Galactic interstellar emission with DC2

1
Galactic interstellar emissionwith DC2

• Jean-Marc Casandjian
• CEA Saclay

2
goals

• how well can we reconstruct the gas cosmic-ray
  distributions in the plane?
• test the linear combination approximation
• learn about systematics
• look for deviations between the interstellar
  model and DC2 data
• compare the DC2 GALPROP diffuse model with linear
  fit
• fit the DC2 data with GALPROP model as a function
  of energy
• fit the DC2 data with the linear ISM model as a
  function of energy

3
linear fit to GALPROP model

• ideal case
• no Poisson noise
• same input gas IC maps as in GALPROP

Galprop map ?qHI.N(HI) ? qCO.W(CO) qICIC
Cst ? qhiHI qHI01.HI01 qHI2.HI2
qHI3.HI3 qHI4.HI4 qHI5.HI5 qHI6.HI6
qHI78.HI78 GALPROP model linear
decomposition result
4
linear fit to GALPROP model

• nice qHI emissivity gradient
• nice X ratio gradient (Strong et al. 04)

5
linear fit to GALPROP model

• small, but systematic and structured differences
• not understood at the moment
• important impact on source detection and
  isotropic background

Residuals () at 320 MeV
6
DC2 data vs. GALPROP model
Ng qnorm gtmodelmap ?qsou Fsources expo
Iext exposr

• isotropic background
• see Andy Strongs report

E2. Intensity (cm-2 sr-1 s-1 MeV)
galprop
Energy (MeV)
7
DC2 data vs. GALPROP model

• again small, but systematic residuals

binning 0.5
?
8
DC2 data vs. linear ISM model
N? ?qHI.N(HI) ? qCO.W(CO) qICIC Cst
? qhiHI qHI01.HI01 qHI2.HI2 qHI3.HI3
qHI4.HI4 qHI5.HI5 qHI6.HI6
qHI78.HI78 resulting model DC2 photon map
Energy range 100 MeV 200 GeV
9
DC2 data vs. linear ISM model

• good residuals, but obvious fitting difficulties
• several rings are not detected
• at low energy because of the poor angular
  resolution
• at high energy because of poor statistics

100 MeV 200 GeV
?
10
a lot more work needed to understand

• the impact of the choice of IS emission model on
• source detection at small scales (low latitude)
• extragalactic intensity at large scale
• effective energy band E1, E2 over which we can
  trust q(E, r) and X(r)