1. Igor V. Moskalenko (Stanford) with Seth Digel (SLAC) Troy Porter (LSU) Olaf Reimer (Stanford) Olaf Reimer (Stanford) Andrew W. Strong (MPE) Andrew W. Strong (MPE) Uncertainties in the determination of the diffuse Galactic  -ray emission

2. Igor V. Moskalenko 2 February 23, 2006DM’06/UCLA GLAST LAT Project Diffuse Galactic Gamma-ray Emission ~80% of total Milky Way luminosity at HE !!! Tracer of CR (p, e − ) interactions in the ISM (π 0,IC,bremss): oStudy of CR species in distant locations (spectra & intensities)  CR acceleration (SNRs, pulsars etc.) and propagation oEmission from local clouds → local CR spectra  CR variations, Solar modulation oMay contain signatures of exotic physics (dark matter etc.)  Cosmology, SUSY, hints for accelerator experiments oBackground for point sources (positions, low latitude sources…) Besides: oForeground in studies of the extragalactic diffuse emission oExtragalactic diffuse emission (blazars ?) may contain signatures of exotic physics (dark matter, BH evaporation etc.) Calculation requires knowledge of CR (p,e) spectra in the entire Galaxy

3. Igor V. Moskalenko 3 February 23, 2006DM’06/UCLA GLAST LAT Project Conventional model vs EGRET data 4a-f Conventional model consistent with local p,e spectra exhibits the “GeV excess:” a factor ~2

4. Igor V. Moskalenko 4 February 23, 2006DM’06/UCLA GLAST LAT Project Systematics of the Unknowns EGRET calibration & systematics Data handling Components of the Galactic propagation model  A theoretician view Defense Secretary Donald Rumsfeld’s classification: “…we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns - the ones we don't know we don't know.”

5. Igor V. Moskalenko 5 February 23, 2006DM’06/UCLA GLAST LAT Project Apr 1991 – Jun 2000 (mainly off in the last years of CGRO) Mass : 1830 kg Size : 2.25m ­ / 1.65m   E/E : ~ 25 % Sensitivity : ~ 5 x 10 -8 cm -2 s -1 (> 100 MeV, 10 6 s) Time resolution: 0.1 ms FoV : ~ 0.5 sr (PSF: 6  above 100 MeV) Source loc acc : 0.1°... 1°, PSRs: arcmin EGRET telescope

6. Igor V. Moskalenko 6 February 23, 2006DM’06/UCLA GLAST LAT Project EGRET performed pointed observations <- exposure : P1-4, <30° off-axis angle <- intensity: exposure corrected photons Diffuse emission model (GALDIF: Hunter et al. 1997) : dynamic balance- CR density ~ gas density Exposure Map & Photon Counts (P1-4)

7. Igor V. Moskalenko 7 February 23, 2006DM’06/UCLA GLAST LAT Project Each pointed observation has been corrected for the varying spark chamber performance (standard candles /diffuse emission !!!) -> systematic uncertainties in high end data products -> implicit dependencies in t and E introduced 1 st gas refill2 nd gas refill 3 rd gas refill partial refill Sc B failure last partial refill 100% corrected 100% in-flight P1-4 Spark chamber efficiency vs. time

8. Igor V. Moskalenko 8 February 23, 2006DM’06/UCLA GLAST LAT Project - functional expressions fitted in order to describe the varying performance of the EGRET spark chamber - assumed to be qualitatively the same in the 10 energy bands -> implicit dependencies resulting in systematic uncertainties, here particularly in source & diffuse emission spectra P1-4 Spark chamber efficiency vs. energy & time

9. Igor V. Moskalenko 9 February 23, 2006DM’06/UCLA GLAST LAT Project Other corrections resulting in systematic uncertainties in EGRET measurements: -> qualitative event classification (A, B, or C) during spark chamber track reconstruction (different response functions) -> corrections for distortions under large incident angles (fisheye) -> earth albedo cut (contaminations) as function of energy and vector to earth from instrumental pointing axis -> point source detections (and removal): -> non-gaussian tails in psf -> narrow/wide field psf -> residuals/artefacts in the diffuse maps -> different detection thresholds… -> different operation modes (wide, narrow, strip modes) DON’T FORGET: SATELLITE BASED GAMMA-RAY ASTROMOMY IS (BESIDES ALL SYSTEMATICS) PHOTON LIMITED ! Other Corrections

10. Igor V. Moskalenko 10 February 23, 2006DM’06/UCLA GLAST LAT Project Convolution with EGRET PSF:  Important below 1 GeV  A large effect at low energies especially in latitude affecting the overall spectral shape  Convolution itself is model dependent - depends on spectrum, not fully accounted for Spectral response: 10% effect Model comparison with data

11. Igor V. Moskalenko 11 February 23, 2006DM’06/UCLA GLAST LAT Project UnconvolvedConvolved Longitude profile |b|<5  Effect of Convolution: 70-100 MeV

12. Igor V. Moskalenko 12 February 23, 2006DM’06/UCLA GLAST LAT Project NB here the spatial convolution correction is applied to the DATA based on the model. Hence the DATA changes, not the model (procedure appropriate for spectra) “Convolved data”De-convolved Effect of De-Convolution: Spectrum |l|<30  |b|<5 

13. Igor V. Moskalenko 13 February 23, 2006DM’06/UCLA GLAST LAT Project Effect of the Energy Dispersion EGRET Calibration (Thompson etal 1993) Dispersion 40% → Effect ~10% ~10% E, MeV % FWHM ~25% Corrected Beam calibration ~25% → ~5% effect on spectrum, but it is energy dependent ! Diffuse emission Energy 1 GeV

14. Igor V. Moskalenko 14 February 23, 2006DM’06/UCLA GLAST LAT Project CR Interactions in the Interstellar Mediume+- PHeCNO X,γ gas ISRF e+- π+- P_LiBeB ISM diffusion energy losses energy losses reacceleration reacceleration convection convection etc. etc. π 0 synchrotron IC bremss Chandra GLAST ACE helio-modulation p 42 sigma (2003+2004 data) HESS Preliminary SNR RX J1713-3946 PSF BHeCNO Flux 20 GeV/n CR species:  Only 1 location  modulation e+- π+- BESS AMS PAMELA

15. Igor V. Moskalenko 15 February 23, 2006DM’06/UCLA GLAST LAT Project What it takes to model CR propagation in the Galaxy  Gas distribution (energy losses, π 0, brems)  Interstellar radiation field (IC, e ± energy losses)  Nuclear & particle production cross sections  Gamma-ray production: brems, IC, π 0  Energy losses: ionization, Coulomb, brems, IC, synch  Assume propagation model (Dxx, Dp, Va)  Source distribution & injection spectra  Solve transport equations for all CR species  Fix propagation parameters

16. Igor V. Moskalenko 16 February 23, 2006DM’06/UCLA GLAST LAT Project How It Works: Fixing Propagation Parameters Using secondary/primary nuclei ratio & flux: Diffusion coefficient and its index Propagation mode and its parameters (e.g., reacceleration V A, convection V z ) Radioactive isotopes: Galactic halo size Z h Z h increase B/C Be 10 /Be 9 Interstellar E k, MeV/nucleon E 2 Flux Carbon E k, GeV/nucleon

17. Igor V. Moskalenko 17 February 23, 2006DM’06/UCLA GLAST LAT Project Effect of Cross Sections: Radioactive Secondaries Different size from different ratios… Has a direct impact on the propagation parameters Z halo,kp c ST W 27 Al+p  26 Al Errors in CR measurements (HE & LE)Errors in CR measurements (HE & LE) Errors in production cross sectionsErrors in production cross sections Errors in the lifetime estimatesErrors in the lifetime estimates Different origin of elements (Local Bubble ?)Different origin of elements (Local Bubble ?) nat Si+p  26 Al W ST T 1/2 = ? E k, MeV/nucleon

18. Igor V. Moskalenko 18 February 23, 2006DM’06/UCLA GLAST LAT Project Peak in the Secondary/Primary Ratio Leaky-box model: fitting path-length distribution -> free function B/C Diffusion models:  Diffusive reacceleration  Convection  Damping of interstellar turbulence  Etc. Accurate measurements in a wide energy range may help to distinguish between the models E k, MeV/nucleon too sharp max?

19. Igor V. Moskalenko 19 February 23, 2006DM’06/UCLA GLAST LAT Project Distributed Stochastic Reacceleration Fermi 2-nd order mechanism B Scattering on magnetic turbulences D pp ~ p 2 V a 2 /D D ~ vR 1/3 - Kolmogorov spectrum I cr E strong reacceleration weak reacceleration ΔEΔE Simon et al. 1986 Seo & Ptuskin 1994 1/3 D xx = 5.2x10 28 (R/3 GV) 1/3 cm -2 s -1 V a = 36 km s -1 γ ~ R -δ, δ=1.8/2.4 below/above 4 GV

20. Igor V. Moskalenko 20 February 23, 2006DM’06/UCLA GLAST LAT Project Convection Galactic wind Escape length Xe E v R -0.6 wind or turbulent diffusion resonant diffusion Jones 1979 problem: too broad sec/prim peak D~R 0.6 D xx = 2.5x10 28 (R/4 GV) 0.6 cm -2 s -1 dV/dz = 10 km s -1 kpc -1 γ ~ R -δ, δ=2.46/2.16 below/above 20 GV

21. Igor V. Moskalenko 21 February 23, 2006DM’06/UCLA GLAST LAT Project Distribution of interstellar gas Neutral interstellar medium – most of the interstellar gas mass –21-cm H I spin flip & 2.6-mm CO (standing for H 2 ) Differential rotation of the Milky Way – plus random motions, streaming, and internal velocity dispersions – is largely responsible for the spectrum This is the best – but far from perfect – distance measure available Self-absorption of HI (21cm) and optical depth effects… Dame et al. (1987) Hartmann & Burton (1997) (25°, 0°) W. Keel CO H I G.C. 25° within solar circle outside solar circle

22. Igor V. Moskalenko 22 February 23, 2006DM’06/UCLA GLAST LAT Project Limitations of velocity as distance measure For Clemens (1985) rotation curve V(R), the pattern of line-of-sight velocities (shown from above with dashed lines indicating different Galactic longitudes viewed from the sun) –Contour interval is 20 km s -1 –This is ~ internal velocity dispersion of a large interstellar cloud In the inner Galaxy, the velocity- distance relation is double valued Near the center and anticenter, the velocity is ~0 for all R In the outer Galaxy the gradient of velocity with distance approaches 0 at large R Line of Sight Velocities from Differential Rotation of the Milky Way

23. Igor V. Moskalenko 23 February 23, 2006DM’06/UCLA GLAST LAT Project Overall distribution of interstellar gas Hunter et al. (1997) Pohl & Esposito (1998) Sun Galactic Center Strong & Moskalenko Two studies that started with essentially the same data disagree in many details No unique answer – owing to distance ambiguity, choice of rotation curve, streaming motions, radiative transfer, …

24. Igor V. Moskalenko 24 February 23, 2006DM’06/UCLA GLAST LAT Project Case & Bhattacharya Deriving the SNR distribution + + = Milky Way

25. Igor V. Moskalenko 25 February 23, 2006DM’06/UCLA GLAST LAT Project Injection Spectra (p, e) Ellison etal 2004 |  |  2 |  |>2 HESS RX1713 Hydro-simulation of particle acceleration in SNRs “Model dependent”

26. Igor V. Moskalenko 26 February 23, 2006DM’06/UCLA GLAST LAT Project  0 -production cross section: Dermer’s 1986 recipe SB scaling model @HE Stecker’s Δ-isobar model @LE Interpolation in between BUT:  The data of 1960’s have large syst. errors  Interpolation may produce error

27. Igor V. Moskalenko 27 February 23, 2006DM’06/UCLA GLAST LAT Project Electron Fluctuations/SNR stochastic events GeV electrons 100 TeV electrons GALPROP/Credit S.Swordy Energy losses 10 7 yr 10 6 yr Bremsstrahlung 1 TeV Ionization Coulomb IC, synchrotron 1 GeV Ekin, GeV E(dE/dt) -1,yr Electron energy loss timescale: 1 TeV: ~300 kyr 100 TeV: ~3 kyr

28. Igor V. Moskalenko 28 February 23, 2006DM’06/UCLA GLAST LAT Project Interstellar Radiation Field Systematic errors:  Star distribution –star counts  Grain properties –lab measurements  Gas/dust proportion –extinction curve  “Reasonable parameters”  Compare with ISRF data only at R  Target for CR leptons (IC) Energy losses Model components:  Geometrical: disk, ring, halo, bar, triaxial bulge, arms  87 stellar types (main sequence), AGB & exotics  Dust: silicate, graphite, PAH (5Å – few  m)  Absorbed light gives mid-IR (small grains +PAH) and FIR (~0.1-1  m grains) SMR00 PS05 Systematic uncertainty Local ISRF (PS05) R=0 Optical Scatt.opt. IR PAH 4 kpc 12 kpc 16kpc

29. Igor V. Moskalenko 29 February 23, 2006DM’06/UCLA GLAST LAT Project Heliospheric Modulation & Charge Sign Effect A>0 Ferreira etal 2003 e/p (2.5 GV) M. etal 2002 pbar/p A<0 pbar p pbar/p Parker’s Equation: Convection, gradient & curvature drifts, diffusion, adiabatic energy changes + diffusion tensor (K , K  r, K  ) Bieber etal 1999 pbar/p

30. Igor V. Moskalenko 30 February 23, 2006DM’06/UCLA GLAST LAT Project More Effects: Local Environment Sun Regular Galactic magnetic field may establish preferential directions of CR propagation Sun GC ~200pc Local Bubble:  A hole in the interstellar gas is formed in a series of SN explosions; some shocks may still exist there…  May be important for radioactive CR species, but D xx =?

31. Igor V. Moskalenko 31 February 23, 2006DM’06/UCLA GLAST LAT Project Consistency Check Every model has to be checked back to be consistent with all kinds of data (CR, astrophysics, nuclear physics)  Finally gamma rays

32. Igor V. Moskalenko 32 February 23, 2006DM’06/UCLA GLAST LAT Project Conventional model vs EGRET data 4a-f Conventional model consistent with local p,e spectra exhibits the “GeV excess:” a factor ~2 Taking into account all the uncertainties – remarkable agreement !!!

33. Igor V. Moskalenko 33 February 23, 2006DM’06/UCLA GLAST LAT Project Optimized Model (CR variations) Intermediate latitudes: l=0°-360°,10°<|b|<20° Outer Galaxy: l=90°-270°,|b|<10° Intermediate latitudes: l=0°-360°,20°<|b|<60° Inner Galaxy: l=330°-30°,|b|<5° l=40°-100°,|b|<5° Milagro Strong etal 2004, ApJ 613, 962

34. Igor V. Moskalenko 34 February 23, 2006DM’06/UCLA GLAST LAT Project Conclusion The systematic effects are numerous (data, astrophysical input, models) Not all of them are equally important May contain “unknown unknowns” Errors are hard to estimate The effects are energy-dependent & intrinsically inseparable GLAST will do better! What to do: Exercise caution and apply consistency checks where possible

35. Igor V. Moskalenko 35 February 23, 2006DM’06/UCLA GLAST LAT Project Thank you !