CZCS and .WHRR meas- 
urements are not simultaneous, there are some se- 
vere constraints imposed on studies such as those of 
c oud and ocean-feature movement between the sat- 
ellite overpasses, and the necessity to access two 
separate data archives. 

There would be many improvements in both 
?cc?'*^^'^'""* »"<• 'he scientific quality of the data 
It SST and pigment measurements were made si- 
multaneously using MODIS. Both SST and phvto- 
plankton pigment are nonconservative tracers- SST 
can change as a result of heating and cooling', and 
phytoplankton can grow, die. and sink. However 
phytoplankton pigment is a passive tracer whereas 
SST IS intimately linked to the dynamics of the flow 
held In addition, pigment concentrations are re- 
lated nonlmearly to the flow (Denman, 1983) Ex- 
amination of the differences in the behavior of these 
two tracers in mesoscale flow mav allow us to sep- 
arate the effects of phytoplankton growth (and 
death) from fluid flow (Bennett and Denman 198S) 
Initial studies of SST and pigment have shown that 
the evolution of the relationship between these two 
tracers can be used to elucidate manv features of 
the flow held (Abbott and Zion, 1985)1 

As SST can be used as an indicator of physical 
prwesscs, simultaneous measurements of SST may 
improve estimates of productivitv derived from pig- 
ment concentrations such as detection of upwelling 
Finally, information on SST and the diffuse atten- 
uation coefficient (closely related to the pigment 
concentration) may be useful in studying mixed 
layer dynamics and surface transport mechanisms 

In prcKcssing. it would be far simpler to obtain 
both data sets for a given study if the measurements 
wore in the same data stream. Both variables will 
be used for studies of mesoscale pnKesses most 
studies will want access to both data sets rather than 
just one. As both clouds and wean features move 
(clouds typically move several l(K of kilometers in 



an hour; ocean features can move a few kilometers 
in an hour), studies that require direct comparisons 
of both data sets will benefit by having simultaneous 
measurements. Finally, it will be possible to use in- 
formation on atmo-;>heric aerosols as derived from 
the visible imagery to correct for atmospheric effects 
in the infrared imagery. This should result in im- 
proved SST retrievals. 

The most stringent requirement for global SST 
appears to be that stipulated by the World Climate 
Research Program (WCRP) (see Table 6 from Har- 
ries et al., 1983). It is not clear that the needed 
accuracy for -large-scale processes" can be 

xS'f'f 1''^ ^"•'^^ ^^^ '''g''*'y modified AVHRR on 
NOAA-K. L, and M or by the Low Frequency Mi- 
crowave Radiometer (LFMR) on the Navy Remote 
Ocean Sensing System (NROSS). The MODIS, with 
Its better and more numerous windows in the 8 5 to 
12.5 Jtm and 3.6 to 4.3 jtm bands, and with bands 
in the visible and reflective infrared for better cloud/ 
aerosol detection and correction, is expected to en- 
able satellite SST retrievals that meet the WCRP 
requirements. 

Oceanographic experiments in the WCRP are 
categorized in three streams of differing time scales 
(Woods. 1983): stream (1) extended range weather 
forecasting (several weeks); stream (2) short-term 
chmate prediction (several years); and stream (3) 
k)nger term climate prediction (several decades) 
For the studies in stream (1). updating SST charts 
with satellite data collected during the previous 
week or month is considered useful if the interan- 
nual thermal anomalies can be resolved. Stream (2) 
experiments include the investigation of oceanic 
•teleconnections" to the atmosphere that are 
thought to be related to oceanic thermal anomalies 
A major international project called TOGA (Trop- 
ical Ocean and Global Atmosphere) will focus on 
such processes over a period of 10 years or longer 
and a recent TOGA Workshop (Bernstein. 1984) 
recommended a satellite SST accuracy of 3°C for 
monthly means over 2(K) x 200 km areas. Another 
WCRP document (ICSUAVMO/IOC, 1983) listed 
estimates of satellite SST accuracies required for 
-^^o"^" '"vestigations. the most stringent being 
±0.5 C over 5° x 5^ areas. Stream (3) investiga 
tions require consistent long-term global times se- 
ries of highly accurate SSTs. since on decadal time 
scales It appears quite certain that atmospheric tem- 
perature and precipitation changes will be linked 
mextricably to changes in sea surface temperature 
rdatcd m turn to CO, increases in the atmosphere.' 
I he above requirements for •mesoscale processes" 

l^^M^?i'""■''"".''' P''"«''-'««--^" can be met bv either 
MOD S- or AVHRR-bascd (see Tables 1 and 2) in- 
frared measurements, although the more frequent 
coverage by the AVHRR gives it some advantage 
over the MODIS for the dynamic "small-scale- 
processes. 



® 



r 



■liVPPi 



Table 6. Accuracies of Measurements of Sea Surface Temperature 
For the World Climate Research Program 



Large-Sctile Processes 



Absolute temperature accuracy 
Spatial averaging interval 
Temporal averaging interval 
Type of data product 



0.2 K 

200-300 km 

20-40 days 

isotherm contours in map coordinates 



Mesoscale Processes 



Absolute temperature accuracy 
Spatial averaging interval 
Temporal averaging interval 
Type of data product 



1.0 K 

5-10 km 

3.5 days 

isotherm contours in map coordinates 



Small-Scale Processes 



Absolute temperature accuracy 
Spatial averaging interval 
Temporal averaging interval 
Horizontal gradient accuracy 
Type of data product 



2.0 K 

1.0 km 

instantaneous 

0.5 K/1.0 km 

gridded images located to 10 km 



The basic instrument requirements (Tables 13 
and 15) for such measurements can be stated as 
follows: 

1 . Sufficient number of bands in the visible por- 
tion of the spectrum (400 to 1,000 nm) to 
characterize the shape of the reflectance 
spcctruni 

2. Sufficiently narrow spectral bands so as to 
detect (and in some cases avoid) features 
that have a narrow spectral signature 

3. Sufficient digitization and signal-to-noise ra- 
tio to be able to measure small changes in 
the level of the reflectance spectrum, given 
the low albedo of the ocean 



4. 



5. 



At least one polarized channel to improve 
atmospheric correction 

Regular calibration of the enti.e instrument 
system to ensure long-term stability of the 
measurements 



To address the scientific questions briefly de- 
scribed earlier there arc observational require- 
ments as well: 

1 . Given the constraints imposed by cloudiness 
(about 50 percent of the ocean is obscured 
by clouds at any one time) and by the time 
scale of phytoplankton growth, it is essential 
that global measurements be obtained on the 



order of once every two days in order to re- 
solve important temporal fluctuations and 
avoid aliasing. 

The spatial resolution of these measure- 
ments should be 1 km in the coastal zone 
and 4 km in the open ocean, given the char- 
acteristic spatial scale of phytoplankton 
variability. 

These measurements should be ':ontinued 
for time scales on the order of 10 years in 
order to resolve synoptic-scale fluctuations 
(time scales of months to a few years). 

The visible radiance field observed by 
MODIS will include contributions from at- 
mospheric scattering (aerosol and Rayleigh 
scattering) that account for 50 to 90 percent 
of the total radiance; observations should be 
made near local solar noon to minimize er- 
rors in the separation of the aerosol scatter- 
ing, Rayleigh scattering, and water-leaving 
radiance components, as well as to minimize 
the amount of specular reflection (glint) 
from the ocean surface and the amount of 
surface obscured by afternoon cumulus 
clouds and low level morning stratus. 

. The sensor should be able to view either fore 
or aft of the spacecraft (as well as at nadir) 
to avoid areas of glint. 



23 



wf^mmmmmm 






tr? 



6. Methods are needed to caRbrate the sensor 
system periodically, using both internal 
sources and reflected solar radiance suffi- 
cient to ensure accurate measurements of 
ocean reflectance. 

Data from other satellite sensors as well as in situ 
measurements will also be necessary if we are to 
understand the mean and fluctuating components of 
ocean primary productivity. These needs are dis- 
cussed in later sections. 



ATMOSPHERE STUDIES 

Clouds 

There are twq atmospheric phenomena that can 
be monitored directly by MODIS: clouds and aero- 
sols. Both play a major role in climate. In addition, 
they affect other important Earth processes; for ex- 
ample, the geochemical and hydrologicalcycles. 
The following section presents some of the physical 
and distributional characteristics of these constitu- 
ents and their temporal variations as they relate to 
MODIS objectives. Atmospheric temperature and 
humidity fields should be monitored by a sounder 
accompanying MODIS: this will also help in cloud 
held determmation as well as sea and land surface 
temperature determination. This is described in 
Chapter III, Terrestrial Studies. 

Impact of Clouds on Weather and Climate 

Clouds have a major impact on the radiation 
balance of the Earth's atmosphere svstem. Most 
clouds are both excellent absorbers of 'infrared ter- 
restrial radiation and reflectors of solar radiation. 
Clouds reduce the amount of outgoing radiation 
from the Earth's surface to space. Also, clouds are 
a major factor in determining the amount of solar 
radiation reflected back into space. Because of the 
importance of clouds, they arc a crucial component 
in al! meteorological and climate models. In order 
to model clouds, information is required on a num- 
ber of cloud properties. Important characteristics 
are the areal distribution, cloud-droplet size distri- 
bution, cloud-top altitude, cloud temperature, and 
optical thickness. Many of these characteristics can 
bo determined on the basis of multispectral meas- 
urement studies in the visible, near-infrared, in- 
frared, and microwave regions to be supplied by 
MODIS iind other Eus sensors. On the basis of such 
studies, it should be possible to develop a cloud^ 
characteristic climatology and to monitor rainfall 
over land and oceans. These data are important in 
understanding the interrelationships of parameters 
such as sea surface temperature, soil moisture, and 
vegetation index with cloudiness and rainfall. 

Also, these studies should, together with sup- 
plemental sounding data, be useful in operational 



weather applications, rainfall forecasts in particular. 
Because of the global coverage of MODIS, such 
forecasts would be extremely important for hydro- 
logical monitoring in areas where rainfall measure- 
ments are unavailable, i.e., in remote areas and over 
the oceans. The Eos sun-synchronous polar orbit 
Will, however, result in a temporal bias that will 
make interpretation of this data a challenge. 

Relation of Clouds to Atmospheric Chemical 
Cycles 

The measurement of cloud characteristics on a 
global scale will also be important from the stand- 
point of understanding global chemical cycles 
Clouds play a very important role in the transport, 
transformation, and removal of chemical species in 
the atmosphere. For example, clouds are the dom- 
inant mechanism for the removal of the oxides of 
sulfur and nitrogen that are important in the acid 
deposition problem, and for removal of soil dust 
radioactive particles, and particles produced in com- 
bustion processes (soot). It is estimated that on a 
global scale 80 to 9() percent of these species are 
removed by precipitation. In addition, the cloud 
droplets themselves play a major role in the atmos- 
pheric chemistry of many gaseous species. The total 
surface area of cloud droplets is very large and 
consequently, gases can be rapidly absorbed once 
they enter a cloud. Once absorbed, the chemistry of 
many species will be dominated bv aqueous-phase 
processes. With regard to the acid deposition prob- 
lem, important species such as SO., N,0„ and NO, 
may go through fast aqueous transformation in 
cloud droplets; ultimately, the end products of such 
reactions are SO, and NO,, the species that are pri- 
marily responsible for the pH of precipitation. How- 
ever, most clouds do not precipitate; they simply 
evaporate. In doing so th.- reaction products in the 
cloud droplets are converted to solid or liquid aer- 
osol particles. Thus, clouds serve as chemical reac- 
tion vessels that efficiently convert gaseous species 
to other chemical and physical forms; they also are 
efficient in cleansing the atmosphere. 

Clouds are also an important contributor to the 
transportation and mixing of materials in the at- 
mosphere. For example, most chemical species of 
interest, both natural and anthropogenic, are emit- 
ted from the Earth's surface into the planetarv 
boundary layer (PBl,), the top of which is usualK 
dehned by an inversion that restricts exchange with 
the free troposphere. Consequently, the concentra- 
tion of these species and their reaction products is 
much higher in the PBL. The PBL is disrupted bv 
strong convcctive activity, which is usually mani- 
tested by characteristic cloud types. 

Because of the importance of clouds in such 
processes, a recent NAS report (Global Tropos- 
pheric Chemistry: A Plan for Action. 1984) recom- 
mends that a major effort be made to understand 
the role of clouds in tropospheric chemistry. This 



41 



J> 



fW.iyi'ivjii-" 



wmmmm 



w 



effort would be but one component of a major pro- 
gram to develop a Tropospheric Chemistry Systems 
Model (TCSM). An important component of the 
TCSM is a cloud transport-transformation-removal 
model that includes detailed treatments of the phys- 
ical and chemical mechanisms involved. 

The physical aspects of the model would include 
the parameterization of radiation, condensation, 
evaporation, stochastic coalescence and breakup, 
and precipitation development. It is clear that the 
development of such a model and its implementation 
will require an extensive knowledge of cloud distri- 
bution and characteristics on a global scale. Such 
information can be obtained only by remote sensing 
techniques. 



Aerosols 

An aerosol is defined as a gaseous suspension 
of hquid and/or solid particles. The aerosol particles 
found in the atmosphere are formed by two primary 
processes: by direct injection (e.g., sea salt parti- 
cles, dust, soot) and as a product of the atmospheric 
reaction and transformation of gaseous materials 
(e.g., conversion of SO, to H.SO, droplets or to SO, 
salts in particles). 

Aerosols can be broadly classified on the basis 
of their production processes or their composition, 
and their distribution in the atmosphere. On this 
basis, the major aerosol types are classified as de- 
scribed in Table 7. 

The major issues in the field of aerosol science 
today, featuring those aspects that might be ad- 
dressed in a remote sensing program, are as follows: 
climate, geochemical cycles, anthropogenic im- 
pacts, temporal trends in aerosol concentrations, 
characteristics of aerosol distributions, and volcanic 
aerosols. 



Climate 

Aerosols can affect climate in two ways. First, 
the particles can alter the radiative properties of the 
atmosphere directly by absorbing or scattering ra- 
diation. In order to assess the role of aerosols in 
climate, the following are required: the size distri- 
bution of the aerosols, the composition and/or op- 
tical properties as a function of size, the vertical 
distribution in the atmosphere, and the global dis- 
tribution. Second, aerosols serve as condensation 
and freezing nuclei in the atmosphere, and they play 
a critical role in the cloud formation process. 

Current understanding of the climatic role and 
impact of aerosols on a global (or even a regional) 
scale is very poor, A major problem is that few data 
on the distribution and composition of aerosols for 
vast regions of the Earth are available. This is es- 
pecially true for the oceanic areas and for most areas 
of the southern hemisphere. 



Table 7. Aerosol Types 



Defined on the Basis of Composition or Sources: 
Natural 

Sea spray residue 

Windblown mineral dust 

Volcanic effluvia (includes both direct 
particle emissions and products derived 
from the subsequent reactions of emitted 
gases) 

Biogenic materials - particles emitted 
directly and particles produced from the 
condensation of volatile organic 
compounds emitted by plants and trees 
(e.g., terpenes) or the reaction products of 
these gases 

Smoke from the burning of land biota 
Natural gas-to-particle conversion products 
(e.g., sulfates derived from reduced sulfur 
emitted from the ocean surface) 

Man-Made 

Direct anthropogenic particle emissions 
(e.g., soot, smoke, road dust, etc.) 

Products from the conversion of 
anthropogenic gases 

Defined on the Basis of Distribution: 

Tropospheric background aerosols - a 
residue of aerosols in remote locations; 
also includes aerosol produced 
continuously from a large-area source such 
as the ocean 

Stratospheric aerosols - primarily volcanic 
effluvia 



Geochemical Cycles 

Aerosols play an important role in many geo- 
chemical cycles and processes. Examples include: 
(I) The dominant aerosol in the atmosphere on a 
mass basis is sea salt: sea spray is important in the 
exchange of a number of substances between the 
atmosphere and the ocean; (2) Wind-transported 
soil dust, another major aerosol species on a mass 
basis, IS the major non-biological component in 
deep sea sediments in most open ocean regions 
around the Earth; and (3) The aerosol phase of the 
sulfur and nitrogen cycles is the dominant mecha^ 
nism by which these species are removed from the 
atmosphere: indeed, the acid rain problem (dis- 
cussed below) to a large degree is concerned with 
the aerosol-related aspects of these cvcles. 



25 



HP 



wfmmmmmmmmm 



Quantitative assessment of the geochemical as- 
pects of aerosols is very difficult because of the lim- 
ited data we have on the concentration and distri- 
bution of these species over the Earth. This is 
particularly the case for remote areas. In the case 
of soil aerosols, the major sources are located in 
arid regions, which are typically located in remote, 
sparsely inhabited regions. 

Anthropogenic Impacts 

Man is a prolific producer of aerosols and of 
gases that are aerosol precursors. Estimates vary, 
but it is generally agreed that about one-third to 
one-half of the atmospheric flux of oxidized sulfur 
and nitrogen species can be attributed to anthro- 
pogenic sources. This is the cause of the acid rain 
phenomenon. There is also reason to believe that a 
substantial portion of the soil aerosols in the atmos- 
phere is derived from human activity, primarily poor 
land use practices compounded with variations in 
climate. 

Another major aerosol constituent is particulate 
elemental carbon (PEC) that is emitted by combus- 
tion processes. PEC can play an important role in 
climate because the particles are highly absorbing 
throughout much of the radiation spectrum. Be- 
cause of the small size of PEC particles and their 
unreactive character, PEC can have a long residence 
time in the atmosphere and it can be transported 
great distances. The global budget of PEC is essen- 
tially unknown, primarily because of the unknown 
impact of the burning of vegetation in remote re- 
gions where it is a common agricultural practice. 

A major problem is that of assessing the extent 
of anthropogenic impacts. This statement applies 
not only to the problem of determining the quan- 
tities of materials involved, but also to gauging the 
areal extent and degree of these impacts. 

The ultimate example of the climatic impact of 
anthropogenic materials in general and of PEC 
(and, to a lesser extent, soil aerosol) in particular is 
provided by the nuclear winter scenario (Turco et 
fii^ 1983). To a major extent, the assessment of the 
climatic impact of nuclear war will depend on the 
ability to understand and model the production of 
PEC and soil aerosols, their subsequent transport 
and distribution in the atmosphere, and their ulti- 
mate disposition on the Earth's surface. 

Temporal Trends in Aerosol Concentrations 

The residence time of aerosol particles in the 
troposphere is of the order of 10 days. This lifetime 
ii about the same as that of water vapor in the 
troposphere. The similarity in lifetimes reflects the 
fact that clouds and precipitation processes play the 
dominant role in the removal of aerosols from the 
atmosphere. Because of this short lifetime, it is dif- 
ficult to develop a synoptic knowledge of aerosol 
distributions using conventional techniques. Con- 
sequently, remote sensing provides the only hope of 



obtaining a large-scale, integrated picture of aerosol 
distributions. 

The temporal time scales of aerosulrelated 
phenomena range from minutes to years. However, 
on the larger geographical scale, aerosol distribu- 
tions are governed by meteorological processes that 
have lifetimes of days (i.e., individual synoptic 
events). Individual parcels of pollution-derived aer- 
osols can be followed by satellite as they move off 
the northeast coast of the U.S. Also, individual dust 
storms can be followed as they emerge from the 
Sahara and Gobi deserts. Twice in 20 years the con- 
centration of dust over the Atlantic has increased by 
a factor of three in response to drought in Africa. 
Over the long term, the output of particles will vary 
with climate or with human activity. Continuous and 
extended global measurements of aerosol concentra- 
tions and diitributions would enable discernment of 
trends and the identification of the time scales of 
these trend: so as to relate them better to causes 
and effects. 

Characteristics of Aerosol Distrilmtions 

It is difficult to obtain a coherent picture of 
aerosol distributions over the continents because of 
the great diversity of sources and because of their 
temporal variability. Over the oceans, the picture is 
somewhat clearer in a general, broad sense. The 
major conclusion is that the continents are the 
source for many classes of substances found in the 
marine atmosphere. Figure 7 shows the distribution 
of Aitken particles over the oceans. Aitken panicles 
have sizes below 0.1 |xm. They are produced pri- 
marily in combustion processes, having a very short 
residence time in the atmosphere, of the order of 
hours to tens of hours. It is clear that the major 
source of these particles is the continents. Figure 8 
shows the distribution of haze at sea for the summer 
season; the numbers represent the percentage of 
meteorological reports that cite the occurrence of 
haze. Haze is caused primarily by particles in the 
size range between about 0.2 ^im and 1 fim. Haze 
is often enhanced by the hygroscopic action of aer- 
osols which serve as nuclei for the haze particles. 
When relative humidity is high, haze is considerably 
thicker. Even though these data were culled from 
meteorological reports collected before the 1930s, it 
is clear that anthropogenic sources are a major 
source of haze-producing aerosols. It is clear that 
deserts are also a major source of aerosols found 
over the oceans. 

Volcanic Aerosols 

Volcanic aerosols generally play a relatively 
small role in the global budget of aerosol species 
because of the sporadic nature of major events such 
as El Chichon. However, volcanoes can have a great 
impact on climate; only those aerosols ihat reach 
the stratosphere are important in this regard-the 
tropospheric particles have too short a residence 



26 



L 



® 



iHPf?^ 



» . 



iii«ip«*p^^"'«nmi«ni^|iiH|i 



- f Ase IS 

OF POOR QUALfTY 




\ 

\ 



Figure 7. Aitken nuclei concentrations in 10* cm 



:j 



time. The emissions that are most important are the 
sulfur gases that eventually become oxidized to sul- 
fate particles. Over geological time scales, volcan- 
oes have been a major determinant of climate. It is 
important to understand the behavior and transport 
of volcanic materials so that the mechanisms by 
which they affect climate can be better understood. 
Such studies also provide an insight into the mech- 
anisms by which other types of aerosols affect cli- 
mate. Because of the sporadic nature of volcanic 
events, remote sensing techniques are the only 
means by which systematic measurements of these 
emissions can be obtained. 

Curren Status of the Remote Sensing of 
Aerosols 

Visible Imagery (qualitative) 

There are many examples of images showing 
large-scale aerosol phenomena. These include large 
haze patches related to pollutant episodes, smoke 
plumes related to major fires, volcanic eruption 
plumes, and dust storms. 

Infrared Iniagery (qualitative) 

Infrared imagery is used t'^ locate dust stv)rms 
over the deserts: the dust clouds being at a higher 
altitude than the surface, radiate ai a lower temper- 
ature. This procedure has been useful for identify- 



ing in a precise manner the location of major areas 
of soil deflation and relating this knowledge to ge- 
ology, geomorphoiogy, soil characteristics, lard use 
patterns, etc. Carried out in association with mete- 
orological studies, this work also provides an insight 
into the mechanisms for larger scale dust storm 
generation. 

Quantitative Measurements 

Surface brightness values can be used to esti- 
mate aerosol optical thickness. Measurements over 
the ocean have been shown to have an accuracy of 
a few hundredths (Griggs and Stowe, 1984). Meas- 
urements over land are most difficult because of the 
non-uniformity of the surface; nonetheless, recent 
work has shown that the measurements of aerosol 
optical depth by satellite correlate quite well with 
ground measurements carried out simultaneously 
Furthermore, the fine particle mass (d<2.5 ^im) at 
low relative humidity correlates well with the dry 
scattering coefficient and, therefore, with the aero- 
sol optical thickness. 

Ultimately, it should be possible to measure aer- 
osol mass concentratitms from satellites. This will 
require some knowledge of the aerosol size distri- 
bution and optical properties. The size distribution 
over water could be determined by making optical 
thickness measurements as a function of wave- 
length. It should also be possible to estimate the 



t) 



ixi^'- 



SPiP 



i«il«|PM 




o 


« 


a» 


^ 


D 


a. 




< 


S 




>• 




o 


^ 




(0 


r> 


« 




••» 


3 




U' 


01 




? 


3 


o 


3 


< 



2 ^• 

O 3 



S <S 






N 



DC 






3 
Dl 



Ml 



2« 



®, 



m'vm^mm 



I^J 111 VII^V^VPVV. ^1 liV «■ !■ 



absorptive properties of aerosols by making meas- 
urements over relatively bright and dark surfaces 

bee«rem'Svi''.^?f" ^^^^ objectives, it would 
be extremely helpful to have information on the ver- 
ncal d.stnbut.on of aerosols and the vertical disTr^- 

obtam the vertical distnbution of aerosok by lidar 
however there are major difficulties in interJSg 
hdar backscatter because the return is very much 

Aerosols are an inportant aspect of tropos- 
pheric chemistry research. Global scale measure- 
ments of a number of important specierare c- 
qu.red. One of the major uses of thei 
measurements will be to provide data for develop 
ment and validation of large-scale atmospheric 
transport models. It is dear that a network "f 
ground stations will provide a minimal data set fo 

plement the ground network and improve the effi- 
ciency of both the surface and Eos measuring s s- 
^ms. It can provide large-scale interpolation 
between the ground stations. The ground network 
couW also make ancillary measurements that wou d 
serve as surface truth for the satellite 
measurements. "itmit 

ni..? "^ '""*"*'' ^^^^' 'P*''''^' •-•mphasis should be 
^ Ti "*" ^J^^'T^ ^"' '*""'"? '" "'•■■ ^^oastal areas. 

affected by continental sources of aerosol materials 
Aerosol measurements in these regions would also 
be helpful to those studying the water surface co o" 
characteristics, whereas satellite measurements wH 
provide estimates of the mass flux of aeios;;^";^ 
rials to the coastal zone, these data will enable one 
to estimate the possible impacts „„ the marine 
environment. ■■•"mt 



Cloud Measurement Requirements 

. 'T'^^^'^fqu'rements presented below have been 
compiled by dividing the interest in clouds and aer- 
oso s into four categories for removing data seri- 
ously contaminated by cloud and aero^l climato - 
Zn^"r. ^""'"^''"f^ '^'"dies of cloud microphyscal 
properties. In order to meet the goals of these cat- 

ofa sin, '"'" n"J" '"' ^°™ «f measurements 
or assumptions will be required about the state of 
the atmosphere. 

will h!!''^ ^ f "mmarizes specific measurements that 

^^''PJ'''^'^ to >ne or more of the categories 

outlined above. In the more detailed discussion^! 

low. other supportive observation products such as 

beSifieT^'''"''- '"^ "'"'^^"^^ P^"'^"^'^ -*" 

MODIS Requirements for Detection of Clouds 
and Aerosol Editing 

Perhaps the simplest treatment of clouds and 
aerosols ,n the MODIS concept is to identify their 
presence m a binary fashion for discarding a^ da'a 
.n K f" ':""'^"""^'^'d. This is an editing proc-ss 
and whjle this task may sound relatively limpti 
IS not. Differentiating between snow, ice or so ne 
desert surface and clouds requires more than simple 
s-ngle-channei visible imagery Likewise cSs 
^'hose spatial extent is less than 20 percent of the 

ar. difhcult to discern. This becomes especially 
higfSdr"^"""^^'^"'^-"^-'»'^^-ye.y 

oM.l n ^^-"^'t""" Experiment- (Curran. 1982). 
outlines the channels required to examine clouds 
and how each is used. 

sneJJtn"*. '', ^" /"'•*^! "'"^mpt to define some 
sptci K spectral information that will lend itself to 
the cloud/aerosol identification task. Undoubtedly 
some revisions and additions will be necessary, par- 
ticularly for the aerosol layers. " 



Table 8. 



Aerosol 
Cloud 
T(p). 0(p) 

0,,,,., la- er 
Reflected .SW 



-^^!!!y!^J?!?»^!!!lfo>JVI^^ Applications 

Cloud/ 
Aerosol 
Editing 



X 
X 



Surface 
Parameter 
Correction 



Cloud/ 

Aerosol 

Climatology 



X 
X 
X 
X 
X 



X 
X 



Cloud/ 

Aerosol 

Properties 

X 
X 
X 
X 
X 



29 



-O^-^.^.V^.Jh' 



"J "J^PIH^P*»»!»^»'"> ■P*W!^IBPII""W^»WX' 



Cluuiacl 



1 

2 
3 
4 
5 
6 
7 
8 



NOTES: 



TSible9. Spectral Channels for Detailed Observations of Clouds 



Bawl Center 



Baudwidtli 

(Jim) 



Range 

(Reflectance) 



0.7540 
0.7607 
0.7632 
1.131 
1.639 
2.06 
2.13 
10.99 



0.0012 

0.0012 

0.0012 

0.075 

0.10 

0.05 

0.05 

1.15 



0.05-1.1 
0.05-0.9 
0.05-0.9 
0.05-0.9 
0.10-0.8 
0.01-0.4 
0.01-0.5 
N/A 



S/N 



400 
400 
400 
600 
600 
300 
300 
0.18 K* 



IFOV 



1 km 
1 km 
1 km 
1 km 
1 km 
1 km 
1 km 
1 km 



Channel 1-0.7.S40 ^m is non absorbing b\ both ice and watt r 

Channel 2-0.7«t7 |im-Oxvgen A band 0J619 iun 

Channel 3-0.763.-! used for altitude-Oxvgen A band ((.7619 ,im 

cZn'^i ^ iii*"" "'^*' '" '"'*' ""•" *"'"> »' *^'" ^"P^r near cloud top 

Channel 8-*NE^T (a 30() K 



Channels 1, 5, and 9 should enable one to dis- 
criminate cloudiness in the field-of-view of the in- 
strument using shortwave information onlv. Channel 
5 IS required to delineate high-albedo ice and snow 
froni cloud features. Channels 8 and 10 will be used 
to discriminate clouds using their thermal radiation 
signature. Channel 10 will be useful for this purpose 
pnmanly during the night-side pass, while channel 
8 will be useful both day and night. Channel 8 will 
be particularly useful for the detection of optically 
thin, high clouds during daylight hours; these clouds 
may otherwise escape detection from the short 
wavelength measurements. 

Channels 1, 2, 5, and 9 may also yield an indi- 
cation of the presence of aerosol in the atmosphere. 
Channel 9 lies in a spectral region associated with 
little aerosol absorption; channel 2 lies in a region 
asscKiatcd with high reflectance bv aerosol species 
containing iron, while channel 5. as well as channel 
9. IS relatively unaffected bv water vapor 
absorption. 



MODIS Requirements for Correcting Rc- 
scHiree Data for Atmospheric Effects - Correction 
of MODIS surface data for atmospheric effects re- 
quires a knowledge of the atmospheric medium be- 
tween the satellite and the surface. The variables 
required to make this correction are listed in Table 
8. For the purpose of this discussion we will assume 
that all pixels including clouds have been edited out 
leaving only pixels with clear atmosphere or aerosol 
or haze-laden air with optical depths less than one. 

One of two tactics for atmospheric correction 
can be adopted. First, one can attempt to obtain an 
independent measurement of the variables listed in 
Table 8 for each pixel, relying solely on simultaneous 
satellite observations; unfortunately, some of the 
variables may be difficult to observe in this manner. 
Second, one could attempt to rely on some average 
distribution of some or all of the variables to develop 
a correction based on these average distributions 
and any real-time observations that might be avail- 
able. In the following comments it is assumed that 



Table 10. 



Spectral Channels for Editing Cloud 
or Aerosol Pixels 



Channel 


Wavelength 


Bandwidth 


IFOV 




(^m) 


O.OKN) 


(m) 


9 


0,52 


5(KI 




0.7607 


0.(KII2 


500 


.^* 


1.639 


1.10 


5IHI 


10 


3.70 


0.20 


5<HI 


H* 


10.99 


l.(N) 


5<HI 



Sec liihk- 9 



M) 



-. V 



^^mmmmmm 



3 



iilHPffi 



"W 



mem. An open question at this time is whether ««-h 
^^o'^^V^ i'"^ simuhaneitv and sS Su 

of diffenng spatial resolution. While thT^ffcc ,1 "f 
aerosols may be monitored on the highest reSu,io„ 
MODIS measurements, i, is likelv that if wS ^ 

e^rsruS^o^'irsSTnr'^^^^^^^^^ 

SSS " *;^ --'"'cting^nTCl^m 
of aS .V ?^ ;"''"'"""" **^'^ 0"« »he presenc^ 
\1 Iht t:" " *'«^»'^'-mmed. one could then go back 
o the high-resolution data listed in T-.h 7x .!. I 
."dividual corrections, h sS3 i.';^^d^'" ^J^^^^ 

or he effects of aertwols is considerably higher o"f 
umtorm. lo*-reflectance surfaces such as oceans 
than mer highly variable surface features 

A««?^'^ Requirements for Cloud Climaloloev 
frJm SSSI ~in"' '^'""'l. cLmatology Te^ 
irom MUDIS will necessarily be a component in a 
more comprehensive cloud climato^rSe ived 
from several satellites. This is because of the Sd 
eir in; "T'^'"^ *■"' ^ ''""-"vnchronouTit 

will be Si f ' P."^"*^'" *^"^'^'' f^"-" MODIS 
w I be compiled ,n a form suitable for merging wi h 
ii^^a from Harth-radiation-budget and ot^Snll 
meteorological satellites. With'this ^fewS as a 



Table 11. 



Mnni^ 'u" '*^**"^ ^ '^^ of data products that 
MODIS might provide thai would be uVefulTn cloud 
climatology studies (Table 11 ). ° 

ClouT" W^lf ^T"*" *" ^^^'^ Studies of 
v.iouds -While clouds represent an obstacle to 

SSem?. '"'"" '^-^ '"^ -telUte Sey e? 
Sersvrm^-r'~7'"^'""'^'« '" 'he elimat^ 
mntf [h- ^"efore it would seem prudent to 

utilize the opportunity when clouds domSe he 
scenes viewed by MODIS to gather deSd cloud 

p.icafir^^^^iraS;j^^^^^^^^^^ 

property v-ariabies observable from viS ?ea Hn 

mm a NASaT' "T V^^'^ '^ -s exTr'a "t d 
irom a NASA Proposal. 'Cloud Climatolocv and 

Complementary Atmospheric Sounding 

Knowledge of atmospheric temperature and hu 
midity structure and cloud fields is important both 
for correcting MODIS surface observTons for a, 
mospheric effects, and for understandingX ro e o 
he atmosphere ,n the land-ocean-atm4phere sys- 

rmbI?;SrisTs:X^^^^^ 

anc^v for more "t^ Sete":aTora"tm:r 

S fn,m ^''"''"^. channels. This impmvement re 
suits from an ability to better correct for atmo! 

£h '"T '".' '" "•'"^'^ narro.™ t : regTons" 
which are less affected by atmospheric aSorpS 

uon, ont has to decrease spatial resolutinn to 
achieve similar signal-to-noise ratio^, S^^^^^^ 
mosphenc temperature-humiditv profiles are mon 
tored operationally bv NOAA ii^ino th! T 
tionof HIRS ^ ^ur/c. ^ '"^ ^"^ combina- 
"on ot HIRS-., SSU (Stratospheric Sounding Unit) 




Fractional cloud cover 
Cloud height stralihcalion 



Integrated solar radiance 



C loud contaminati(>n 
algorithm 

Cloud contamination 

algorithm ^ window 

radiance 

0. 76 ^m 

».(>6 ^m radiance 

11-14 ^m sounder data 

MODIS SW channels or 
measured directly 



H**» X KM) km 
l^*«' ^ KM) km 



Ouring day 
Day and night 



See spectral channels 
under cloud detection 



31 



® 



mmm 



m^ m 



^mmmmm 



"ff" 



and MSU infrared and microwave sounding instru- 
ments flying on the NOAA series meteorological 
satellites. NOAA-K, L, and M will include the Ad- 
vanced Microwave Sounding Unit (AMSU) in place 
of the MSU and SSU. 

Susskind et ai (1984) have developed a retrieval 
system based on finding solutions of the radiative 
transfer equation representing the observations in 
the channels of HIRS-2 and MSU. The retrieval 
system produces atmospheric temperature profiles, 
and in addition, provides sea and land surface tem- 
peratures, cloud heights and amounts, and ice and 
snow cover. The increased spectral information pro- 
vides the possibility of producing improved surface 
soundings, even at the expense of losing high spatial 
resolution, once thought necessary for finding clear 
columns, because cloud effects can be treated as 
part of the analysis. Completely clear columns are 
not necessary with this technique. 

A recently completed sea surface temperature 
workshop intercomparing monthly mean fields of 
sea surface temperature derived from AVHRR, 
HIRS-2/MSU, SMMR (Scanning Multifrequency 
Microwave Radiometer), and ships and buoys 
shows the H1RS-2/MSU fields to be of comparable 
accuracy to those produced from AVHRR (in the 
rms sense) when compared to ship measurements 
The errors in HIRS-2/MSU fields are more random 
spatially and do not give large areas of small (().5°C) 
but significant anomalies as are observed in the 
AVHRR data, especially in the western tropical 
Pacific. 

Land surface temperatures are harder to mea- 
sure than ocean temperatures for a number of rea- 
sons. The land surface is highly variable with regard 
to both temperature and emissivity In addition, 
land temperatures may have large and variable dif- 
ferences from the air in the PBL. Consequently 
multi-channel regression approaches such as those 
used in operational analysis of AVHRR sea surface 
temperatures may not work as well over land. Even 
the physical approach of analysis of sounding data 
has limitations in the absolute accuracv of retrieved 
ground temperature due to uncertainty in the sur- 
face emissivity 

Ciround temperatures over land are also hard 
to verify or even define, because of high spatial and 
temporal variability Their day-night difference does 
give a good measure of the thermal inertia of the 
ground, which is related to evapotranspiration rates 
and soil moisture. Monthly mean fields of the dif- 
ference between the 3:(M» p.m. and 3:(K)a.m. ground 
temperatures derived from analysis of HIRS/MSU 
show good agreement with climatological tempera- 
ture, moisture fields, and derived cloud fields. Mintz 
et al. (iy«5) have developed a theory relating the 
gri)und temperature differences to evapotranspira- 
tion and soil moisture and have derived reasonable 
fields for a number of months. 



Table 12. Summary of Passive 

Techniques to Detemiine Cloud Physical 

Parameters 



Pftnuneter 



Technique 



Optical Thickness 



Thermodynamic 
Phase 

Particle Size 



Cloud Top Altitude 



Volume Scattering 
Coefficient 



Temperature/Height 



Reflectance at 0.754 fim 
together with theoretical 
relationship 

Reflectance ratio R(1.61 
^m)/R(0.754 jim) 
compared with theory 

Reflectance ratio R(2.125 
fim)/R(0.754|tm)as 
compared to theory 

Agreement in matching 
0.763 Jim altitude and 2.06 
fim altitude from theory 

Agreement in comparison 
of 0.763 \Lm and 2.6 \ixn 
altitude determination 

10-12 Jim split window; 14 
M-m temperature inversion 
product 



The HIRS-2 has channels in the ranges 14.96 to 
13.37 Jim, 11.1 to 6.73 ^m, and 4.57 to 3.75 fim, 
with a resolving power (X/AX) of about 100. The 
spatial resolution of the instrument is about 20 km 
at nadir and 50 km at 50^ look angle. MSU has four 
channels, and a spatial resolution varving from 100 
km to 3(X) km. Two of the channels,' at 50.3 GHz 
and 53.7 GHz, are very important for optimized 
multispectral sounding capability These channels 
enable the determination of ice and snow fields and, 
even more importantly aid the accuracy of infrared 
soundings under partially cloudy conditions. Cur- 
rently retrievals are done on a 125 km grid, and the 
potential exists for going to a 50 km grid with the 
HIRS-2/MSU data. 

Cloud height and fractional cloud-cover fields 
are derived from HIRS-2/MSU data primarily from 
the 14 Jim HIRS-2 sounding channels and the 11 jim 
window. The ice and snow-cover fields are deter- 
mined by a combination of surface temperature 
measurements from the window channels in HIRS- 
2 and the 50.3 GHz surface emissivity as determined 
from MSU. HIRS-2 also contains a channel in the 
near infrared and red. Monthly mean reflectance 
fields, derived from the visible channel, sho^v good 
consistency with the infrared-derived cloud fields. 
In addition, other features are apparent, such as 
deserts and ice and snow effects, though the latter 
arc difficult to distinguish from the clouds at high 
latitudes. Scenes are selected as clear or cloudy 



32 



"T^TT" 



Miifa 



® 



m 



mmmmmmmmm 



^ 



based only on the thermal channels, so as to use the 
same cloud algorithms day and night and not bias 
the day-night cloud difference. To check the accu- 
racy of this procedure, all scenes determined to be 
cloud contaminated were deleted before creating 
the monthly mean reflectance field. The resulting 
field showed no clouds but did have excellent repro- 
duction of the deserts, as well as ice and snow fields 
that matched those determined from the surface 
emissivity and surface temperature. Thus, a multi- 
spectral sounding complement can not only give 
temperature humidity profiles necessary to correct 
MODIS measurements for atmospheric absorption, 
but also can provide accurate estimates of ground 
temperature and its day-night difference, ice and 
snow cover, surface reflectivitv, and cloud cover. 

While the HIRS-2/MSU, or HIRS-2/AMSU 
(an advanced 50 km to 150 km resolution microwave 
sounding unit with more stratospheric sounding 
channels and humidity sounding capability) will be 
flying on operational satellites at the time MODIS 
is launched, it is preferable to have a sounding ca- 
pability either as part of MODUS, or at least on the 
same platform, because humidity and clouds are 
highly variable in space and time (even on scales of 
5 to 10 minutes). The potential exists for significant 
improvement over the current sounding capabilitv 
or that scheduled to fly on NOAA NEXT. In partic- 
ular, a design exists for an advanced high spectral 
resolution (X AX as l,2(K») infrared sounder with a 
spatial resolution of 10 km (Chahinc et ai. 1*^84), 
which will significantly improve sounding capabilitv. 
particularly in the lower troposphere and at the sur- 
face. As shown in Table 4. high spectral resolution 
enables the selection of very clean atmospheric win- 
dows, with atmospheric transmitlance of the order 
of 0.95 even in very humid atmospheres. The pres- 
ence of three clear windows also allows for the de- 
termination of surface emissivicy. The high spectral 
resolution allows for a set of atmospheric tempera- 
ture and humidity sounding channels with much 
sharper lower tropospherie weighting functions 
than those of current svstems or svsiems planned 
for the IW(K. 

Simulation studies have shown atmospheric 
temperature retrieval accuracy to be of the order of 
1 to 1.5 C in up to 90 percent cloud cover (Halem 
and Susskind, I9S4). Retrieval accuracy in the lower 
troposphere will be considerably higher than that 
expected from the operational AMSU-HIRS system 
as currently configured. The accuracy of retrieved 
single-spot sea surface temperatures was shown lo 
range from 0.2C' under clear conditions to 0.«X' 
under 90 percent cloud cover Monthly mean sea 
surface temperature fields should have accuracies of 
a; least 0.2 (' at a 50 km scale. This would further 
increase MODISs utility In addition, ground tem- 
peratures and their diurnal variations sh<>uld have 
accuracies of the order of I ( . These simulation 
studies included the simultaneous use of a micro- 



wave instrument of the quality of MSU to aid in 
cloud filtering. The advanced infrared temperature 
sounder concept instrument would have 28 channels 
in the ranges 16.48 to 14.94, 11.43 to 8.12, 6.06 to 
5.18, and 4.20 to 3.72 jim. The current design calls 
for 10 km spatial resolution with contiguous cover- 
age on an 833 km orbit. Considerable cooling (de- 
tectors to 75 K, instrument to 160 K) is required to 
meet the signal-to-noise requirements for the small 
footprint and narrow bandpass. 

Addition of such capabilities to Eos would 
greatly enhance the experiment. Accurate surface 
temperatures and day-night temperature differences 
can be retrieved at 10 km spatial resolution. The 
MODIS 1 km ground temperature measurements, 
which may have local biases because of uncertainty 
owing to humidity and emissivity effects, can be 
used to interpolate fine structure within the 10 km 
X 10 km box. The high-accuracy lower tropospherie 
temperature humidity structure determination will 
improve the ability to compute surface-atmospheric 
heat and moisture flux. 

Two options exist for incorporation of temper- 
ature sounder-type capabilities data into the Eos 
system. It may be included either as a stand-alone 
instrument, as referred to above, or appropriate 
channels can be added to the MODIS instrument. 
While the intrinsic spatial resolution of MODIS is 1 
km, it is unlikely that the appropriate signal-to- 
noise ratio for high spectral resolution sounding re- 
quirements en be met at that spatial resolution. 
Nevertheless, observations in a number of spots can 
be averaged to give accurate soundings on a de- 
graded spatial resolution. In either event, it is desir- 
able to have a complementary microwave sounding 
capability to aid in cloud filtering and determine ice 
and snow cover. In addition, a proper choice of mi- 
crowave channels will also give rain indications, es- 
pecially in conjunction with the temperature soun- 
der channels. The temperature sounder can provide 
accurate estimates of cloud height and fraction on a 
10 km scale. Such measurements have been shown 
to give good estimates of convective rainfall (Rich- 
ards and Arkin. 1981). Concurrent measurements 
at 10 km resolution, at 37 GHz and 90 GHz, will 
give passive estimates of rainfall (Spencer et ai. 
1983). 

Ozone amount is another atmospheric property 
that affects surface imaging in the near infrared. 
Addition of a co-liKated total ozone monitoring ca- 
pability such as the Total Ozone Mapping Spectrom- 
eter (TOMS) to the Eos platform is also desirable. 
A global total ozone field gives good indications of 
important circulation features such as the jet 
streams and tropopause heights, TOMS can be used 
to further improve the atmospheric sounding capa- 
bility of the temperature sounder(s). While TOMS 
measures ozone only during the day, the tempera- 
ture sounder has day and night ozone sounding ca- 
pability. Comparisons with TOMS data during the 



41 

^1 



33 



g) 



mmmm 



w 



day will give an indication of the accuracy of the 
temperature sounder total ozone measurements and 
indicate whether the accuracy is good enough for 
nocturnal ozone monitoring. TWo-dimensional total 
ozone measurements are essential for these appli- 
cations. Nadir viewing instrumentation such as 
SBUV-2, which will fly operationally on the NOAA 
satellites, is not adequate for this purpose because 
of large gaps in coverage. Limb-viewing ozone 
sounders are not designed to give adequate horizon- 
tal resolution for this purpose and coverage and 
therefore do not satisfy this need. 

The ultimate in vertical resolution and accuracy 
of atmospheric temperature-humidity profiles, as 
well as aerosol distribution, will come from lidar 
instruments, although lidar will not be able to give 
information about sea surface temperature, or 
ground temperature diurnal variations. Lidar, at 
least at first, will not provide the complete spatial 
coverage given by the passive sounders. Simultane- 
ous analysis of spot lidar soundings, with their in- 
trinsically high vertical resolution together with a 
field of passive temperature soundings, should pro- 
vide a much more accurate field of temperature and 
humidity profiles than would be achievable by either 
instrument type alone. The high-accuracy lower 
tropospheric temperature-humidity soundings will 
improve the ability to compute surface-atmosphere 
heat and moisture flux. 



SNOW AND ICE RESEARCH 

Using 1 km visible and infrared imagery, the 
seasonal distribution of snow and ice can be docu- 
mented both on land and oceans during cloud-free 
conditions. MODIS observations will be used di- 
rectly to aid in assessment of snow cover and sea ice 
coverage, and will complement all-weather obser- 
vations made with microwave radiometers (25 km 
resolution globally) and SAR (30 m resolution re- 
gionally). Knowledge of snow cover and thickness 
is important for terrestrial radiation budgets, me- 
teorology, and hydrology, and is an important envi- 
ronmental parameter in ecosystem assessment. For 
the latter application, the availability of snow cover 
extent with measurements of green plant material 
provided by MODIS is of great importance. 

Sea ice coverage has a major effect on air-sea 
heat flux calculations and is a sensitive indicator of 
climatic change. Furthermore, it has important op- 
erational considerations for shipping and fishing ac- 
tivities as well as for identifying locations of ice fea- 
tures for research expeditions. Areal coverage of 
open water within the ice pack and distributions of 
various types of sea ice are of prime concern to ice 
scientists. Distinguishing between sea water and 
ptK>ls of meltwatcr on the surface of the larger ice 
floes is one area where MODIS can contribute sig- 



nificantly to the elimination of ambiguities in micro- 
wave observations. 

The marginal ice zone (MIZ) the area of active 
ice formation and melt, is a region of intense bio- 
logical production resulting from ocean mixing pro- 
cesses associated with the MIZ and growth of algae 
on the bottom surface of the ice. Enhanced biolog- 
ical production associated with ice formation and 
melt leads to a very rich food web supporting large 
fish, mammal, and bird populations. There is a po- 
tentially opposite feedback mechanism operating 
here as well, namely, enhanced production causes 
solar heating to be confined to a shallower depth, 
leading to higher temperatures in the upper few me- 
ters of the ocean, which would temporarily serve to 
inhibit ice formation during the spring and fall. 
Availability of ocean color and surface temperature 
data along with ice extent on a continuing basis will 
enable these interdependences to be investigated to 
a much greater extent than is now possible. 

The MIZ and polynyas are also sites of forma- 
tion of the cold, dense, deep waters of the world's 
oceans. Better knowledge of seasonal and interan- 
nual variability in their regional extent and pro- 
cesses occurring in sites of deep water formation are 
critical for unravelling the role of the ocean in global 
heat flux and climate cycles. The chemical compo- 
sition and radionuclide content of the deep water 
are used to trace its decadal-scale motion and mix- 
ing in the deep ocean basins. Since these properties 
are strongly influenced by biological processes op- 
erating in the upper layers at the sites of formation, 
better knowledge of the initial surface bio-optical 
properties of incipient deep water will be useful to 
the tracer effort. 

The MODIS instrument requirements for snow 
and ice research are exceeded by the requirements 
for ocean color, SST, and land assessment. The 0.5 
to 1.0 km resolution using visible and near-infrared 
bands, with daily coverage in the polar regions, is 
fully adequate to complement microwave sensors for 
distributional assessment. Research applications of 
the remaining infrared bands are also considered 
important. Depth of snow pack is not addressable 
with MODIS. Potentially, discrimination of sea ice 
thickness up to ten* of centimeters may be possible 
based on spectrally dependent reflectance in the vis- 
ible region. For obvious geographic and seasonal 
reasons, use of MODIS in snow and ice observation 
will require high radiometric sensitivity and near 
noon equatorial crossing to deal with the low-inci- 
dent light levels (somewhat offset by the high al- 
bedo) and good atmospheric correction routines to 
deal with the long atmospheric incident path length 
and multiple scattering by the atmosphere. Since 
atmospheric scattering at high latitudes is strongly 
polarized, the pi^larization discrimination ability of 
MODIS will be quite useful in this regard. 



34 



® 



mm 



^ 



OPERATIONAL NEEDS 

The operational needs that could be met by 
MODIS measurements are largely those outlined in 
Chapter II, where complementary operational ca- 
pabilities are discussed. Meeting the requirements 
for studying living marine-resource activities, in 
particular, would depend on MODIS if no Ocean 
Color Imager were carried on an operational space- 
craft during this period. Aerosol distributions and 
aerosol corrections to satellite-derived sea surface 



temperatures are important for operational prod- 
ucts. Plant growth-health indices derived from 
MODIS data are expected to be significant improve- 
ments over the rather crude vegetation index cal- 
culated from AVHRR measurements, and these 
could lay the basis for future operational products 
for agriculture and forestry. The scientific success of 
MODIS in the above research activities covering 
land, oceans, cryosphere, and atmosphere will 
make this instrument a prime candidate to replace 
the currently used NOAA/ AVHRR. 



1 



1 



j 

1 



35 



© 



^mmmmmmmm 






- I 



IV. THE MODIS SENSOR SYSTEM 



BACKGROUND 

As discussed in the preceding chapter, MODIS 
must be capable of conducting global surveys to sup- 
port terrestrial, oceanographic, snow and ice, and 
atmospheric science. The two attributes of MODIS 
that are crucial to its mission are its numerous spec- 
tral channels in the region between 0.4 and 12.0 
>tm, and its revisit time of two days for channels 
viewmg reflected solar radiance and one day for 
thermal channels. The goal is to develop a sensor 
system that will address the widest possible variety 
of research tasks that further the science objectives 
of Chapter III, within the limitations of the available 
resources and consistent with the overall goals of 
the Earth Observing System. 

Heritage for MODIS includes the Coastal Zone 
Color Scanner (CZCS) of Nimbus-7, the Ocean 
Color Imager (OCI) being planned as a follow-on 
to the CZCS, and the various models of the Ad- 
vanced Very-High-Resolution Radiometer 
(AVHRR, AVHRR-2, and AVHRR-3) being used 
on the NOAA scries of operational weather satel- 
htes. Each of these has demonstrated advances both 
in remote sensing technology and in the numerous 
scientific prob»-ms that can be addressed with fre- 
quently repeated. multispectraL 1 km resolution 
global surveys. 

The CZCS is a six-channel imaging spectrom- 
eter :hat has demonstrated the abilitv to convert 
rcmotely-scnscd data to maps of oceanic chloro- 
phyll. The OCI is an eight-channel enhancement of 
the CZCS design. The AVHRR series of sensors 
has been utilized as the operational ocean temper- 
ature sensor and cloud imager during the last six 
years, and will continue as such into the mid-1990s 
through NOAA-K, L. and M. These sensors have 
lully demonstrated an ability to generate global sur- 
veys of a wide variety of ocean*", land, and atmos- 
phere parameters at a resolution of I to a few km 
on a daily to weekly basis. The experience gained 
from the AVHRR and C^ZCS in technology utiliza- 
tion, sensor and data calibration, space operations, 
and ground data process.ng is directiv transferable 
to MODIS. 

In developing the scientific requirements for 
MODIS. !i was apparent that the requirements for 
( 1 ) the ocean color sensing channels to view 2if fore 
or aft of nadir in order to avoid specular reflections 
ot sunlight (glint) from the ocean surface, (2) the 
need for uninterrupted Kmg-term surveys of ocean 
chlorophyll, and {}) the desire lor the terrestrial 
viewing channels to have minimum atmospheric 
path radiance for most applications were incompat- 
ible with a single sensor package. This is especially 
true since the planned polar orbit and \MH) km 
swath width would result in numerous passes along 



36 



the United States coastlines that would include both 
ocean and terrestrial sites in each scan line. There- 
fore, it is proposed that MODIS be implemented in 
two packages to be designated MODIS-T (tilt) and 
MODIS-N (nadir), the former containing the re- 
quired ocean color channels and to be pointable fore 
or aft of nadir, and the latter to contain those chan- 
nels with no requirement for off-nadir pointing. The 
two packages are discussed in some detail in the 
following paragraphs. The sensors are not indepen- 
dent as they will require a joint data multiplexer so 
that the data from T and N can be transmitted 
jointly. 



MODIS-T 

MODIS-T (tilt) will address those science re- 
quirements that call for viewing the surface at pre- 
determined angles forward or aft of the subsatellite 
pomt (nadir). These requirements include: (a) min- 
imizing the amount of specular reflectance of solar 
radiance from the surface, (b) examining the BRDF 
of large homogeneous targets, and (c) pertbrming 
atmospheric studies by examining the spectral signal 
at several optical depths The high-priority ocean 
color requirements result in a need for both fore 
and aft pointing in P to 2^ steps to a maximum of 
J* . a set of 17 spectral channels in the visible and 
near-infrared region ((1.4 to 1.0 ^m) (see Table 5) 
a spectral width of approximately 10 nm for all chan- 
nels, a signal-to-noise minimum of 6(K):1 for the 
visible channels, frequent revisits, and acquisition 
ot long-term global data sets. The oceanographers 
on the Panel stated that spatial resolution of 1 km 
m the coastal regions is sufficient, and that the res- 
olution in open oceans could be reduced to 4 to 10 
km. It was decided, however, that any such reduc- 
tion in resolution would best be done durimz cround 
processing. The BRDF requirements include view- 
mg to angles as large as 60^ both fore and aft Other 
requirements, both for BRDF and for the proposed 
atmospheric studies, are less restrictive and there- 
fore fit within those for ocean color 

Sensor Concept 

The MODIS-T requirement for a minimum of 
17 spectral bands with 10 nm width, and the desire 
tor additional bands in the region from 0.4 to 1 (i 
M-m, can be satisfied bv any of' several types of im- 
aging radiometers. A practical system in terms of 
size, complexity, technology availability, and overall 
utility IS that of the imaging spectrometer shown in 
Figures 9 and 10. The system consists of a crosstrack 
scan mirror, collecting optics, spectrometer, and a 
M X 64 element silicon detector array, with the 



r1 



@ 



SCAN MIRROR 



FIELD STOP 



SPECTROMETER 




WHISK-BROOM TRACK 
^9.47 sec PERIOD 



Figure 9. IV10D1S-T scan geometn and conceptual system layout. 



entire system capable ot :tf>(P rotation about the 
optics axis to give the required fore-aft tilt. The scan 
mirror views the ^f swath required for two-day 
global coverage in ^.5 seconds, the time required 
tor the platform subsatellitc point to advance 64 km. 
The collecting telescope can use either rettective or 
refractive optics or a combination of the two. The 
spectrometer disperses the beam from the entrance 
slit along one dimension of the djtector array. 
Therefore, the image of the spectrometer slit on the 
surface consists of 64 elem'^nts along track, with 
each element being dispersed into 64 perfectly reg- 
istered contiguous bands of approximatelv 10 nm 
width covering the range from 4(K» to l.iKM) nm. 

The parameters used to calculate the signal-to- 
noise anticipated for MODIS-T are listed in Table 
I.V It is worth noting that the entrance aperture 
(telescope diameter), which sizes the entire system, 
is only 5.0 cm. Also, note that the signal-to-noise 
calculations were made using a sensor Fook angle of 



20' and an atmospheric model with total nadir op- 
tical thickness of 0.72. 

The calculated signal-to-noise ratios for 25 of 
the 64 channels of MODIS-T, using the parameters 
listed in Table 13. arc tabulated Tn Table 14. The 
resulting numbers arc in excess of l.(HMI: 1 from 4(MI 
thrtuigh 540 nm and greater than 3(K>:I for wave- 
lengths shorter than ^M) nm. 



MODIS-N 

MODIS-N (nadir) will address those scientific 
tasks that do not require the system to be pointed. 
The requirements generated in Chapter III cover 
the spectral range from 0.4 to 12.0 ^m and include 
a strong justification for 500 m nadir resolution in 
several channels in the visible, near-infran d. and 
shortwave infrared. M()I)IS-N spectral wi(» hs vary 
from 1.2 to .5(K) nm. These requirements u suit in a 



1 



37 



® 



DETECTOR 
ARRAY 164x64) 




f I 

41 



4 



HOLOGRAPHIC GRATING - 

Figure 10. Conceptual optical system for MODIS-T (Shafer, 1981). 



system that is more complex than MODIS-T hut 
capable of solving a wide variety of problems in the 
areas of terrestrial ecosystems, climatology, and 
oceanography. 

Sensor Concept 

The requirements outlined above result in a sys- 
tem with at least 35 spectral bands. Owing to the 
range of spectral widths, and a requirement to mea- 
sure polarization, it is impractical to use the imaging 
spectrometer concept of MODIS-T Instead, a more 
conventional imaging radiometer is envisioned. This 
system consists of a crosstrack scan mirror and col- 
lecting optics similar to those shown in Figure 9 for 
MODIS-T and a set of individual detector elements 
with spectral interference filters located in the focal 
plane of the collector. The layout of the focal plane 
is illustrated schematically in Figure 1 1, which shows 
the detector/filter layout for 36 channels, IS main- 
tained at a temperature of 3(K) K and IS at SO K. 
Each of the 24 l,()(K) m resolution channels has a 
single filter/detector module, whereas each of the i2 
5(K) m resolution channels has four detector ele- 
ments with a single spectral filter over alt of them. 
Sinre the platform advances only I km along track 
during the swath scan period (I4S milliseconds), 
which is 64 times faster than that of MODIS-T. the 
optics for MODIS-N must be larger than those of 
MODIS-T in order for the system to collect the re- 
quired number of photons in the limited lime 
available. 

Signal-to-noisc calculations for M()I)IS-N were 
made using an optics diameter of 40 cm. rhis results 
in a scan mirror as large as 40 x 104 cm. Ihe t>ptics 
size could be reduced by increasing the number of 
detectors for each band and thereby scanning more 
than one line for each pass of the mirror. This, how 
ever, increases the mechanical and electrical com- 
plexity of the fociil plane and the amount of testing 
and calibration required. This tradeoff must be 
examined. 



Examples of the parameters used to calculate 
expected signal-to-noise are listed in Tables 15 and 
17 for a shortwave infrared and thermal infrared 
channel, respectively. The results for these and the 
remainder of the MODIS-N channels are listed in 
Tables 16 and IS. Typical, but by no means com- 
plete, applications are also listed for each band. 



Table 13. MODIS-T Parameters for 
Sensitivity Calculations 



Satellite Height 
Ciround Resolution 
Swath Width 
Wavelength 
Spectral Bandwidth 
Solar Zenith Angle 
Sensor Look Angle 
Optical Transmission 
Detector Size 
Fele scope Diimieler 
Optical f-Number 
Surface Relkctivity 
Quantum lifticiency 
Saturation Radiance 

{Integration TinuM 

(Dweinimc) 
Time \o Map the Earth 

Number of Detectors Per 
Spectral Band 

Scanning E^fficicncy 

Expected Ni:D() 



705 km 
l.(HK)m 
I..S|3km 
See Table 14 
10 nm 

20 
0.1 
104 ^m 

5.0 cm 
1.5 

See lable 14 
See lable 14 

2.1 mw cm - 
sr-fim 

1.0 

2 days 

M 

0.S5 

See Table 14 



}H 










Table 14. MODIS -T Twenty-Five of Sixty-Four Spectral Bands 



h f 



II \ 



Band 



Wavelength 

(nm) 



1 


410 


2 


420 


3 


430 


4 


440 


5 


450 


6 


460 


y 


490 


12 


520 


14 


540 


17 


570 


19 


590 


22 


620 


24 


64(» 


27 


670 


2S 


680 


30 


7(M» 


34 


750 


37 


780 


4(1 


8(K) 


44 


840 


47 


870 


50 


9<M) 


55 


950 


60 


IJKN) 


64 


K040 


C'tncr tins row 


ot detector 



Detector 
Quantum 
Efficiency 



0.43 
0.44 
0.47 
0.50 

0.51 
0.52 
(».57 
0.60 
0.63 
0.66 
0.67 
0.67 
0.67 
0.67 

0.67 
0,66 
0.64 
0.59 
0.57 
0.43 
0.39 
0.29 
0.16 
(».07 
0.(K) 



Water 
Reflectance 



4.7 
5.0 
5.0 
5.1 

5.1 
5.1 
3.5 
2.8 
2.1 
1.7 
1.0 
1.0 
1 () 
1.0 

1.0 
1.0 
1.0 
1.0 

t.o 

1.0 
1.0 
1.0 
1.0 
1.0 
1.0 



S/N 



Comments 



l;irk 



MOO 
1,120 
1,140 
1,170 


Dissolved organic material 

Dissolved organic material 

Dissolved organic material 

Chlorophyll absorption 
maximum 


1,180 




1,2(K) 




1,180 
1.070 
1,030 


High chlorophyll concentratio 
Low suspended sediment 


980 
-930 


Chlorophyll minimum 


-870 
-840 


High suspended sediment 


-790 


Chlorophyll absorption 
maximum 


-780 
-760 


Chlorophyll fluorescence 


-670 


Atmospheric correction 


-600 
-560 


Atmospheric correction 


-490 




-430 
-350 


Atmospheric correction 


-230 


H:0 


120 




-0^ 





4 



current m.m.tor NOTC: RdKt.incc ohta.ncd Irom Wolfe ;md Zis.is. I^>7S 



>f 



CALIBRATION 

C onversion of digital counts to radiance enter 
mg the sensor requires complete characterization of 
the system prior to launch; including response to an 
extended standard source, in-flight system monitor- 
mg and response to known sources, and constant 
vigilance over the life of the mission to detect 
changes m the system and to interpret and compen- 
sate for these changes. 

The principle standards for prelaunch radio- 
metric calibration will be a visible and near-infrared 



integrating sphere similar to those used for MSS 
FM, CZCS. and AVHRR, and ?. calibrated black 
body. 

In-flight visible and near-infrared calibration 
has typically taken the form of monitoring system 
response to incandescent lamps and referring'thcsc 
changes to the prelaunch values. Thermal in-flight 
cahbration usually is based on viewing a black body 
built into the backsean portion of the sensor and the 
near-zero temperature of space. Inclusion of an ap- 
erture filling visible and near-infrared sources that 
IS external to the sensor is highly desirable for both 



39 



s) 



4 



FLIGHT 



1 


1 


















6 


6 


1 


1 


















6 


6 














13 










18 






T = 300K 








SILICON DETECTORS 



S/C ADVANCE 
IN 1 MIRROR 

SCAN 
» 



SCAN 



19 


19 


















24 


24 


19 


19 


















24 


24 






i 

76: 










31 










36 



T = 80K 

Fi}>ure II. MODIS-N focal plane layout konccptuah showing 18 cooled (JM» K) an«] 18 un- 
cookd iJm K) spectral channels - Channels 16 and 19-24 have 5iM» m spatial resolution. 



MODIS-N and MODIS-T, but may be difficult for 
the former owing to aperture size. Inclusion of more 
than one blackbody in MODIS-N is also desirable. 

For many of the planned applications of 
MODIS, the spectral radiance reflected from the 
surface is compared against the incoming solar spec- 
tral radiance. Direct measurement of the solar ra- 
diance using a calibrated diffuse reflector bypasses 
many of the difficulties inherent in radiometric cal- 
ibration. Therefore, inclusion of a deployable cali- 
brated diffuser is required for M()DIS-T and highly 
desirable for MODIS-N. The 4(1 cm MODIS-N op- 
tics diameter will make the latter difficult. 

The third type of in-flight radiometric calibra- 
tion is accomplished using suifacc targets that ar*; 
well characterized and located in areas with a gen- 
erilly clear atmosphere. Careful measurements of 
the surface and atmosphere at the time of spacecraft 
overpass coupled with radiative transfer modeling, 
can result in accurate sensor radiometric 
calibration. 

In addition to in-flight radiometric calibration, 
it is imperative that any post-launch degradation in 



the spectral and spatial response of the system be 
known. Spatial response may best be measured us- 
ing known surface targets; as with radiometric sur- 
face targets, atmospheric measurements and radia- 
tive transfer modeling must also be utilized. In- 
flight spectral characterization may be accomplished 
using absorption lines in rare Earth glasses or input 
from light emitting diodes (LEDs). 

Precise cross-calibration of MODIS-T and -N is 
required since the science algorithms will use chan- 
nels from both sensors, A single calibration source 
will be used prior to launch. Methodology for in- 
flight cross-calibration will be developed using si- 
multaneous observations. 

Use of as many of the calibration techniques 
described above as are practical for both MODIS-N 
and MODIS-T will ensure the greatest possible sci- 
entific return for the resources expended. 

DATA RATES 

Eos guidelines call for a UK) percent duty cycle 
for MODIS. This implies that the thermal channels 



40 



® 



v*l^>>^ 



TaWe 15. MODIS-N - Example of Performance Calculations (Channel 25) 



Satellite Height 
Ground Resolution 
Swath Width 
Wavelength 
Spectral Bandwidth 
Solar Zenith Angle 
Sensor Lo<jk Angle 
Optical Transmission 
Detector Size 
Telescope Diameter 
Optical f-Number 



705 km 
500 m 
1.513 km 
2.13 Jim 
20 nm 
22° 
0° 

0.35 
382 Jim 
39.8 cm 
1.35 



Surface Reflectivity 


0.10 


Quantum Efficiency 


0.40 


Saturation Radiance 


0.71 mw/cm--sr- 




M-m 


(Integration Time)/ 


1.0 


(Dwell Time) 




Time to Map the Earth 


2 days 


Number of Detectors Per 


2 


Spectral Band 




Scanning Efficiency 


0.25 


Calculated 


247 



r J 



41 

4>1 



Channel 



Table 16. MODIS-N Visible/Near IR Channels (Preliminary) 



X 

(nm) 



AX 

(nm) 



IFOV 

(m) 



Surface 
Reflectance 



S/N 
Calculated 



Comments 



1 


470 


20 


5(K) 


3 


(B) 


740 


Soil- Vegetation Differentiation 


2 


550 


20 


5(K» 


10 


(B) 


920 


Green Peak Chlorophyll 


3 


670 


20 


5(K) 


6.5 


(B) 


770 


Chlorophyll Absorption 


4 


710 


20 


5(M) 


9 


(B) 


8.10 


RED-NIR Transition 


5 


880 


20 


5(M) 


25 


(B) 


8.50 


Vegetation Max Reflectance , 


6 


960 


20 


5(K) 


24 


(B) 


520 


H,0 Peak 


7 


435 


10 


l.(KN) 


5.1 


(C) 


1.4X0 


Low Chlorophyll , 


8 


490 


10 


l.(NM) 


3.5 


(C) 


1.520 


Nonlinear Chlorophyll 


9 


520 


10 


l.(NN) 


2.8 


(C) 


l..'^90 


High Chlorophyll 


10 


565 


10 


l.(NM) 


1.8 


(C) 


1.290 


Chlorophyll Baseline 


11 


590 


10 


l.(MN) 


0.6 


(C) 


1.160 


Sediment 


12 


665 


10 


l.(MM) 


0.17 


(C) 


950 


Atmosphere/Sediment 


13 


765 


10 


l.(MM) 


0.1 


(C) 


720 


Atmosphere Correction 


14 


865 


10 


l.(XH) 


0.1 


(C) 


470 


Atmosphere Correction 


15 


754 


1.2 


l.(M)0 


30 


(D) 


920 


Cloud Altitude 


16 


761 


1.2 


l.(MM) 


90 


(D) 


1..S.50 


Cloud Altitude 


17 


763 


1.2 


l.(MN) 


50 


(D) 


1.160 


Cloud Altitude 


18 


5(H) 


KM) 


l.(NM) 


2.5 


(B) 


2.880 


Polarization 


19 


5(M) 


KM) 


l.(NN) 


2.5 


(B) 


2.880 


Polarization 


20 


1.080 


20 


5(K) 


25 


(B) 


1.120 


Leaf Morphology 


21 


1.131 


20 


5(K) 


10 


(A) 


520 


Cloud H,0 Absorption 


22 


1.240 


20 


5(M) 


10 


(A) 


7.50 


Leaf H,0 Absorption 


23 


1 550 


20 


5(K) 


14 


(B) 


480 


Leaf H,0 Absorption 


24 


1,640 


20 


5(M) 


10 


(A) 


375 


Snow/Cloud Differentiation 


25 


2.130 
(A) Colwell 


5U 

. mi 


5(K) 
(B) HxKkeial. 


10 
I9R4 


(A) 


2-50 


Cloud Penetration 


References: 


(O Wolfe and Zissis. WH 


(0) personal (ommuniciilion. W'.L. Barnes 



41 



® 



'''^Me 17 MODIS-N Thermal Channel S/N Calculation (Channel 35) 






Satellite Height 
Ground Resolution 
Swath Width 
Wavelength 
Spectral Bandwidth 
Solar Zenith Angle 
Sensor Look Angle 
Optical Transmission 
Optical Depth of Atmosphere 
Detector Size 



7(J5 km 
l,(KK)m 
1,513 km 
12.0 ^lm 
0.5 ^lm 
22"" 
if 

0.35 
0,10 
763 p,m 



39.8 cm 

1.35 

270 K 

4.8 amps/watt 



Telescope Diameter 

Optical f-Number 

Surface Temperature 

Responsivity 

(Integration Time)/(Dwell Time) 1.0 

Time to Map the Earth 2 days 

Number of Detectors Per i 

Spectral Band 

Scanning Efficiency u,25 

Expected NEAT o.on K 



Table 18. 



Channel 



26 
27 
28 
30 
33 
34 
35 



X 

(nm) 



3,750 

3,959 

4,050 

8,550 

10.450 

11.030 

12,020 



AK 

(nm) 



MODIS-N Thermal Cha nnels (Preliminary) 

Comments 



90 
50 

50 
5iH) 
500 
500 
500 



IFOV 

(m) 



ISKh 
1,0(X) 
l.O(K) 
1,0(K) 
1,0(K) 
1 .0(K) 
1,(KK) 



NEAT 

(K(a 270 K) 



0.14 
0.14 
0.13 
O.OI 
0.01 
0.01 
0.02 



* Temperuiurc 



Clouds & Surface Temp* 
Clouds & Surface Temp 
Clouds & Surface Temp 
Stratospheric Aerosol Detection 
Stratospheric Aerosol Detection 
Clouds & Surface Temp 
Clouds & Surface Temp 



will be on at all times and the reflected solar chan- 
nels will be on for one-third of each orbit. Assuming 
lO-bit digitization, 40 percent over-sampling in the 
crosstrack direction, anc contiguity at nadir, 
MODIS-N, with 24 channels having KCXK) m reso- 
lution and 12 channels with 5{)() m resolution, will 
output 76 Mbs (megabits per second) in daylight 



and 1.2 Mbs during the remainder of the orbit. 
MODIS-T using the same assumptions and output- 
tmg 17 of its 64 spectral channels, has a data rate of 
1.2 Mbs (day only). These rates resuh in a total of 
3.4 X 10'' bits per day This is equivalent to 200 
CCT (computer compatible tapes) per day (at 6 250 
bits per inch (BPI)). 



42 






®, 



V MISSION OPERATIONS REQUIREMENTS 



TILTS 

The MODIS-T instrument will be capable of 
operation with a tilt of 60" fore and aft as well as 
nadir. The entire instrument will be tilted, not just 
the scan mirror, which will lead to a scan pattern 
different from that of the present instruments. The 
time to tilt from fore to aft of 20° over the (Kean to 
avoid glint will be a few seconds, so as to minimize 
the data loss. Tilts in excess of 20'' will be used over 
land for BRDF measurements. 

The onboard priKessing system would know the 
position of the sunglint point, and could control the 
acquisition along the scan to acquire the data on the 
appropriate part of the scan, even near the ghnt. 



which 17 of the 64 available bands would be ac- 
quired. It would also permit control of the resolu- 
tion of the data acquired, i.e., 1 km in coastal areas 
and 4 km in the open tKcan. Over land a few se- 
lected bands (or synthesized bands by summing se- 
lected channels) from the fore- and aft-looking in- 
strument could be acquired to provide some 
bidirectional information. 

The system would have a minimum solar illu- 
mination angle defined to identify Earth day or night 
and the solar reflectance channels would not be ac- 
quired at night. The world map could have an effect 
on the definition of acceptable angles since some 
very-low -angle data may be desired over the ice- 
covered regions. 



GAINS 



CALIBRATION 



The required accuracies can be achieved with 
10 to 12 bits digitization. If 12-bit digitization is 
achieved, there will be no need for commandable 
or programmable gain control within the data sys- 
tem If 12-bit quantizing is not included, then it may 
be necessary to have some commandable gains to 
minimize the quantizing noise in the signal. These 
would be selected using the solar illumination angles 
and world map information to best use the available 
dynamic range of the analog-to- digital converter. 



ONBOARD PROCESSING 

The onboard prcKessing system of MODIS will 
have information about the current and future piv 
sitions of the spacecraft, the attitude t>f the space- 
craft and the sensors, sun position, and surface il- 
lumination angles. The system will store a world 
map that identifies some essential characteristics of 
the Earth in regions varving in size from approxi- 
mately 30 km X 30 km to 300 km x 300 km. de- 
pending OP their liKation. Each region car. identify 
between four and eight different surface cases, such 
as deep ocean, coastal water, estuaries, barren land, 
natural vegetation, cultivated land, nominally snow 
or ice. urban areas, special investigation region, etc. 
The specific bands selected for acquisition and any 
processing for bandwidth reduction can be opti- 
mized for each of the classes. This would primarily 
apply to the MODIS-T instrumentation, selecting 



The various applications of the data from 
MODIS will require very precise calibration of the 
system both before launch and in flight. Both the 
MODIS-N and -T instruments are scanners, and this 
will allow calibration of the total optical system by 
observing space before and after the Earth scan, and 
integration spheres and/or black body targets during 
the back scan. Occasional looks at a diffuse solar 
reflector or the lunar surface could provide addi- 
tional useful calibration information (see Chapter 
IV, Calibration). Every two years, when Eos is vis- 
ited by the Shuttle, a recalibration using STS-borne 
precision calibration systems should be done. 

Electronic systems will be used to check the 
linearity of the signal processing electronics and 
step-size uniformity of the analog-to-digital 
converters. 



OPERATIONS 

The MODIS instrument will routinely acquire 
data over the world using ii stored acquisition strat- 
egy that is a function of its world map. The strategy 
can be changed in-flight by loading new tables into 
the onboard processing system. In addition, there 
will be a limited number of investigations of specific 
test areas underway at any time. These will be iden- 
tified in the world map with special identifiers, and 
Jata acquisition will be optimized for these 
investigations. 



43 



t) 



VI. GROUND SYSTEM PROCESSING AND ARCHIVING 

REQUIREMENTS 



The processing requirements for MODIS are 
defined by considering both the volume of data that 
will be produced by the instrument and the wide 
range of investigations with their attendant differ- 
ences in both spectral and spatial requirements. The 
diversity of applications, both for land and water, 
leads ti» the generalized case where each application 
can require a separate product tailored to the phys- 
ics of the observables. Given the large volume of 
data that will be produced by such an instrument, a 
balance is needed between achieving a suitable 
breadth in product generation and restricting the 
computation to usable, e.g., cloud-free, observing 
periods. 



OVERVIEW 

The MODIS Panel recommends a multi-tiered 
approach to processing and analysis of the instru- 
ment data. A layered network should be established 
to meet the need for a limited number of widely 
used standard products. Simultaneously the low 
level (Level (1 or data 1) would be distributed to 
local or regional processing centers for generation 
of research or specialized products. This approach 
has the potential to make optimal use of the im- 
proved satellite data handling and analysis capabil- 
ity that is projected to exist within interested labo- 
ratories and institutii>ns. Thus a central requirement 
necessary to implement this strategy is the dehnitii^n 
and implementation of a network topology. 

The network can serve multiple uses ranging 
from simple catalog searches to distribution of the 
low level satellite data for further priKCssing. While 
the network concept provides a flexible mechanism 
to generate a range of products, distributed access 
to supporting data bases (satellite management, 
sensor instruments, allied instruments) needs to be 
managed in a coherent manner. Data base coordi- 
nation with respect to the sensors (MODIS. the plat- 
form, allied instruments) and the central and dis- 
tributed processing centers is a prime concern. The 
range of science questions that will be addressed by 
MODIS requires provision for synergistic melding of 
liarth- and space-based observations. Studies ad- 
dressing thev? concerns have been initialed by the 
I:os Data Panel. 



LEVELS OF DATA PROCESSING 

The MODIS Panel has formulated a set of level 
definitions that are used to provide a framework for 
discussion of data products (see Table 1*^). 



MODIS ARCHIVAL AND 
DISTRIBUTION REQUIREMENTS 

Level I B data, together with the cloud and land/ 
ocean masks, would be archived permanently. The 
routine Level 3 products given above would also be 
archived permanently 

The Eos Data and Information System would 
have the responsibility for producing user-specific 
products on request, and would utilize the Level IB 
data for this. An example of such user-specific prod- 
ucts are those for investigators who would like to 
develop improved algorithms, or algorithms for new 
properties, requiring the generation of data sets of 
particular regions over varying time periods. Storing 
the Level lA data in geographically useful regions 
(Level IB) would facilitate this, as well as requests 
for distribution of Level I A data. Anticipated an- 
nual data requests arc listed in Table 21. 

A substantial application for MODIS data will 
stem from current practice of use for AVTIRR and 
CZC'S data where daily coverage of large geograph- 
ical regions is used to study evolution of the physical 
and biok>gical regimes of the ocean's upper mixed 
layer. Daily coverage is necessary for avoidance of 
clouds or other contaminating features and to re- 
solve the space time variability inherent in the 
observables. This class of investigation is currently 
undertaken in a number of ocean-oriented research 
institutions. Daily coverage is also utilized for com- 
putation of LAI by National Aeronautics and Space 
Administration Goddard Space Flight Center 
(NASA GSFC). Basin-wide and global analyses are 
undertaken utilizing GAC data from the NOAA 
platforms for both terrestrial and oceanic applica- 
tions. These studies are increasing in scope and 
number of investigators with the advent of low-cost 
high-coverage-frequency access based on data dis- 
tributed bv domestic communications satellites 
(DOMSAT). As global programs such as TOGA, 
wort (World Ocean Circulation iixperiment). 
Cilobal Flux. Eos, etc.. enter the research arena. 
the need for rapid and varied access to satellite 
products increases and the use of large-scale, syn- 
optic observations of a region or process becomes a 
routine and indispensable component of a complete 
observation program. 

Data cost and access have been and probably 
will continue to be prime factors influencing the use 
of satellite observations. Since the user community 
spans an experience and capanility range from inose 
who need simple images of prepriKCssed data to 
those whose research requires large volumes of 
Level I A data, the range of capabilities afforded b\ 
networking various processing organizations 



44 



® 



TaUe 19. Definition of MODIS Data Products Levels 



Level Level data represents the basic telemetry stream as received from the spacecraft for the 

MODIS instrument. 

Level 1 A Level lA data contains MODIS instrument data augmented with all ancillary data necessary 

to compute Earth-located, geophysical parameters. Potential ancillary parameters include 
calibration information, satellite ephemeris, attitude , time, sensor information (gain, tilt, 
channel selection). Sufficient information or pointers to easily accessible auxiliary data bases 
should be present within the data stream to allow subsequent processing at an appropriate 
center. These data would be available in orbit-sequential data bases. 

Level IB The MODIS Panel reviewed existing and anticipated prxtice for prcKessing satellite data 

and recommends that satellite swaths be ordered by relation and then by time. This 
recommendation assumes large-volume, low cost storage media are available in the MODIS 
time frame, but the total data volume will still be large. Segmenting the data regionally will 
permit ready access to tha* portion of the overall data archive that can be logically grouped. 
This regional-area data segmentation can be a function for cither the central or regional 
center. Thus geographical blocks such as continents, ocean basins, polar regions, etc., 
should be established. Coastal zones represent an area where MODIS and HIRIS data can 
be used to address a range of problems over various space scales. Data should be flagged for 
presence of land or clouds. 

Level 2 Level 2 data are derived geophysical parameters in orbital serial format. Atmospheric 

corrections and derived-product algorithms are applied here. This level is not reversible to 
Level 0. A minimal set of derived properties is computed routinely. The number of these 
properties is expected to be on the order of 25. Some examples are: 

1. Terrestrial Leaf Area Index 

2. Ocean Chlorophyll Pigment 

3. lerrestrial Surface Temperature 

4. Sea Surface Temperature 

5. Aerosol Optical Depth (over oceans) 

Additional properties for which algorithms are expected to be well developed by launch: 

6. Chlorophyll Fluorescence 

7,S. Additional Terrestrial Vegetation Indices 
y. Bioluminescence 

10. Oceanic Cyanobacteria Index 

11. Terrestrial Aerosol 
12-15. Atmospheric Properties 

16. Oceanic Particulate Calcium Carbonate Concentration (Coccoliths) 

Items 17-25 are reserved for new properties of general and routine interest for which 
ilgorithms are not presently under development. 

Level 3 Level 3 data products are spatial and or temporal composites of Level 2 mapped to a fixed 

Earth grid. Sample time and space scales for compositing arc given in Table 20. Each Level 
3 grid point should contain the mean, number of pixels used to compute the mean, standard 
deviation, and skewness. Other Level 3 products may be requested and required to support 
AO investigators on the MODIS team. 



F«l 



permits the information needs to be met in a cost- 
effective manner as part of the Eos Data ; nd Infor- 
mation System. These MODIS requirements 
recognize: 

I . A need for some rapid data delivery, specif- 
ically in support of field programs and 
(Kcanic expeditions 



2. The expected demand for standardized 
products of vo-ne basic derived properties, 
for global-scale interdisciplinary studies in 
particular 

The MODIS Panel has therefore recommended 
that (near) real-time distribution of low-level data 



45 



z. 



® 



Si&t^i^^ 



••X"^ 



Property 

LAI 

Ocean Pigment 

Temperature Land 
Sea Temperature 
Ocean Aerosol 



Table 20. Compos iting Scales 
Spatial Bins 



0.5 km 
10 km 

0.5 km 

10 km 

100 km 



Temporal Bins 

week 
week, month, 
annual 
week 
week 
week 



be provided as well as routine production and ar- 
chivmg of standard products. The real-time distri- 
bution could be effected in seveial ways, including: 

1. Onboard processing to quick-look products 
with direct transmission 

2. Some rapid data center processing and trans- 
mission to users via communication satellites 



3. Distribution (possibly of selected channels) 
of Level lA data to networked processing 
centers for the purposes of large-scale re- 
gional studies, algorithm development where 
such development requires substantial vol- 
umes of data, and global studies requiring 
specialized processing not compatible with 
standardized, central service-produced 
products ^ 



__JaWe21^MODIS Data Requirements-Expecled 



Requests for Data 



Access to Level IB I . Algorithm developers 

2. Field experiments 
anywhere 

3. Demand for special 
Level 3 

4. Operational product 
improvement 

5. Reprocessing of 
Level 3 sets 

^ Regional distributed 
archives 

Level 2 cloud masks 

-every Level IB special request 
-cloud statistics 



5(»-75 

2-5, anytime 

H), up to 50 

10 

once every I to 
3 years 

2-3 



Level 3 (Na of requests depends on success of regional centers) 
-suridte temperature 
land 
ocean ^* 



-vegetation indices 
land 
ocean 

-aerosols 

-othcrs-undefmable. less than above 
potentially nearly equal 



100s 
250 



Comments 



all channels, regional 
lime series (L(HK) km) 
<I day, level 3, L(X)0 km 

highest resolution, 
random regions 

I month, all data, 
selected regions 

improved algorithms, 
data updates 

rapid access to all storage 
limited, up to 50 centers 

regional requests/yr 



(subset of chlorophyll 
pigment) 



46 



® 



•1*1 



M > 



terrestrial remote s^ensinSbvT?'"' ''P'*'''«y for 
provide an invaJuaWe dfta ir ^^"''^'"""'^"dwill 

studies. The justification/^ ' ^? " *'<*<^ ^^nge of 

to fully understand the scen/rH""y '" ^"' "^^d 
resolution data. The soectr^i '*''*'^ ''^ ""^^ '««'- 
spatial resolution pixels J nnl^'"^^''!!''' ^'""^ '«* 
tance from a diversity of surT"''^'' ^^ ""■* ^^^Aee- 
•ng of such comply l^e'afS- ^"' ""^erstand- 
spatial-spectral SoSeC slS'^s:"'''.'-'^'"^'' f'"'" 
agmg by MODIS and H rk ^ ,h '""' '•"■ 
data sets necessary to addrei T''^ P"'**"*-'^- the 
questions, to testour mJde,, of .""^^T'^^'^'fo" 
reflectance, and to calibm?e ?hn ? """P'*^" ''"^^^e 
more accurately than S '"^-resolution data 
juration of low splLl r. ? '^''^^^ '" ^ate. Cal- 
surface measuremen s has t"""1 "^'"-'^"tions by 
•ow accuracy of kfcat „„ fo?'sl" '''"P-r-d ^y .he 
multaneous high-resohi»iMn i ^'''^^ sampling. Si- 

•niportant intermedial '""ia,! 7" "" P^"^''^' -" 
twcen ground measuremTnu ""'["P^^^ral link be- 

data. The inter-dtrnaT v".ri.'ri '''' '"^-''""'"'•"n 

for a multilevel samnlino ^ l^ P'""de the basis 

of applications SlST'' '"^^ ^'^'-' ^«"^tv 
heirjg made available i'Vt"Sj:,!;:;'^' '"^"^-"^'-n 
level sampling by remote ^-n^u "'""■^'" ^ulti- 
vcKated for range S forest m^ *"' '""^ ^''" «d- 
.heen used in such apph SjoL "f^""*^"'- ^"d has 
•toring and crop foS ifnoM'' ""'" '^'^"'*' '"«"- 
«H.rdinating. ..btainin, "„^i ""''^'^'^- difficulties in 
aneous c<,yeragc fm' d.f."^ registering contcmpor- 
dered such multih^?.^ *""' P'^'^^'"'* have ren- 
MODIS will cX d t iXir'" P^"h'«-''-atic 

-er a swath I.5(K;S''w'^ '^'^.rH" o';;'"' P'''*^''' 
data oyer a 50 km swath wiih J^^ "^'^ acquires 
and. by ayeragin^. 5 'm p cK l*" ^"'^' "'" '^"d 
to attain adcuuafe radionf . *■"'' ''"'" <'" ''rdcr 

The qucK of s n "' '"-'"•^"^vity). 

intheccitTp 2:^'!';;'" '•■■ ""^-^^'-'d 
addressed. The f<5l< m „« ,"' "."'^ measurables ,o be 

**'ays in which the JwolnlTr '"'''''''•■ '"^J*'' 
complementary data '"''"^"'"ents could pnnide 



VII. MODIS/HIRIS SYNERGISM 



DYNAMIC PHENOMENA 

whi.rs;s^S::,;;pKr^^^^^ 



X&JrSS SJSr^ -" as insect 
events such as volcan sm and &"*:.f "'^ ^P«odic 
by MODIS. HIRIS can th.„^' *'" ^ detected 
the phenomena in detail HIP%'''«'''^ ^ ^»"dy 
i20°crosstrack to reach any .^S. """ ?" P^'^'^d 
a maximum of six days ^ ^ °" "'^ «'<*e in 

CONTEXT AND WXEL STRUCTURE 

H.R7s/,5j)SdrrrirrH"-^''""''--- 

each instrumem benefiTinTn 5 '"T'S^^" for 
vides data for HIRIS inve"il^ t^'' ^^^^ Pro- 
=-ene context. The efoTmIS/ "^^"8 'he 
rounding the HIRIS «rl„l^ "^^'^ variabi! ty sur- 
<tf l-rger'scale fS^^^^ «"d the effect 

HIR^afcquS-^S^^^^ 

dataSl'SfSre^l?^"^'' ^'-^ "'^^S 
veal the inhomogene ,. aid "" ""'«" *'" ''■ 
MODIS pixels. At^ongX" th ?''"'w^*'"''" 
aid m the study of coasfalfronL ^^^P^l^'lity will 
plumes and the studJ of mT '^"^ '° '*''" o^ river 
as Sargassum and kelp Wn "h '"^'^'-^Phytes such 
sampling for Moors'^ the hTrk H '''^''-^""'"tion 
dude programmable poLino ^IT '''""'d in- 
Permit selectable covS Ir'^'^^'^'y '^^' ^'H 
within the MODIS scene '''"'' '""tions 

^'S^atS'^IJ^^ension and 

SPATIAL EXTRAPOLATION 

Simultaneous MODIS and Hiriq ^ 
make possible surface-relkctant ^ ^ ^^^'^ *'" 
sion using HIRlS-general^Hfii "8"ature exten- 

•nembers (signatures)" o„e„era^ fh 'P^^''^' ^"^- 
signaturesofMODIS In Jh?.! ^^ mixed-pixel 
to extrapolate pmcesseS fnf "' ''^'" ^P««^^ 
from HIRIS <iat^u"Zr I^TtT'' .|!?^'°P-d 
instance, to cioh w m .^ r ^"''* ^"" ead, for 

hetter undeSlTP^fjf •«''«" indices and a 

composition detSa^Li or*'" °^ "^^^'^ ^r 
determination of Sidesi? v'''"P'" ^^^ 'he 
and sediment .ran^XtK.ssS.'''''^^" «^a''«" 

ATMOSPHERE 

L'nderstandin? of the um,, u 



47 



t 



® 



ssa.t^^ 



WBT-^ 



pptnui^H. . \ I ^titm'fr^^tt^mmfm/imf^mMKA J.^wprTPwiipni 



-vr 



proper utilization of data from either HIRIS or 
MODIS. Intercomparison of results from both in- 
struments, and from companion instruments such as 
LASA, will provide better atmospheric corrections. 



particularly over land-water interfaces. MODIS will 
provide information on cloud cover surrounding the 
areas imaged by HIRIS for analysis of adjacency 
effects. ^ 



48 







^v^n^mm^i^m^ 



TT 



^^^p^« 



APPENDIX A: ATMOSPHERIC CORRECTIONS OVER LAND 



tt? 



Satellite measurements of the characteristics of 
land surfaces depend significantly on the optical ef- 
fects of the atmosphere. This section discusses such 
effects for a cloudless atmosphere and methods for 
conecting for the effects in the spectral range below 
3 ^im. The essence of the atmospheric effects can 
be discussed with the aid of the following accurate 
expression for the radiance (L) of the Earth-atmos- 
phere system: 

L - L,, + Tr 

where L,, is the path radiance of the atmospheric 
column, T is the transmission of sunlight to the sur- 
face and then to a satellite, and r is the surface 
reflectance. All quantities are functions of wave- 
length, polar angles from the surface to both the 
sun and the satellite, location, and time. Since the 
radiance is nearly a linear function of the surface 
reflectance, if the latter is known for dark and bright 
surfaces, then the iwo atmospheric parameters L,, 
and T can be estimated from the satellite measure- 
ments of radiance. Although the method seems sim- 
ple, it is difficuh to apply because the surface re- 
flectance is not usually known with enough accuracy. 

The optical effects of the gaseous components 
of the atmosphere alone can be calculated accu- 
rately. The MODIS spectral bands will be chosen in 
the atmospheric windows, where molecular absorp- 
tion is weak. McClatchey et al. (1971) and Kneizys 
et ai (1983) give methods for calculating atmos- 
pheric transmission. Well developed radiative trans- 
fer models exist for calculating molecular and aer- 
osol scattering (Lenoble, 1977), Since aerosols are 
always present in the atmosphere, the molecular 
scattering should not be considered independent of 
light scattered by aerosols, when the aerosol optical 
density is large on either the path from the ground 
to the sun or to a satellite. 

The difficulties in making atmospheric correc- 
tions are caused by aerosols, since their optical 
properties are difficult to estimate during satellite 
observations: their properties are not known accu- 
rately and they are variable. The aerosol optical 
parameters are their optical thickness, single scat- 
tering albedo, and scattering phase matrix. The 
scattering phase matrix, which accounts for the po- 
larization properties of scattered light, is required 
instead of just the phase functions, if any of the 
following three conditions apply: 

1. The MODIS radiometer is sensitive to 
polarization 

2. The polarization of light reflected from 
plants is measured 



3. Accurate atmospheric corrections are cal- 
culated for atmospheres containing moder- 
ate amounts of haze 

Some idea of the accuracy required for the aer- 
osol optical parameters can be given for two atmos- 
pheric states and observations near the nadir direc- 
tion. Assume that the surface reflectance will be 
measured with an accuracy of O.Ol. A rather com- 
mon state is one where the aerosol optical thickness 
is 2, its albedo of single scattering is 0.96, and the 
surface reflectance is 0.1. The required accuracy of 
the optical thickness is 0.1, and an accurate value of 
the single-scattering albedo is unimportant. This 
implies that atmospheric corrections are not re- 
quired for near-nadir observations, if the aerosol 
optical thickness is less than 0.2 (Schowengerdt and 
Slater, 1979). To take another example, consider the 
problem of dense haze (an optical thickness of 0.6) 
that is common in such places as the eastern United 
States during the summer, or the Sahara region. The 
optical thickness is still an important parameter, but 
now the radiance is sensitive to the aerosol single- 
scattering albedo, which has to be specified with an 
accuracy of 0.02, when the surface is bright (r = 
0.4) (Fraser and Kaufman. 1985). The reflectance 
measured at a satellite, however, depends on both 
the optical thickness and the single-scattering al- 
bedo when the zenith angle at the ground of a ray 
from the ground to either a satellite or the sun is 
large. 

The aerosol optical properties are a function of 
wavelength, but the correlation of the same param- 
eter at two different wavelengths is generally good. 
The aerosol optical thickness can vary from hun- 
dredths to values large enough to obliterate surface 
features. Usually, the visible optical thickness range 
over land is 0.05 to 1.0. The aerosol single-scattering 
albedo ranges from 0.5 in some urban environments 
to 0.99 in rural environments (Shcttic and Fenn, 
1979). The scattering phase matrix depends on mo- 
lecular scattering and on aerosol size, composition, 
and shape. This matrix has large variations (Sekcra, 
1957). 

The small amount of experimental data indi- 
cates that the spatial gradients of aerosol parameters 
may be important when moderate to dense haze is 
present. The vertici>l profiles of the parameters are 
important for calculating the transfer of radiant en- 
ergy from outside to inside the instantaneous field- 
of-view (IFOV); but this adjacencv effect is signifi- 
cant for IFOVs smaller than thai of MODIS (Kauf- 
man. 1984). The vertical profiles become more im- 
portant with increasing amounts of haze and large 
polar angles from the point of observation to either 
the sun or satellite. The horizontal gradients of the 
optical parameters depend on the locations of 



49 



O 



^^^m 



Ff^^P^^PR-^^P^^^piPiiiW^WPPP 



U^ 



^m. 



WIWi' 



'W^ 



^^mi^mm 



^ 



sources such as cities, forest fires, agricultural burn- 
ings, and dust storms. The smallest significant scale 
seems to be about the depth of the mixed boundary 
layer (about 1 km) (Stull and Floranta, 1984). Since 
the aerosols are frequently hygroscopic, their opti- 
cal properties depend on the relative humidity, 
which changes diurnally. 

Two general approaches have been used or con- 
sidered for making atmospheric corrections to the 
radiances measured from satellites. One approach 
is essentially empirical, whereas the other involves 
^computations with radiative transfer models. In 
either case, analyses can be simplified by accounting 
for the variation in flux of solar radiant energy in- 
cident at the point of observation by ratioing the 
spectral radiance to the solar spectral irradiance. 
Then atmospheric effects can be reduced by ratioing 
the normalized radiances in the various bands. This 
method is most successful when the radiance is 
strongly dominated by light reflected from the 
ground. Another empirical method is to make a 
principal component analysis of the normalized ra- 
diances in the various bands (Lambeck, 1977). The 
components with information about the surface pa- 
rameters tend to be independent of atmospheric ef- 
fects. Such empirical methods, however, do not have 
general applications. 

The other approach utilizing radiative transfer 
models requires specification of the aerosol optical 
properties, which is difficult to do. The accuracy of 



the computed radiance is not restricted by the nu- 
merical methods, but by the agreement between the 
models of the atmosphere and its true state on the 
occasion of satellite observations. In addition to cal- 
culating the atmospheric transmission and path ra- 
diance, the modulation transfer function of the at- 
mosphere can be calculated, and applied to the 
Fourier transform of the measured radiances, in or- 
der to find the Fourier transform of the surface re- 
fllectance (Kaufman, 1984). Atmospheric correction 
procedures may complicate processing of satellite 
measurements so much that corrections will be 
made only on limited sets of measurements. 

When atmospheric corrections are computed 
with radiative transfer models, only climatological 
data on aerosol optical parameters will be available 
(Shettle and Fenn, 1979) unless special efforts are 
made to measure the optical parameters. The cli- 
matological data are always incomplete and also ex- 
tremely sparse for some regions of the world. More 
accurate data will have to be derived from measure- 
ments made by satellite radiometers themselves, 
such as MODIS and LASA. The cost and manpower 
required are too great for making auxiliary meas- 
urements from the ground or aircraft, except during 
special experiments. Methods are being developed 
to measure the optical thickness and scattering 
phase function, and albedo of single scattering from 
satellite data (Slater, 1980; Fraser etal. , 1984; Fraser 
and Kaufman, 1985). 



,^(l 



® 



miiv 



rmmfmmmm^mmfmmm 



W^ 



^mm 



APPENDIX B: ATMOSPHERIC CORRECTIONS OVER OCEANS 



The Coasta! Zone Color Scanner (CZCS), a 
precursor of MODIS, utilizes an algorithm that cor- 
rects for the atmosphere and determines chlorophyll 
concentrations in ocean waters with little or no sus- 
pended sediment. This algorithm will, with some 
improvements, be used by MODIS. The Nimbus-7/ 
CZCS is a scanning radiometer that views the ocean 
in six coregistered spectral bands, five in the visible 
and near infrared (443, 520, 550, 670, and 750 ijim), 
and the sixth, a thermal infrared band ( 10.5 to 12.5 
|xm). The sensor has an active scan of 78° centered 
on nadir, and a field-of-view of 0.0495°, which, from 
a nominal height of 955 km, produces a ground 
resolution of 825 m at nadir. The satellite is in a sun- 
synchronous orbit with ascending node near local 
noon. The sensor is equipped with provision for 
tilting the sci»n plane ±20° from nadir in 2° incre- 
ments along the satellite track, in order to minimize 
the influence of direct sunglint (the contribution to 
the sensor radiance from photons that were specu- 
larly reflected from the sea surface without interact- 
ing with the atmosphere). 

The CZCS provides estimates of the near-sur- 
face concentration of phytoplankton pigments (de- 
fined to be chlorophyll-a and its associated phaeo- 
pigments) by measuring the spectral radiance 
backscattered out of the ocean (Gordon and Moiel, 
1983). This radiance scattered out of the ocean ind 
reaching the top of the atmosphere comprises only 
a small portion of the total radiance measured at 
the sensor. In general the sensor radiance L^ at 
wavelength (X) can be decomposed into Li(X), the 
radiance due to photons that never penetrated the 
sea surface, and t(X)L^(\), the radiance due to pho- 
tons that were backscattered out of the water (the 
water-leaving radiance) and diffusely transmitted to 
the top of the atmosphere, i.e.: 

L,(X) = L,(X) + t(\)U(X). 

All of the information relating to the oceanic con- 
stituents, such as the chlorophyll concentration, is 
contained in L^(X), which is usually an order of 
magnitude smaller than L,(X). 

Schemes for extracting L^(X) from L,(X) arc re- 
ferred to as "atmospheric correction" algorithms. 
To facilitate the discussion of these, it is often more 
convenient to work with reflectance, rather than ra- 
diance. We define the reflectance according to: 

p = ttL/F„cosB^„ 

where L is the radiance in the given viewing direc- 
tion, F.. is the extraterrestrial solar irradiance, and 
0., is the solar zenith angle. With this normalization 
for L, p determined at the top of the atmosphere 



would be the albedo of the ocean-atmosphere sys- 
tem if L were independent of the viewing angle. 
(Note that this is just a normalization of the radiance 
to the extraterrestrial solar irradiance, and has no 
other significance.) 

The CZCS signal is eight-bit digitized aboard 
the spacecraft. The reflectance corresponding to 
one digital count for the least-sensitive gain (1) and 
the most sensitive gain setting (4) is given in Table 
B.l. (The near infrared band at 750 nm has Landsat 
MSS sensitivity and is not used in oceanic studies 
except as a land/cloud discriminator.) In practice the 
gain is set based on the mean value of the solar 
zenith angle for a given set of scenes, i.e., for small 
zenith angles Gain 1 is used, and for very large 
angles Gain 4 is used. Thus, the red band (670 nm) 
saturates at an ocean-atmosphere reflectance about 
0.06, i.e., 6 percent. 

The correction algorithm as it is presently being 
used is most easily understood by considering only 
single scattering (Gordon et ai, 1983). In this ap- 
proximation, ignonng direct sunglint and assuming 
that the sea surface is flat, the reflectance measured 
by the sensor p,(X) can be divided into its compo- 
nents: p,(X) the contribution arising from Rayleigh 
scattering; p,,(X) the contribution arising from aer- 
osol scattering; and ,(X) p^(X) the contribution from 
the water-leaving radiance diffusely transmitted to 
the top of the atmosphere; i.e.: 



p,(X) = p,(X) + p,(X) + t(X)pJX). 



(I) 



Tvpical values of p^, p,, and p^ arc given in Table 
B.2. 

The values of p, and p, correspond to points near 
the center of the scan (there is considerable limb 
brightening pj, and the p^, values correspond to an 
aerosol optical thickness at 670 nm of about 0.1 (p, 
varies considerably with the specific properties of 
the aerosol, e.g., for a given optical thickness it can 
vary by more than a factor of two depending on the 
aerosol model used in the computations). The val- 
ues of p^ are given for both high and low pigment 

Table B.l. Reflectance for One CZCS 
Digital Count 



X 

(nm) 



Gain 1 



Gain 4 



443 


().(HK)75 


().(K)()36 


520 


().()(HI53 


().(MK)25 


520 


().(KK)42 


().(HX)2() 


670 


().(KK)24 


0.(MK)ll 



51 



® 



T^ 



Table B.2. Typical Values of p,/ p./ 
andp« 



X 

(nm) 


Pr 


P. 


P. 




C = 0.03 


CSM 


443 


0.10 


0.015 


0.035 


0.0008 


520 


0.05 


0.013 


0.008 


0.010 


550 


0.04 


0.012 


0.005 


0.015 


670 


0.02 


0.010 


0.0001 


0.002 



C is the chlorophyll concentration in mg/m\ 

concentrations (this is the significant range of vari- 
ation); however, there are water types (called Morel 
Case 2 waters) for the most part in coastal areas for 
which p, ^ 0.05 to 0.07 and rather featureless for 
the spectral range 450 <X< 600 nm, and === 0.01 
near 670 nm. In most situations of mterest, the pig- 
ment concentration is determined from the ratio of 
reflectances: p,(440)/p,(550) at low chlorophyll con- 
centrations; and pJ520)/p,(550) for high pigment 
concentrations. There is no generally accepted 
method of extracting the pigment concentration 
from CZCS-measured p,(X) for the Morel Case 2 
waters mentioned above. Thus, the data in Table 
B.2 suggest that we need to extract p, from p. to 
within about 0.0001 in order to obtain a useful es- 
timate of p^, under most conditions for two of the 
three bands 443 nm, 520 nm, and 550 nm. 



p, and Pa in Equation 1 are given by 



^ a).(x)T,(x)T(x)p,(e,e,,x) 
P^ 4cos e cos e,„ 



(2) 



where 

p.(B.H.,.X) = {P,(«.A) + 1p(B) + p(HJ1 
xP.(«^.X)}. 

and cose± = ±cose.,cos H + sin 6.. sin e cos (d)- 
^J, 9,, and (t>,, arc, respectively, the solar zenith and 
azimuth angles, and d) are the zenith and azimuth 
angles of a vector from the point on the sea surface 
under examination (pixel) to the sensor. p(0) is the 
Fresnel reflectance of the interface for an incident 
anftle , Px(0^X) is the scattering phase function of 
component x (x = r or a) at X. a>,(X) the single- 
scattering albedo of x (w, = 1 ), and t,(X) the optical 
thickness of x. T(X) is the two-way transmittance 
through the ozone layer, i.e., 

T - cxp|-T,»,(l/cosB + 1/coseJl, 



where t^^ is the ozone optical thickness. The term 
involving 6- in Equation 2 provides the contribution 
owing to photons that are backscattered from the 
atmosphere without interacting with the sea surface. 
The term involving 0+ account for those photons 
that are s attered in the atmosphere toward the sea 
surface ( Ky radiance) and then specularly reflected 
from the surface into the field-of-view of the sensor 
(p(0) term), as well as photons that are first specu- 
larly reflected from the sea surface and then scat- 
tered by the atmosphere into the field-of-view of the 
sensor (p(ej term). If the assumption of a flat sur- 
face is relaxed, these terms involving p become in- 
tegrals over solid angle of essentially the product of 
the reflectance, the phase function, and the surface- 
slope probability density function. 

t(X) is the diffuse transmittance of the atmos- 
phere between the sea surface and the sensor. It is 
given by: 

t(X) = exp[-(V2 + Tj/cos0]t,(X), (3) 

where 

ta(^) = exp [-(1 - a),(X)F(X)) T,(X) /cos 0], 

and F is the probability that a photon scattered by 
the aerosol will be scattered through an angle less 
than 90°. The upper limit to the factor (1 - 
o)3(X)F(X)) is about 1/6, so t, depends only weakly 
on the aerosol optical thickness. The rationale for 
using the diffuse transmittance rather than the di- 
rect transmittance is to account for the fact that 
when the sensor is viewing a given pixel, some of 
the radiance it receives originates from neighboring 
pixels. The only unknowns in these equations (other 
than p^) are w,, t^, and the aerosol scattering phase 
function. 

Examinations of CZCS imagery in the red band 
over low chlorophyll waters (pJ670) ^ 0) shows that 
the aerosol reflectance is dependent on position. 
This means that knowing the aerosol reflectance (or 
even the aerosol phase function and optical thick- 
ness) at one point in an image does not provide 
sufficient information to compute the aerosol reflec- 
tance everywhere in the image. This dependence of 
p,(X) on position is believed to be due to variations 
in the aerosol optical thickness, implying that such 
variations must, at least implicitly, be taken into 
account in any atmospheric correction scheme. 

From Equation 2 it is seen that: 

P..(X:)p,(X,) = e(X,,X,)lT(\:)/T(X,)l, (4) 



where 



A 



e(X:.X,) 



w.,(x,)T.(x,)p.,(e.e.,.x,) 



(5) 



52 



® 



M > 



i^ . • 



S .f 'l^ <>f «(^>A.) is of central importance in 
the atmosphenc correction procedure. E2 for 
a given aerosol type, defined here to t? a^ven 

dent of position within an image '"oepen 

Equations 1 and 2 are rigorously correct in th*. 
hmit that the slam paths VcosO. VcoSe x W 
and Vcose.. all approach zero. AltLgh EquS 

T,(A) where multiple scattering becomes imoortant 

multiple scattering computations show tharSa: 

S h ?'" ^PP^'^^i'^^'ely valid even for largeT 

ica thicknesses as long as the radiometer i?vieT 

r.r'fecTS^ 'I ''- ^HntavSeeVSelr 
on the CZCS. Deschamps el al. ( 1983) have derived 

creases tE '' ^^"f ^" ' "'''^'' ^'gnifictntly Tn 
creases the accuracy: however, the simpler Equation 
1 IS sufficient for the CZCS AUn in th:^ ^ ^ 
cCk \ \ .,„i • ""',^^'->>- /Mso. in this case the 
€(V>,X,) value given by Equation 5 is only an ao 
proximation and multiple scattering introducefa 

In what follows, it is assumed that c(X, X \ i. 
constant* even in the presence of a horizontallv in 
homogeneous aerosol. This is equivalem to a sum 



t(X,)pJ\.) = p,(x.) - p^(x,) 

- S(X..XJ [p,(\J - p^(x^> 

- UX4) pjXJJ; 



(6) 



!nH'4~ *• ^" ''u'^ ^- *''*'^*^ 'he indices i = 1 7 1 
and 4 refer to the four visible CZCS bands inortcr 
of increasing wavelength, and 

S(X„X,) = e(X,XJ li^ 
T (X,) 

Equations 6 are 3 in number but there appear to be 
II unknowns, t(X.) for i = I ,0 4. S(X ,xTfor i - 1 
to 3. and pJX.) for i = I to 4. HowU for a Uen 
aerosol type the three S(X.,XJs can be determ^neS 
everywhere once ;hey arc determined at oncZsi 

reigCX^rrf^'^'''--^""'"""^^^^^^ 
lu tigm. Also. t(\.) IS unknown only because the 



bunds available to en. brl, ■•''"'■' "''^ '"" f*"'* "'P'-'^'"" 

and MODIS. additional snectr, h.„ . ^ *" ^""^" <"^''> 

facilitate this detelmll^Lrn^nd 'hc":^^",^ T '"'""' *'" 
not be needed t<>nst.int « ;issumption will 



pu a ion of t,(X,). in most cases of practical interest 

4m will tet ? """/ because the entire algo: 
rithiti wii break down for other reasons before t 

rtiTth ''^'h'"°"«.!? '° •"«"^"'^^ signmcanS; the 
£ ?aken o"S T '^^'" ^" '"^^■)- '^f'"^- t(^*) can 

use an empirical equation of the form ^ 

f(Pw(X,), p,(x,), pJXJ) = 0, (7a) 



while Gordon et al. (1983) used 

t(^4) P.(X4) = 



(7b) 



01 €(X,.\4). which provides S(X .\J 
P7 '" ^^^ 'nitial application of this algorithm to 
CZCS imagery (Gordon et al.. 1980) the Srxx ^ 
were determined from ship measuremems of p^ tx 
at a single location in the image. This reliance on 
surface measurements was. hotever umatis?vin. 

'tlVSnSfiri 17''' '''' woulS"enabt df: 
ilom" Th °^^^^-^^) f""""! satellite measuremems 
/V-Td J''^'^''""Pt °f clear water reflectance 
(CWR) provides the basis for such a determlSlZ 

rca l^cl:^ bands can be written: 
PJX) = (p„(X)J.,.cose„ 

X exp[-(V2 + T,J/cose,.|; 

^l^'ctS'^hu'-'T' '''''''• -^"»- and 6^ nm'^ : 
spective y. Thus, if a region of image for which C < 
0.25 mg/m' can be located, equations 6^nd 8 r.n 
be used to determine €(520 670) J?W) l^m a 
.(670.670). .(443.670) cL then be e t maS ly'ex 

hr r^rxTr ''^ "''''''' calculation 'be'l^w 
. itwt €(X,.XJ IS a smooth function of X Tn 

iWtha,:'"^ procedure, it is assumed foj a'nven! 



(9) 



e(^,.X4) = (X,/X4)n (X,) 
and then €(X,.X4) is determined from: 

n(X,) = |n(X.,) + n(x3))/2. (lo) 

An important aspect of this algorithm is that no 

ti ":,f";rr "'■T^ "^ ''''" '^^'^^ - - p «";: 

trtits of the aerosol arc required to affect the -,t 
mosphenccorcction with this scheme. 



0^ 

\ 



53 



®; 



^..■^■- J*»- 



\ 



APPENDIX C: MODIS INSTRUMENT PANEL 
STATEMENT OF WORK 



1. Clarify and refine the science and measure- 
ment objectives outlined in the Science and 
Mission Requirements Working Group Re- 
port for the Moderate-Resolution Imaging 
Spectrometer (MODIS). 

2. Specify detailed observational requirements. 

3. Define characteristics of a candidate instru- 
ment and alternative approaches including: 

(a) typical observing scenarios 

(b) operating characteristics and require- 
ments 

(c) data acquisition, processing, and inter- 
pretation strategies 

(d) refined definition of spectral bands, res- 
olutions, and sensitivities 

( e ) requirements for correlative data for im- 
age correction, calibration, and 
interpretation 

(f) appropriate use of array detector tech- 
nology and selectable spectral bands 



(g) onboard processing opportunities and 

requirements 
(h) strategy for on-orbit servicing 
(i) determination of which instrument 
functions can be integrated into a com- 
mon optical train versus which require 
separate hardware implementation 

4. Coordinate with ongoing studies and devel- 
opment of related instruments. 

5. Make recommendations to the Earth Sci- 
ence and Applications Division/NASA 
Headquarters and to the Eos Project on the 
feasibility, development timing, limiting 
technologies, and possible follow-on defini- 
tion and development activities of this 
instrument. 

6. Produce an interim oral report in October 
1984 and a written study report by March 



.S4 




IIIPMHIilMipPHmMPH 



REFERENCES 



Abbott M.R., and P.M. Zion, Satellite observa 
tions of phytoplankton variabili,; durinfan uS-' 
ing event, Cont. Shelf. Res.. 4, ^1, 1985. ^ 

Th^ll^^u '^f '*^*^"' «"d PJ. Richardson, 
The relationship of environmental variability to the 
spat^l patterns of phytoplankton biomass iVuJe 
lahoe, J. Plankton Res., 4, 927. 1982. 

MAcT^ ^^' ' <^*^"«'- ^^.d H. Lang, The joint 
Special Publication, in press, 1985. 

Barker J.L.. (Ed.), LANDSAT-4 science charac 

Sness^inVerf"' ^^P''^"^-^' Phytoplankton 
Bernstein. R., Report on the Toga Workshop on sea 

blooming off the U.S. east coast: A aSe descrrp 
tion. Science. 2?.9. 163, 1985. aescnp- 

fncl'anHM-- '' '''u ^""'^ ^^^'^^^*"8 System: Sci- 
ence and Mission Requirements Working Group 
Report. NASA TM 86129, 1984. '^ 

Campbell. J.W, and WE. Esaias. Spatial patterns 
s"h *r''f"''"'l'"^ chlorophyll on Nantucket 
9 m{ rJS "''J«'""'^/,'='"««e ''^^nsing data. Mav 7- 
"^^ iy«l, y. A/£?a: Res., 43. 139, 1985. 

Chahine. M.T. N.I.. Evans. V Gilbert, and R D 
Haskins, Requirements for a passive IR advanced 
979 XV '*''"f^'^«'"^«^ ''«"nder. Appl. Opt.. 23. 

Shnf \" • "'"'L^ ^""'"=^- Confirmation of the 
airborne biogeophysical mineral exploration tcch- 
723"'i983"^ '«t«'ratory methods. Earn. GeoL. 78. 

Kornfield A comparative study of albedo change 
1366 1977 '" '"''""""'^ '■'»'""''• ^ '^"""^ '^" • S 

Colwell R.N.. (Ed.). Spectral transmissivity of 
clouds of various thickness, in Manual of Remote 



Sensing, 2nd ed.. Figure 5-30, American Society of 
Photogrammetry, Falls Church, VA, 1983. 

Courel M.. Etude de revolution recente des mil- 
ieux sahehens a- paitio des mesures fouries pouTs 
PartiSs'"""''' ''•^- "'''"'^" l^nivLTy of 

Cox S.C, (Ed.), The multispectral imagine science 
working group: Final Report, NASA LKS 
Publication 2260, 3.20, 1983. 'J^^rence 

I;HiItLV-; '''"P"'^' ^"' '^ '^'°"'^ climatology and 
radiation budget experiment for Space Lab-2. 53 
pp., Submitted to the Office of Planning and Pro- 
gram Integration, December 1976. 

di'mate ^H^i Jh'''' ''°"'^ '■''''^''°" <^''Pe"ment as the 

t:rrni^^8?"-°^^p''^^^--'^^^^^^^^^^ 

penman, K.L.. Predictability of the marine plank- 
omc ecosystem, in The Predictability of FuIdMo- 
fons, edited by G. Hollowav and B.J. VVfest p 60^ 
American Institute of Physics. NY. 1983. " 

penman. K.L.. Covariability of chlorophvll and 
temperature .n the sea. Deep .Sea Re.s.!^23 m 

penman. K.L. andTM. Powell. Effects of physical 
pmcesses on planktonic ecosystems in the^coa a 
ocean. Oceanogr. Mar. Biol. Ann. Rev.. 22. 125. 

t ling of the atmospheric etfccts and its application 

S. .575K mT'"""' "' "'''" ^■'""^- -^'/'- ''P'- 

DeVlK)ys. C.G.N.. Primary production in aquatic 
environments, m Scope 13. The Global cTrbol 
Cycle, edited by B. Bolin. E.T Degans S S 
Kempe. and P Ketner. p. 259. J. Wiley. 1979." 

Hcyman Estimating ocean primary production 

SiZen- "' f "^"P^vil: Int/oduction'to legS 

h It / n/",'^ ''"'"^"^'^ ^"'- 'he Southern California 
bight, J. Plankton Res.. 7. 57. 1985. 

Epplcy. R.W Estimating phytoplankton growth 
rutes in the Centr.l Oligotrophic Oceans in Prl 
mary Productivity i„ the .SV«.' edited by PG S- 
kowski. p. 213. Plenum Press. NY. 1980: 



W\i 



55 



® 



Eppley, R.W, and B.J. Pfeterson, Particulate or- 
ganic matter flux and planktonic new product on?n 
the deep ocean. Nature, 282, 677, 1979. 

Esaias, WE., Remote sensing of oceanic phvto- 
plankton.-^Present capabilities and future goak in 
Primary ProducUvity in the Sea, edited by PJ F^ 
kowski, p. 321, Plenum Press, NY, 1980. 

Fraser, R.S., and Y.J. Kaufman, The relative im- 
portance of aerosol scattering and absoqjtion in r" 

222^^9^^" ^''"'- ^^''^"- ^-- ^- . 

^at^Ihte measurements of aerosol mass and trans- 
port. Atmos. Environ., 18. 2577 1984. 

Fung. I.Y. C.J. Tucker, and K.C. Prentice AddH 
catjons of AVHRR vegetation index "osmdyl 
mosphere-troposphere exchange and CO,, /. Gel 
phys. Res. . m press. 1986. 

Gathn. J.A.. R.j. Sullivan, and C.J. Tucker Con- 
s.derat.ons of and improvements to large-scak veg- 
e at.on momtonng. /£££ 7>am. GeLi. ReZl 
Jem.. Ut-22. 496. November 1984. 

Goetz. A.F.H.. D.N. Rock, and L.C. Rowan Re- 
^.:T^i^:'l^-^-- ^" overview^^eo: 

Sovfs°"ph",'^ .• ^.^ ^^''^- ^ L Mueller, and WA. 
coTs ll ;^>'"P'^"''»«" P'g"-.ents from the Nimbus-7 
coastd! zone color scanner: Comparisons with sur- 
face measurements. Science, 210. 63. 1980. 

Gordon. H.R.. and D.K. Clark. Clear water radi- 
ances for atmospheric correction of coastal zone 
color scanner imagery. Appl. Opt.. 20. 299. 1981 

Brown"p"H^- ^^ ^^''^- J ^ B^o^"-- O.B. 
nl n^.?- '"'""'• ""'^ W^- Broenkow, Phyto- 

piankton pigment concentrations in the Middle At- 
lantic b,ght: Comparison of ship determinations and 
CZCS estimates. Appl. Opi.. 22. 20. 1983. 

Gordon H.R.. and AY Morel. Remote Asse,, 
'-^n^of Ocean Color for Interpre.atSof tlZ 

Goward. .S.N.. D.G. Dye. and C.J. Tucker North 

7 AVHPP ";?''''"'"" f''"*^'"'* "*"*'-''"^<^d by NOAA- 
/ AVMKK. Vci'eiaiion. in press. 1985. 

Sofon,.- .• """^ ^^ ^'"'''- Measurements of aer- 
osol opt cal parameters from satellites, in Proceed- 
ings Of, In- International Radiation Symposium^. 
rugia. Italy. August 21-28 1984 P"'"m,t^ 



Gurney, R.J., J.R Ormsby, and D.K. Hall A com- 
panson of remotely-sensed surface temperature and 
biomass estimates for aiding evapotranspirat"on de 
termmation m Central Alaska, irProce!ding^of7he 
Permafrost: Fourth International Conference, 1983 

NoIa/N^-^a"'' J- S"«kind, Findings of a joint 
H?^/^ of K'°""f^' comparison: (1) AMTS vs. 
HiKb-2 (2) physical vs. statistical retrievals, in Ad- 
vances m Remote Sensing Methods, A. Deepak 
publisher, in press, 1984. ^^c^cik, 

Hansen, J., D. Johnson, and A. Lads, Climate im- 

Harries, J.E., D.T. Llwellyn- Jones, P.J. Minnett 

s^a^urS f '^' '"^ ^-^s ^^"•'^y' Observations" 
^a surface temper.iture for climate research, Phil. 
Royal Soc. t^ondon, A309, 381, 1983. 

Harris, G.R, Temporal and spatial scales in ohvto- 
p^nkton ecology: MechanisrJs, methods modds 
and^management, Can. J. Fish. Aquat. Sci., SyTl 

Hayes. L., The current use of TIROS-N series of 
meteorological satellites for land cover suidfes /m 
J. Remote Sens., 6, 35, 1985. 

Hayward. TL., and E.L. Venrick, Relationship be- 

nd?ntl";^f r*"°^°P''y"- '"'^8^^»^'' 'Chlorophyll 
247 1982 ^"'"^"'^ production. Mar. Biol , 69, 

Heimburg. K.F, L.H. Allen, and WC. Huber. Eva- 
potranspiration estimates based on surface temper- 
ature and net radiation: Development of remote 
sensing methods Florida Water Resour. Res Or 
66. University of FL. Gainesville. FL. 1982. ' 

Holligan. RM., M. Viollier. D.S. Harbour R Ca 
mus. and M. Champagne-Phillipe. Satellite and ship 
studies of coccolithophore production along a coi 
tinental shelf edge. Nature. 304. 339. 1983 

Hovis W A The Nimbus-7 coastal zone color scan- 
ner (CZeS) program, in Oceanography From 

r^or'^R F^- -f ^ ^Tn ^^ ^^'^^- « M. Aher- 
ron R.F Ardv.m. and R.W Samms. Study of re- 
mote sensor spectra! responses and data processing 
m"\9M "" *" cL-ssification. Opt. Eng.. 23. 

r^Esfem^?'''" "P;r'"'"« *" '^' C^''^«^"i« cur- 
rent system. Prog, m Oceanogr. 12. 259. 1983. 






56 



® 



iCSU/WMO/lOC, Large-scale experiments in the 
WCRP, Volume I, 121 pp., WCRP Publ Series No, 
/, WMO. Geneva, Switzerland, 1983. 

Justice, CO., J.R Malingreau, and J.U. Hielkema, 
The application of satellite remote sensing tech- 
niques for monitoring natural vegetation, crops and 
rainfall, in Proceedings of the Workshop on Early 
Warning Systems in Asia and the Pacific Region. UN/ 
FAO, Rome, 1984. 

Justice, CO., J.R.G. Townshend, B.N. Holben, 
and CJ. Tucker, Analysis of the phenology of global 
vegetation using meteorological satellite data. Int. 
J. Remote Sens., in press, 1985. 

Kaufman. Y.J., Atmospheric effects on remote sen- 
sing of surface reflectance, Soc. Phot. -Opt. Instru. 
Eng., 475.20. 1984. 

KidwcIL K.B., NOAA polar orbital data users 
guide (TIROS-N, NOAA-6, 1 8), NOAA National 
Climate Center, 2-5, Washington, DC, 1984. 

Kiefer, D.A., Fluorescence properties of natural 
phvtoplankton populations. Mar, Biol.. 22. 263, 
1973. 

Kncizvs, FX., E.P. Shettle, WO. Gallery. JH. 
Chetwynd, Jr., L.W Abreu, J.E.A. Selby, S.A. 
Clough, and R.W. Fenn, Atmospheric transmit- 
tance/radiancc: Computer code LOWTRAN 6, 
AFGL-TRH'3-0187. 200 pp., AFGL, Hanscom 
AFB, MA. 1983. 

Ki>blentz-Mishke, O.J., VV Volkovinsky, and J.G. 
Kabanova, Plankton primary production of the 
world iKcan, in Scientific Exploration of the South- 
ern Pacific, edited by W.S. Wcwster, p. 183, National 
Academy of Sciences. 1970. 

Labovitz, M.L., E.J. Masuoka, R. Bell, A.W. 
Siegrist, and R.F Nelson, The application of remote 
sensing to geobotanical exploration for metal sul- 
fides. Econ. (ieoi, 78. 750, 1983. 

Lambeck. R.F. Signature extension preprocessing 
for Landsal MSS data. 74 pp.. Environmental Re- 
search Institute of Michigan, Ann Arbor. MI, 1977 

I.aurs. R.M.. PC Fiedler, and DR. Montgomery, 
Albacore tuna catch distributions relative to cnvi- 
ronmcnial teatures observed ^rom satellites. Deep 
Sea Res... ^L 1085. 1984. 

Lem>blc, J., Standard priKcdnres to compete at- 
mospheric radiative transfer m scattering atmos- 
phere, 125 pp . NCAR, Boulder, CO. 1977 



Mackas, D.L., K.L. Denman, and M.R. Abbott. 
Plankton patchiness: Biology in the physical vernac 
ular. Bull, Mar Sci.. 37. 652, 1985. 

Malingreau, J.P, G. Stevens, and L. Fellows, Re- 
mote sensing of forest fires: Kalimantan and North 
Borneo in 1982-83, Ambio. 14. 314, 1985. 

Matson, M., S.R. Schneider, B. Aldridge, and B. 
Satchwell, Fire detection using the NOAA series of 
satellites, NOAAINESDIS TR 7, Washington, DC, 

1984. 

McCarthy, J -J., Measuring oceanic primary produc- 
tivity, in Proceedings of a Global Ocean Flux Study. 
p. 151, National Academic Press, Washington, DC 

1984. 

McClatchey, R.A., R.W Fenn, J.E.A. Selby, FE. 
Volz, and J .S. Garing, Optical properties of the at- 
mosphere, 85 pp., AFCRL-7I02'79. AFCRL, Han- 
scom AFB, MA, 1971. 

McElroy, J.H., and S.R. Schneider, Utilization of 
NASA's space statii^r. program for operational 
Ear^h observations, NOAA/NESDIS TR 12. Wash- 
ington, DC, 1984. 

Mintz, Y., J. Susskind, J. Dorman, and G. Walker, 
Global fields of transpiration and soil moisture as 
derived from TIROS-N satellite measurements, in 

preparation, 1985. 

Mitchell, G., and D.A. Kiefer, Determination of 
absorption and fluorescence excitation spectra for 
phytoplankton, in Marine Phytoplankton and Pro- 
ductivitw edited by O. Holm-Hansen, L. Bolis, and 
R. Gilles, p. 157, Springer-Verlag, 1984. 

Morel, A., and L. Prieur, Analysis of variations in 

ocean color, Limnoi and Oceangr. 22. 709, 1977 

National Academy of Sciences (NAS), Global tro- 
pospheric chemistry: A plan for action, p. 3-49, Na- 
tional Academic Press. 1984. 

National Academy of Sciences (NAS), Global ocean 
flux study: m Proceedings of a Workshop. 360 pp.. 
National Academic Press, 1984. 

National CXeanic and Atmospheric Administration 
(NOAA), Global Vegetation Index Users Guide. 
SDSD/NESDIS, 27 pp.. National Climate Data 
Center. Washington, DC 1986. 

Norwine, J., and D.H. Groegor, Vegetation classi- 
..ation based on AVHRR satellite imagery. Remote 
Sens. Environ.. /.?. 69. 1983. 



57 



O'Reilly, J.E., and D.A. Busch, Phytoplankton pri- 
mary production on the northwestern shelf, Rapp, 
FV. Reun. Cons. Perm. Int. Explor Mer, 183, 255, 
1984. 

Otterman, J., Satellite and field studies of man's 
impact on the surface in arid regions, Tellus, 33, 68, 
1981. 

Peterson, D.L., J.G. Lawless, P. Matson, J.D. 
Aber P Vitousek, and S. Running, Biogeochemical 
cycling in terrestrial ecosystems, in Proceedings of 
(he 36th Congress of the lAF, Stockholm, Sweden, 

1985. 

Piatt, T, and A.W Herman, Remote sensing of 
phytoplankton in the sea: Surface-layer chlorophyll 
as an estimate of water-column chlorophyll and pri- 
mary production, Int, J, Remote Sens., 4, 343, 1983. 

Piatt, T, and D.V Subba Rao, Primary production 
of marine microphytes, in Photosynthesis and Pro- 
diutivity in Different Environments, International 
Biological Programme 3, edited by J. P. Cooper, p. 
24^^, Cambridge University Press, 1975. 

Richards, F. and P. Arkin, On the relationship be- 
tween satellite derived cloud cover and precipita- 
tion, \fon. Mea. Rew. 109, 1(»81, 1981. 

Richardson. A. J.. J.H. Everitt, and H.W Gauss- 
man, Radiometric estimation of biomass and nitro- 
gen content of alicia grass. Remote Sens, of Envi- 
ron.. 13. 179. 1983. 

Rock, B.N.. Mapping of deciduous forest cover us- 
ing simulated Landsat DTM data. International 
Geosdence Remote Sensing in Proceedings of the 
Symposium, 5, 3.1, 1982. 

Rock. B.N,, and J.E, Vogclnann, Stress assess- 
ment and spectral characterization of suspected acid 
deposition damage in red spruce (Picea rubens) 
from Vermont. Amer Soc. Photogrammetrx Tech. 
Prog.. 19S5a. 

Rock. B.N.. Remote detection of geobotanical an- 
omalies associated with hydrocarbon microseepage 
using Thematic Mapper Simulator (TMS) and Air- 
borne Imaging Spectrometer ( AIS) data, in Remote 
.Sensing for (ieoiogic Mapping. IS. edited by Teleki 
and Weber, p. 299, International Union of Geolog- 
ical Sciences, 19K5b. 

Running, S.W. Microclimate control of forest pro- 
ductivity: Analysis by computer simulation of tran- 
spiration photosynthesis balance in different envi- 
ronments. Agric. and Forest Meteor, 32. 267. 1984. 



Running, S.W, D.L. Peterson, M.A. Spanner, and 
K.B. Teuber, Remote sensing of coniferous forest 
leaf area. Ecology , in press, 1985. 

Ryther, J.H., Photosynthesis and fish production in 
the sea, Science, 166, 72, 1%9. 

Schneider, S.R., S.R. McGinnis, Jr., and J. A. Gat- 
lin. Use of NOAA/AVHRR visible and near-in- 
frared data for land remote sensing, 48 pp., NOAA 
TR NESS 84, Washington, DC, 1981. 

Schowengerdt, R.A., and PN. Slater, MRS litera- 
ture survey of atmospheric corrections, 52 pp., 
NAS5-25606, NASA/GSFC, Greenbelt, MD, 1979. 

Schwalb, A., The TIROS-N/NOAA-A to G satel- 
lite series, 73 pp., NOAA TM NESS 95, Washing- 
ton, DC, 1982. 

Sekera, Z., Polarization of Skylight, in Handbuch 
der Physik, 48, p. 288, 1957 

Shafer, O., Simple infrared telescope with stray- 
light rejection, SPIE, 304, 171, 1981. 

Shettle, E.P, and R.W Fenn, Models for the aero- 
sols of the lower atmosphere and the effects of hu- 
midity variations on their optical properties, 94 pp., 
AFGL-TR79-02I4, AFGL, Hanscom AFB, MA, 
1979. 

Slater, PN., MRS "Proof-of-Concept" on atmos- 
pheric corrections using an orbitable, pointable im- 
aging system, 80 pp. , NAS5-25606-TR-1653, NASA/ 
GSFC, Greenbelt, MD, 1980. 

Smith, R.C., and W.H. Wilson, Bio-optical re- 
search in the southern California bight, in Ocean- 
ographxfrom Space, edited bv J.F.R. Gower, p, 281, 
Plenum Press, NY. 1981. 

^mith, R.C, R.W Eppley, and K.S. Baker, Corre- 
lation of primary production as measured aboard 
ship in southern California coastal waters and as 
estimated from satellite chlorophyll images. Mar 
Bioi, 66,281, 1982. 

Soer, G.J.R., Estimation of regional evapotranspir- 
atioii and soil moisture conditions using remotely- 
sensed crop surface temperatures. Remote Sens. 
Environ., 9,21, 1980. 

Spanner, M.A., K.W. Tuber, W. Ecevedo, S.W 
Running, D.L. Peterson, D.H. Card, and DA. 
Mouat. Remote sensing of the leaf area index of 
temperate coniferous forests, in Proceeditigs of the 
lOth International Symposium on Machine Process- 
ing of Remotelx-Sensed Data, p. 3(»2, Purdue Uni- 
versity, Lafayette, IN. 1984. 



58 



® 



Spencer, R,W, D.W Martin, B.B. Hinton, and 
J. A. Weinman, Satellite microwave radiances cor- 
related with radar rain rates over land. Nature. 4 
141, 1983. 

Steele, J.H.. (Ed.), Spatial Pattern in Plankton 
Communities. 470 pp.. Plenum Press, NY, 1978. 

Stceman-Nielsen, E. , The use of radioactive carbon 
(C^) for measuring organic production in the sea, 
y. duCons.. IS. 117, 1952. 

Stull, R.B., and E.W Eloranta, Boundary layer ex- 
periment- 1983, Bull. Amen Meteor, Soc, 65, 450, 
1984. 

Susskind, J.J., J. Rosenficid D. Reuter, and M.T 
Chahine, Remote sensing of weather and climate 
parameters from HIRS-2/MSU on TIROS-N, J 
Geophys. Res., H9D. 4677, 1984. 

Tarpley, J.D , S.R. Schneider, and R.L. Monev, 
Global vegetation indices from NOAA-7 meteo»^o- 
logical satellite, J. Climate Appl. Meteor, 2?, 491 

I9H4. 

Townshend, J.R.G,, and CJ. Tucker, Objective as- 
sessment of AVHRR data for land cover mapping, 
int. J. Remote Sens., 5, 492, 1984. 

Tucker, C.J,, C. Vanpraet, E. Boerwinkic, and A. 
Gaston, Satellite remote sensing of total dry matter 
accumulation in the Senegalese Sahel, Remote Sens. 
Environ,, A?, 461, 1983. 

Tucker, C.J., B.N. Holben, and TE. (Joff. Intensive 
forest clearing in Rondonia, Brazil as delected bv 



satellite remote sensing. Remote Sens, Environ,, 15, 

255, 1984a. 

Tucker, CJ, J.A. Gatlin, and S.R. Schneider, Mon- 
itoring vegetation in the Nile Delta with NOAA-6 
and NOAA-7 AVHRR, Photogram, Eng. Remote 
Sens., 50, 5X 1984b. 

Tucker, C.J., J.R.G. Townshend, and T.E. Goff, 
African land cover classification using satellite data. 
Science, 227^ 110, 1985a. 

Tucker, C.J., J.U. Hielkema, and J. Roffey, Satellite 
remote sensing monitoring in desert locust breeding 
areas. Int. J. Remote Sens., 6, 127, 1985b. 

Tucker, CJ., Remote sensing of leaf water content 
in the near infrared. Remote Sens, of Environ., 10. 
23, 1980. 

Turco, R.P, OB. Toon, T.P Ackerman, J.B. Ptil- 
lack, and C. Sagan, Nuclear winter: Global conse- 
quences of multiple nuclear explosions. Science, 

222, 1283, 1983. 

Waring, R. H,. Estimating forest growth and effi- 
ciency in relation to canopv leaf area. Adv. Ecot. 
Res., /.?. 327, 1983. 

Woods, J.D,. Satellite monitoring of the tKcan for 
global climate research, Phil. Trans, of the Royal 
Soc. London, A'309, 337. 1983. 

Wolfe, WL, and G.J. Zissis. (Eds.), Calculated 
change in bulk reflectance of (Kcan water with in- 
creasing concentration of phytoplankton. in The In- 
frared Handbook, Figure 3- 124. Environmental Re- 
search Institute of Michigan, Ann Arbor, MI, 1978. 



59 



/!Sl^W»fi^ 



A,^" 



®