Skip to main content

Full text of "TeV Gamma-Ray Sources from a Survey of the Galactic Plane with Milagro"

See other formats

TeV Gamma-Ray Sources from a Survey of the Galactic Plane 

with Milagro 

A. A. Abdo/ B. Allen,^ D. Berley,^ S. Casanova,^ C. Chen,^ D. G. Coyne,^ B. L. Dingus,^ 
R. W. Ellsworth,'^ L. Fleysher/ R. Fleysher/ M. M. Gonzalez,^ J. A. Goodman,^ E. Hays,^ 
C. M. Hoffman,"^ B. Hopper,"^ P. H. Hiintemeyer,^ B. E. Kolterman,'' C. P. Lansdell,^ 
J. T. Linnemann/ J. E. McEnery,^ A. I. Mincer/ P. Nemethy/ D. Noyes,^ J. M. Ryan/° 
P. M. Saz Parkinson,^ A. Shoup," G. Sinnis,^ A. J. Smith,^ G. W. Sullivan,^ V. Vasileiou,^ 
G. P. Walker,^ D. A. Williams,^ X. W. Xu^ and G. B. Yodh^ 


A survey of Galactic gamma-ray sources at a median energy of ~20 TeV has 
been performed using the Milagro Gamma Ray Observatory. Eight candidate 
sources of TeV emission are detected with pre-trials significance > 4.5a in the 
region of Galactic longitude I E [30°, 220°] and latitude b G [-10°, 10°]. Four 
of these sources, including the Crab nebula and the recently published MGRO 
J20 19+37, are observed with significances > 4cr after accounting for the trials 
involved in searching the 3800 square degree region. All four of these sources are 
also coincident with EGRET sources. Two of the lower significance sources are 
coincident with EGRET sources and one of these sources is Geminga. The other 
two candidates are in the Cygnus region of the Galaxy. Several of the sources 
appear to be spatially extended. The fluxes of the sources at 20 TeV range from 
~ 25% of the Crab flux to nearly as bright as the Crab. 

^ Michigan State University, East Lansing, MI 

^ University of California, Irvine, CA 

^ University of Maryland, College Park, MD 

^ Los Alamos National Laboratory, Los Alamos, NM 

^ University of California, Santa Cruz, CA 

^ George Mason University, Fairfax, VA 

^ New York University, New York, NY 

^ Instituto dc Astronomia, Universidad Nacional Autonoma dc Mexico, D.F., MEXICO 
9 NASA Goddard Space Flight Center, Greenbeh, MD 

University of New Hampshire, Durham, NH 

Ohio State University, Lima, OH 


Subject headings: gamma rays: observations 


A survey of the entire Northern Hemisphere sky for so urces of TeV gamma rays has been 
performed using the Milagro Gamma Ray Observatory (lAtkins et aLl |2004| ). This paper 
reports on observations of sources in the region of Galactic longitude / G [30°, 220°] and 
latitude b G [—10°, 10°]. This survey is at a higher energy of ~ 20 TeV and in a different 
region than the survey performed by the HESS atmospheric Cherenkov telescope (ACT) 
above 200 GeV. The HESS su rvey covered I G [-30°, 30 °] and b G [-2°, 2°] and resulted in the 
detection of 14 new sources (lAharonian et al. Il2006bl ). At even lower ener gies, EGRET on 

the C ompton Gamma Ray Observa tory detected 80 sources ab ove 100 MeV (IHartman et al. 

19991 ) and 28 sources above 1 GeV (ILamb and Macomb 1119971 ) within 10 deg of the Galactic 
plane. Fourteen of these GeV sources are in the region surveyed by Milagro. Most of the 
EGRET sources and several of the TeV sources are unidentified without counterparts at lower 
energies. Howeve r, the improved localizati ons of HESS have led to identific ation of supernova 

remnants (SNR) fjAharonian et al. Il2004af). pulsar wind nebulae (PWN) (lAharonian et al 

2005al ). and molecular clouds (lAharonian et al. Il2006al ) as emitters of TeV gamma rays. 

In the Milagro data, four sources (including the Crab) are detected with post-trials 
significance greater than Aa, and four additional lower significance candidates are identified. 
Six of these eight TeV excesses are coincident with the locations of EGRET sources. Many 
of these sour ces have stringent upper limits at TeV energies, such as those from the Whipple 
observatory ( Fegan et al.lboosi ). However, these upper limits weaken for extended sources 

by the ratio of the radius of the source to the angular resolution. The angular resolution of 
Milagro is approximately an order of magnitude larger than ACTs, but is less than half that 
of EGRET, resulting in little reduction in sensitivity for sources up to ~ 2° diameter. The 
TeV flux, spatial morphology, and potential counterparts of these new sources are discussed 

2. Analysis 

The analysis was performed on 2358 days of data collected by Milagro during the opera- 
tional period from July 19, 2000 through Jan 1, 2007. Data from the last 3 years of operation 
were collected after the completion of the outrigger array, which substantially increased the 
sensitivity of Milagro, particularly at high energies. The Milagro data were analyzed using 


the method described in lAbdo. A. et aL I (120071 ). in which the events are weighted based 

on the gamma/hadron separation parameter {A4). The signal and background maps are 
smoothed with the point spread function (PSF), which varies based on the number of hits 
in the events. The st atistical significance of the excess or deficit is computed using eqn. 17 

Li and Ma I (119831 ). 

The Milagro detector is located at latitude 36° N, where the Galactic center and the 
central Galactic bulge are outside the field of view. In this survey of the Northern Galactic 
plane, only events with zenith angle less than 45° are included, which covers declinations 
north of 6 = —7°. In Galactic coordinates, this region is longitude / G [30°, 220°] and 
latitude b G [—10°, 10°], which subtends 9.2% of the entire sky and covers more than half 
of the Galactic plane. The energy threshold and sensitivity of Milagro vary with zenith 
angle. Table 1 shows the median energy and relative sensitivity of the instrument as a 
function of declination for an assumed differential photon power law spectrum of spectral 
index a = —2.3. 

The ba ckground at a given location in celestial coordinates is found use the method 

described in lAtkins et al.l (120031 ) with the modification that the events are weighted. The 
signal map is searched for excesses over the background map with a statistical significance 
> 4.5(7. When a source candidate is found, the region in the vicinity of the candidate 
is removed to prevent double counting, and the search is repeated. With this automated 
procedure, two nearby sources may be identified as a single source candidate or a single 
extended source may be broken up into multiple candidates. 

The event excess in the vicinity of each source candidate is fit to a two-dimensional 
Gaussian to determine the location, spatial extent and flux. The average angular resolutioE0 
of these weighted events is 1.1° and has been measured with observations of the Crab. 
Astrophysical sources are better described by a top hat function than a Gaussian; however, 
a Gaussian approximates a top hat of less than a few degrees diameter that is convolved 
with the Milagro PSF. Therefore, the Gaussian sigma is used to constrain the diameter of 
the top hat function that best fits the angular extent of the source. 

The excess above the assumed isotropic background is calculated from the volume of 
the fit two-dimensional Gaussian. This calculation results in a larger uncertainty than im- 
plied by the significance of the detection, but properly accounts for the flux of extended 
sources. The excess is converted to a flux with a Monte Carlo simulation of extensive air 

^To be consistent with HESS and EGRET, the angular resolution is described by the radius that contains 
68% of the signal events from a point source. In previous publications, the sigma of the Gaussian function 
was referenced, which is ~ 0.7°. 


showers using CORSIKA (Heck et al. I ll998r) and of the Milagro detector using GEANT4 
( lAgostinelli. S. et al. II2003I ). The source fluxes are computed assuming a differential photon 
spectrum of a power law with spectral index a = —2.3 with no cutoff. This is the average 
spectrum for Galactic sources observed by HESS (lAharonian et al. Il2006bl ). The flux is 
quoted at 20 TeV, which is approximately the median energy of the gamma rays from the 
sources detected by Milagro, as seen in table 1. The median energy of Milagro varies with 
source spectra and declination, however the uncertainty in the flux at 20 TeV is relatively 
insensitive to the assumed spectrum and varies by only < 20% when a varies from —2.0 to 

3. Results 

Figure [1] shows a PSF-smoothed map of the Galactic plane, with the color scale in- 
dicating the statistical significance of the Milagro excess or deficit at each point. Table [2] 
gives the locations, fluxes, statistical significances, angular extents, and counterparts for the 
eight source candidates identified with a pre-trials significance in the PSF-smoothed map 
of > 4.5cr. Because several of the candidates are extended beyond the PSF of Milagro, the 
maximum pre-trials significance for a search with a larger bin of size 3° x 3° is also given 
in Table [21 However, the post-trials significances are not based on this a posteriori obser- 
vation, but are calculated from the PSF-smoothed pre-trials significances. A Monte Carlo 
simulation is used to account for the trials involved in searching this 3800 square degree 
region. This simulation predicts that 4% of such searches would result in at least one source 
with > 4.5(7 pre-trials significance anywhere in the region due to background fluctuation. 
The list o f eight candidates inc ludes the Crab and MRGO 2019+37, which was previously 

reported (lAbdo. A. et al. 1120071 ). Excluding these previously known gamma-ray sources, six 
new candidates are identified. The two most significant of these, MGRO J1908-I-06 and 
MGRO J2031+41, exceed 4.5a after accounting for trials. Therefore, the four most signif- 
icant sources in Table 1 are considered definitive TeV gamma-ray source detections. The 
remaining four source candidates, labeled as C1-C4, have post-trials significances less than 
4.5cr and are regarded as lower confidence detections. 

The non-isotropic Galactic diffuse gamma-ray background will contribute to the ob- 
served flux of a source. Diffuse emission is particularly bright in the C ygnus region, where 

five of the eight candidate sources are located. The GALPROP model (jStrong et al. II2004I ) 

predicts that the TeV gamma-ray diffuse emission flux is 3 to 7 times lower t han the observed 

flux in the large Cygnus region defined as I G [65°, 85°] and b G [— 2°. 2°]f lAbdo. A. et al 

20071 ). For individual sources near high concentrations of matter, the diffuse emission will be 

- 5 - 

a larger fraction of the quoted flux. However, due to uncertainties in the model predictions, 
the diffuse emission was not subtracted from the source fluxes. 

The Crab Nebula, a standard reference source for TeV astronomy, is detected at 
15.0(7. The flux derived from the Milagro data is consi stent, within errors, with the flux 
results obtained from atmospheric Cherenkov telescopes (lAharonian et al. Il2004bl ). The fit 
location is 0.11° from the pulsar location, which is consistent with the statistical error. 

MGRO J2019+37 is one of five Milagro sources in the Cygnus region of the Galactic 
plane. It is the most significant sou rce detection by Milagro after the Crab and was discussed 
in depth in lAbdo. A. et al. I (120071 ). More data have been analyzed since the publication of 
that paper, so the flux and significance in Table [2] have been updated. 

MGRO J1908+06 is observed with a pre-trials significance of 8.3a, with a flux that 
is ~80% of the Crab flux. This location is the closest of the eight candidates to the inner 
Galaxy, where the diffuse emission is expected to increase. This sourc e is coincide nt with 
GEV J1907+0557 and with the bright radio, shell-type, SNR G40.5-0.5 JGreen Iboooh . GEV 
J1097-I-0557 was observed for 87 hours by the HEGRA TeV Ob servatory, resulting in a n 
upper limit for a point source of 2.6% of the Crab flux at 700 GeV (lAharonian et al. Il2005bl ). 
The Milagro data are consistent both with a point source and with an extended source with 
diameter up to < 2.6°. The Milagro data are consistent both with a point source and with 
an extended source of diameter up 2.6°. The Milagro flux and HEGRA upper limit taken 
together imply a harder spectrum than other Galactic sources detected in this energy range 
by HESS. However, if the source is extended, the HEGRA upper limit would be increased, 
allowing for a spectrum typical of other TeV sources. The Tibet Air Shower Observatory 
has a similar energy threshold and angular resolution as Milagro, and reports a location 
within 0.9° of MGRO J1908-|-06 as one of eight locations in their survey of the Northern 
Hemisphere sky that is above 4.5(7 (jAmenomori et al. II2005I ). 

MGRO J2031-|-41 is observed with a pre-trials significance of 6.6cr and is located in 
the area with the largest concentration of molecular and atomic gas in the Cygnus region. 
Its location is consistent with EGRET sources GEV J2035+4214 and 3EG J2033+4118 and 

with TEV J2032+413 (lAharonian et al. Il2005dr ). The flux of TeV J2032+413 is measured 
up to ~10 TeV, and is only about one third of the Milagro flux when extrapolated to 20 
TeV. The low energy counterp art for TEV J2032+413 is unclear, but sever al possibilities 
have recently been postulated (IButt et al. I l2006l : lAnchordoqui et al. 1120071 ). The spatial 
extent of the Milagro detection, at 3.0° ± 0.9°, is much larger than the few arcminute extent 
of TEV J2032+413. There must be another source or sources contributing to the Milagro 

- 6 - 

There are two source candidates in the Cygnus region - CI and C2 - that are less 
significant and have no obvious EGRET, PWN, or SNR counterparts. CI is located farthest 
from the Galactic plane at b = —3.9°. C2 may be an extension of MGRO J2019+37 but 
is 2.2° away. C2 is not well fit by a two dimensional Gaussian. For this source, the bin in 
the PSF-smoothed map with the highest excess is used to determine the best location, and 
the fiux is obtained from the excess in a 2.0° x 2.0° angular bin. 

C3 is positionally consistent with Geminga, which is the brightest EGRET source in 
the North ern Hemisphere sky. G eminga is a pulsar that is only 160 pc away and is ~300,000 
years old (jCaraveo et al. Ill996l ). The 5.1a source detected by Milagro is consistent with the 
pulsar location and is extended with a diameter of 2.8° ± 0.8°. The significance increases to 
5.9a in a 3° X 3° bin. The significance of the Milagro excess at the location of the pulsar 
is 4.9a in the PSF-smoothed map. Only pulsed em i ssion was detected by EGRET, but 
a PWN has been observed in X-rays (jCaraveo et al. II2003I ) that delineates the bow shock 
created by the pulsar's motion. The diameter of the excess implies an ~ 8 pc source extent, 

which is consistent with the observations by HESS of more distant PWN ( Aharonian et al. 



a ). Upper limits on the TeV fiux from Gemi nga of ~ 100 mCrab (lAharonian et al 

Akerlof et al. Ill993l : IVishwanath et al. Ill993l ) assumed the emission was from a point 

source and were set at a much lower energy than this observation. If the spectral index is 
~ —2.3 or harder, or if the source is extended, the Milagro detection is consistent with the 
previously reported limits. 

C4 is the least significant source in table [2] at 5.0a. However, as seen in Figure 1, 
the source appears very elongated and the significance increases to 6.3a with the larger 
3° X 3° bin. The source location is consistent with 3EG J2227+6122, GEV J2227+6106, 
an d the SNR G106.6+2.9 with the accompanying Boomerang PWN. This PWN is predicted 
by iBednarek fc BartosikI (120051 ) to be the third brightest TeV PWN in the Northern sky, 
surpassed only by the Crab and PWN G75. 2+0.1, which both have high significance Milagro 

4. Discussion 

A survey of 190° of the Galactic plane at ~ 20 TeV has been performed with Milagro 
with a sensitivity from 3 to 6 xlO~^^ (TeV~^cm~^s~^). Eight sources are identified with 
> 4.5a pre-trials significance above the isotropic cosmic ray background. All four of the high 
confidence sources and two of the lower significance candidates are associated with EGRET 
GeV sources, of which there are only 14 in the search region. Four of these six sources 
appear extended, and the significance of the other two lower significance sources increases 

- 7- 

substantially with a larger bin of size 3° x 3°. The significance for a true point source would 
be expected to decrease slowly for larger bin sizes due to the addition of gamma rays beyond 
the edge of the optimal bin. On the other hand, the significance of a statistical fluctuation 
of background would be expected to decrease inversely proportional to the bin radius. The 
remaining two lower confidence source candidates are within the Cygnus region, where the 
diffuse emission is high and there is a large concentration of EGRET sources. The differential 
photon spectrum that connects the fluxes observed by EGRET at 1 GeV with those observed 
at 20 TeV by Milagro is described by a power law with spectral index a ^ —2.3, except for 
Geminga which is steeper. This spectrum implies a break from the harder spectra observed 
for most of these sources by EGRET above 100 MeV. Four of these sources are associated 
with PWNe, of which only 11 are known in this region (IRoberts II2004I ). and one is associated 
with a SNR. Other members of these classes of sources have also been detected by HESS. 
The angular extents of the Milagro sources are large, as would be expected if the sources 
are nearby, like Geminga. The enhanced sensitivity of GLAST should allow these objects to 
be resolved at energies above a few GeV. In addition, a Milagro energy spectrum analysis is 
being developed that will measure spectra of these sources up to 50 TeV. The morphology at 
different energies provides important clues about the energy of the particles producing the 
radiation and the energy loss mechanisms of these particles. Combining spatial resolution 
over a broad range of energies is key to understanding the Galactic sources that emit the 
highest energy gamma rays. 

We thank Scott Delay, Michael Schneider, and Owen Marshall for their dedicated efforts 
on the Milagro experiment. We also gratefully acknowledge the financial support of the 
National Science Foundation (under grants PHY-0245234, -0302000, -0400424, -0504201, - 
0601080, and ATM-0002744), the Department of Energy (Office of High Energy Physics), Los 
Alamos National Laboratory, the University of California, and the Institute for Geophysics 
and Planetary Physics at Los Alamos National Laboratory. 


Abdo, A. A., et al. 2007, ApJ, 658, L33 
Agostinelh, S., et al. 2003, NIM A 506 2003, 250 
Aharonian, F., et al. 2006, Nature, 439, 695 
Aharonian, F., et al. 2006, ApJ, 636, 777 
Aharonian, F. A., et al. 2005, A&A, 442, L25 

Aharonian, F., et al. 2005, A&A, 439, 635 
Aharonian, F., et al. 2005, A&A, 431, 197 
Aharonian, F. A., et al. 2004, Nature, 432, 75 
Aharonian, F., et al. 2004, ApJ, 614, 897 
Aharonian, F., et al. 1999, A&A, 346, 913 
Akerlof, C. W., et al. 1993, A&A, 274, L17 
Amenomori, M., et al. 2005, ApJ, 633, 1005 
Anchordoqui, L. A., et al 2007, Phys. Rev. D, 75, 063001 
Atkins, R., et al. 2004, ApJ, 608, 680 
Atkins, R., et al. 2003, ApJ, 595, 803 

Bednarek, W., & Bartosik, M. 2005, Journal of Physics G Nuclear Physics, 31, 1465 

Butt, Y. M., et al. 2006, arXiv:astro-ph/0611731 

Caraveo, P. A.,et al. 1996, ApJ, 461, L91 
Caraveo, P. A., et al. 2003, Science, 301, 1345 
Esposito, J. A., et al. 1996, ApJ, 461, 820 
Fegan, S. J., et al. 2005, ApJ, 624, 638 

Green D. A., 2006, 'A Catalogue of Galactic Supernova Remnants (2006 April version) 
Astrophysics Group, Cavendish Laboratory, Cambridge, United Kingdom 

Hartman, R. C, et al. 1999, ApJS, 123, 79 

Heck, D. et al.. Report FZKA 6019, 1998 

Lamb, R. C, & Macomb, D. J. 1997, ApJ, 488, 872 

Li, T.-R, and Ma, Y.-Q. 1983, ApJ, 272, 317 

Roberts, M.S.E., 2004,'^pulsar/pwncat.html 

Strong, A.W., Moskalenko, I.V., & Reimer, O. 2004, ApJ, 613, 962 

- 9 - 

Vishwanath, P. R., et al. 1993, A&A, 267, L5 

This preprint was prepared with the AAS IM^t;X macros v5.2. 

- 10 - 

Galactic Longitjde (degrees) 

Galactic Longitjde {degrees) 

Galactic Longitude (degrees) 

Fig. 1. — Significance map of tlie Galactic plane. The color code shows the pre-trials 
significance in this PSF-smoothed map. The maximum positive value of the color code 
saturates at la although three of the gamma-ray sources are detected with much higher 
statistical significance. The Crab image is inset with the same x and y scale in the bottom 

- 11 - 

Table 1. Sensitivity and Energy 
Response for a ^ cx E'"^-^ 




Energy Range (TeV)'^ 
Median 10%-90% 
































Tlux sensitivity for a point source in units of 
lO-^^TeV^^s^^cm^^ quoted at 20 TeV for a 5(7 

''Energy below wliicli tlie indicated percentage 
of weighted events arc included in the analysis. 

Table 2. Galactic Sources and Source Candidates 




Significance {(7 

Flux'^ at 20 TeV 






post- 3 

= X 3° 

X 10"^'' 









184.5, -5.7 





10.9 ± 1.2 


MGRO J2019-I-37 

75.0, 0.2 





8.7 ± 1.4 

1.1° ±0.5° 

GEV J2020+3658, 
PWN G75. 2+0.1, (1) 

MGRO J1908-I-06 

40.4, -1.0 





8.8 ± 2.4 

< 2.6°(90%CL) 

GEV J1907+0557, 
SNR G40.5-0.5 

MGRO J2031-I-41 

80.3, 1.1 





9.8 ± 2.9 

3.0° ± 0.9° 

GEV J2035+4214, 
TEV J2032-I-413 (2,3) 


77.5, -3.9 





3.1 ± 0.6 

< 2.0°(90%CL) 


76.1, -1.7 




3.4 ± 0.8 


195.7, 4.1 





6.9 ± 1.6 

2.8° ± 0.8° 



105.8, 2.0 





4.0 ± 1.3 

3.4° ± 1.7° 

GEV J2227-I-6106 

SNR G106.6+2.9 

^Thc tabic lists statistical errors only. The systematic pointing error is < 0.3°. 

'^Thc post trials significances account for the trials incurred in searching the 3800 square degree region. 
'^The table lists statistical errors only. The systematic flux error is 30%. 

'^For this high significance detection, the extent was computed using only large events. See lAbdo. A. et al. 1 i2007l") for details 
^Gaussian fit of excess failed for this candidate. 

References. — (1) Abdo ct al. 2007, (2) Butt et al. 2003, (3) Anchordiqui et al. 2007.