The Fermi -LAT Sky with focus on interstellar emission Gulaugur - - PowerPoint PPT Presentation

The Fermi-LAT Sky

– with focus on interstellar emission – Guðlaugur Jóhannesson gudlaugu@hi.is Dark Matter with Machine Learning, Trieste, April 8 2018 LAT collaboration, Troy Porter & Igor Moskalenko

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Talk overview

This talk is not about DM. This talk is not about ML. The focus of the talk will be high-energy interstellar emission:

An overview of the required “ingredients”. Specifjc issues that may possibly be solved with new analysis techniques.

These issues impact searches for DM signal in the Fermi-LAT sky.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Talk overview

This talk is not about DM. This talk is not about ML. The focus of the talk will be high-energy interstellar emission:

An overview of the required “ingredients”. Specifjc issues that may possibly be solved with new analysis techniques.

These issues impact searches for DM signal in the Fermi-LAT sky.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Talk overview

This talk is not about DM. This talk is not about ML. The focus of the talk will be high-energy interstellar emission:

An overview of the required “ingredients”. Specifjc issues that may possibly be solved with new analysis techniques.

These issues impact searches for DM signal in the Fermi-LAT sky.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Talk overview

This talk is not about DM. This talk is not about ML. The focus of the talk will be high-energy interstellar emission:

An overview of the required “ingredients”. Specifjc issues that may possibly be solved with new analysis techniques.

These issues impact searches for DM signal in the Fermi-LAT sky.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Talk overview

This talk is not about DM. This talk is not about ML. The focus of the talk will be high-energy interstellar emission:

An overview of the required “ingredients”. Specifjc issues that may possibly be solved with new analysis techniques.

These issues impact searches for DM signal in the Fermi-LAT sky.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

8 years of LAT data above 1 GeV (P8 PSF3)

A simple question How many point sources are there in this image?

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

8 years of LAT data above 1 GeV (P8 PSF3)

A simple question How many point sources are there in this image?

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

The fourth Fermi-LAT source catalog (4FGL)

https://fermi.gsfc.nasa.gov/ssc/data/access/lat/8yr_ catalog/ 5098 sources 75 spatially extended 3009 AGNs (2938 Blazars) 564 Other associations 1525 unassociated No DM sources There is no confjrmed DM source at the moment.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

The fourth Fermi-LAT source catalog (4FGL)

https://fermi.gsfc.nasa.gov/ssc/data/access/lat/8yr_ catalog/ 5098 sources 75 spatially extended 3009 AGNs (2938 Blazars) 564 Other associations 1525 unassociated No DM sources There is no confjrmed DM source at the moment.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Pulling the sky apart

Usually modeled in terms of interstellar emission, sources, and isotropic background

= +

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

What is high-energy interstellar emission?

Emission processes Gas p e e Stars π0 γ γ γ γ Typical defjnition Interstellar emission arises from interactions between cosmic-rays (CRs) and the interstellar medium (gas and radiation). CR nuclei:

π0–decay from interactions with gas.

CR electrons (e+ and e−):

Bremsstrahlung from interactions with gas. Inverse Compton (IC) from interactions with radiation.

Very simple and useful “Only” need to know the distribution of CRs, the targets, and the interaction processes.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

What is high-energy interstellar emission?

Emission processes Gas p e e Stars π0 γ γ γ γ Typical defjnition Interstellar emission arises from interactions between cosmic-rays (CRs) and the interstellar medium (gas and radiation). CR nuclei:

π0–decay from interactions with gas.

CR electrons (e+ and e−):

Bremsstrahlung from interactions with gas. Inverse Compton (IC) from interactions with radiation.

Very simple and useful “Only” need to know the distribution of CRs, the targets, and the interaction processes.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Interstellar Matter

Gas and Dust Split into dust and gas phase with a gas-to-dust ratio of ∼ 100.

The gas provides most of the mass. The dust is more important in terms of dynamics.

Interstellar gas is mostly hydrogen (∼ 70% of mass) and helium (∼ 28% of mass). Helium is really diffjcult to observe.

Assumed to follow the same distribution as hydrogen.

Use 21-cm line emission of H i and 2.6-cm line of CO to constrain the distribution. Components by mass

Gas (99%) Dust (1%) H (70%) He (28%) Metals (1.5%)

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

The Interstellar Gas

Trivia Hydrogen observed in three phases:

Atomic (H i): The most massive phase with a large fjlling factor. Scale height approximately 200 pc. Molecular (H2): The densest phase, very

• clumpy. Scale height approximately 100 pc.

Ionized (H ii): The least signifjcant component with a large scale height. Scale height approximately 1 kpc.

Helium assumed to have the same distribution as hydrogen. Rest of the interstellar medium is not interesting as targets for CRs, but it can provide important information on the distribution of Hydrogen. Radial distribution in and near the plane

Moskalenko et al. 2002, ApJ 565

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Atomic Hydrogen

Hyperfjne splitting of the lowest energy state causes line emission at 21 cm that can be used to estimate its column density. Not optically thin along the plane so we need to correct for optical depth

Usually done using the approximation of a homogeneous line of sight NH i(v) = − log

• 1 −

T(v) TS(v) − Tbg

• TS(v)C

where v is the observed Doppler velocity, TS(v) is the spin temperature, T(v) is the brightness of the emission expressed as temperature, Tbg ≈ 2.7 K, and C is a constant. Need to know TS(v) for all lines of sight but usually assume a single value for the entire sky.

This is the “easy” component! Works very well at high latitudes where optical depth is small. Signifjcant issues in the plane where optical depth is larger, especially the inner Galaxy.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Atomic Hydrogen

Hyperfjne splitting of the lowest energy state causes line emission at 21 cm that can be used to estimate its column density. Not optically thin along the plane so we need to correct for optical depth

Usually done using the approximation of a homogeneous line of sight NH i(v) = − log

• 1 −

T(v) TS(v) − Tbg

• TS(v)C

where v is the observed Doppler velocity, TS(v) is the spin temperature, T(v) is the brightness of the emission expressed as temperature, Tbg ≈ 2.7 K, and C is a constant. Need to know TS(v) for all lines of sight but usually assume a single value for the entire sky.

This is the “easy” component! Works very well at high latitudes where optical depth is small. Signifjcant issues in the plane where optical depth is larger, especially the inner Galaxy.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

HI4PI survey (Ben Bekhti, N. et al. 2016, A&A 594)

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Efgects of TS

The fjgure before used the optically thin assumption with TS ≫ TB. Getting the value of TS correct can have a signifjcant impact on the derived column density. The efgect is not uniform across the sky. Map below shows ratio of HI column density assuming TS = 125 K over that using the optically thin assumption.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Measuring TS

TS can be measured if the 21–cm line is seen in both absorption and emission. Observed range from few 10 K up to thousands of K. Limited coverage and strong variations from LOS to LOS. TS values from Strasser & Taylor 2004. Example distribution of TB and TS for two LOS separated by less than half degree.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Molecular Hydrogen

No line emission from cold H2 – Need to use a surrogate tracer.

The most common tracer is the CO molecule that forms under similar conditions as H2. The CO line emission is collisionally excited by H2. The column density of H2 is found observationally to be roughly linearly dependent on the integrated line intensity of the 12CO J = 1 − 0 line emission NH2(v) = XCOWCO(v). XCO has been shown to vary throughout the Galaxy and even in the local ( 1 kpc) medium.

CO is not a perfect tracer of H2 The 12CO J 1 0 line is optically thick:

The line width of the emission from a molecular cloud is correlated with its size and hence the mass. This has to do with interstellar turbulence (Bolatto, Wolfjre & Leroy, 2013, ARA&A 51). There can be large variations in XCO if velocity dispersion is large (e.g. tidal disruption near the GC).

C depletion into CO depends on density on the periphery of clouds because of photo dissociation.

Need other tracers, such as dust emission, to get a complete picture.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Molecular Hydrogen

No line emission from cold H2 – Need to use a surrogate tracer.

The most common tracer is the CO molecule that forms under similar conditions as H2. The CO line emission is collisionally excited by H2. The column density of H2 is found observationally to be roughly linearly dependent on the integrated line intensity of the 12CO J = 1 − 0 line emission NH2(v) = XCOWCO(v). XCO has been shown to vary throughout the Galaxy and even in the local ( 1 kpc) medium.

CO is not a perfect tracer of H2 The 12CO J = 1 − 0 line is optically thick:

The line width of the emission from a molecular cloud is correlated with its size and hence the mass. This has to do with interstellar turbulence (Bolatto, Wolfjre & Leroy, 2013, ARA&A 51). There can be large variations in XCO if velocity dispersion is large (e.g. tidal disruption near the GC).

C depletion into CO depends on density on the periphery of clouds because of photo dissociation.

Need other tracers, such as dust emission, to get a complete picture.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

CO survey (Dame et al. 2001, ApJ 547)

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

The Dark Neutral Medium

Defjned as gas not traced by H i and CO emission line surveys

Was revealed for the fjrst time using γ-rays and dust in analysis of EGRET data (Grenier et

• al. 2005, Science 307).

Has since then been confjrmed in many analysis combining Fermi–LAT data and dust emission.

This gas is likely low density H2 that is not traced properly by CO emission because of photo dissociation. Interstellar dust is mixed with interstellar gas and their column density is roughly linearly related.

Can be used as an alternative tracer of interstellar gas to probe the dark neutral medium.

Dust comes with its own set of issues No distance information in dust emission, need absorption measures for distance estimates. Dust emission is strongly temperature dependent that can be diffjcult to correct for near star-forming regions. The dust to gas ratio is not constant throughout the Galaxy and the column density of dust is not linearly related to that of gas.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

The Dark Neutral Medium

Defjned as gas not traced by H i and CO emission line surveys

Was revealed for the fjrst time using γ-rays and dust in analysis of EGRET data (Grenier et

• al. 2005, Science 307).

Has since then been confjrmed in many analysis combining Fermi–LAT data and dust emission.

This gas is likely low density H2 that is not traced properly by CO emission because of photo dissociation. Interstellar dust is mixed with interstellar gas and their column density is roughly linearly related.

Can be used as an alternative tracer of interstellar gas to probe the dark neutral medium.

Dust comes with its own set of issues No distance information in dust emission, need absorption measures for distance estimates. Dust emission is strongly temperature dependent that can be diffjcult to correct for near star-forming regions. The dust to gas ratio is not constant throughout the Galaxy and the column density of dust is not linearly related to that of gas.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Planck τ353 (Planck Collaboration XI 2014, A&A, 571)

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

HI4PI + Dame CO using XCO = 2 · 1020 cm−2 (K km/s)−1

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

The DNM Sky

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

CO survey (Dame et al. 2001, ApJ 547)

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Kinematic Distances and Rotation Curves

Doppler shift of emission lines used to place gas given a model for its velcotiy fjeld in the Galaxy. Cylindrical rotation is a good approximation for the gas motion. Some known issues Near–far ambiguity in the inner Galaxy. Does not work for directions near dotted line. Limited distance resolution because of thermal and turbulent motion. Non–circular motion. GC LOS SUN

• v1
• v2
• v3

|v2| = |v3|

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Kinematic Distances and Rotation Curves

Doppler shift of emission lines used to place gas given a model for its velcotiy fjeld in the Galaxy. Cylindrical rotation is a good approximation for the gas motion. Some known issues Near–far ambiguity in the inner Galaxy. Does not work for directions near dotted line. Limited distance resolution because of thermal and turbulent motion. Non–circular motion. GC LOS SUN

• v1
• v2
• v3

|v2| = |v3|

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

The Interstellar Radiation Field (ISRF)

Three main components:

Stellar light. Dust re-emission of stellar light. The cosmic microwave background.

Only directly observable from our position ⇒ Need modeling codes to predict its distribution.

Stellar distribution and properties. Dust distribution and properties. Radiative transport.

Porter et al. 2008, ApJ 682 Inverse Compton (IC) cross section is angle dependent so we need angular dependent SEDs throughout the Galaxy.

A skymap of SEDs at each grid point.

Current models are axisymmetric but three dimensional models are in the pipeline. Signifjcant freedom in model properties, especially in the inner Galaxy.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

3D Interstellar radiation fjeld (ISRF)

R12

|b| < 5◦

−180 −135 −90 −45 45 90 135 180 Longitude (deg) 0.01 0.1 1 10 100 1000 Intensity (MJy sr−1) DIRBE 2.2 µm DIRBE 4.9 µm IRAS 25 µm IRAS 60 µm IRAS 100 µm

F98

|b| < 5◦

−180 −135 −90 −45 45 90 135 180 Longitude (deg) 0.01 0.1 1 10 100 1000 Intensity (MJy sr−1) DIRBE 2.2 µm DIRBE 4.9 µm IRAS 25 µm IRAS 60 µm IRAS 100 µm

R12 includes stellar disc, ring, bulge, 4/2 major/minor arms + dust disc with inner hole toward GC. F98 includes ’old’ and ’young’ stellar discs that are warped, spheroidal bar, and warped dust disc with inner hole toward GC. Full radiation transfer modelling using FRaNKIE code. Both models consistent with data. Porter et al. ApJ 846, 67 (2017) /arxiv:1708.00816

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

3D ISRF in the plane

⋆

R12

−15 −10 −5 5 10 15 X (kpc) −15 −10 −5 5 10 15 Y (kpc) 0.1 1 10 Energy Density (eV cm−3)

⋆

F98

−15 −10 −5 5 10 15 X (kpc) −15 −10 −5 5 10 15 Y (kpc) 0.1 1 10 Energy Density (eV cm−3)

(X/kpc, Y/kpc) = (0, 0) (4, 0) (8, 0) (12, 0) (16, 0)

R12

0.1 1 10 100 1000 Wavelength (µm) 0.001 0.01 0.1 1 10 100 Energy Density (eV cm−3 µm−1 µm) (X/kpc, Y/kpc) = (0, 0) (4, 0) (8, 0) (12, 0) (16, 0)

F98

0.1 1 10 100 1000 Wavelength (µm) 0.001 0.01 0.1 1 10 100 Energy Density (eV cm−3 µm−1 µm)

Difgerent integrated energy density distributions that refmect the stellar and dust distributions. In and about the inner Galaxy there is a factor ∼ 5 difgerence between the models.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Cosmic Rays – The Origin

Source classes Require signifjcant power and ability to accelerate particles up to PeV energies. Most likely candidates are:

Supernova remnants Pulsars Stellar wind …

Most associated with massive stars, sparse and transient in nature. SNRs

Ackermann et al. 2013, Science, 339, 807

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Cosmic Rays – The Origin

Source classes Require signifjcant power and ability to accelerate particles up to PeV energies. Most likely candidates are:

Supernova remnants Pulsars Stellar wind …

Most associated with massive stars, sparse and transient in nature. Pulsars

Abeysekara et al. 2017, Science, 358, 911

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Cosmic Rays – The Origin

Source classes Require signifjcant power and ability to accelerate particles up to PeV energies. Most likely candidates are:

Supernova remnants Pulsars Stellar wind …

Most associated with massive stars, sparse and transient in nature. Stellar Wind

Judy Schmidt - https://www.spacetelescope.org/images/potw1608a/

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Propagation of CRs

CRs move at the speed of light, but are afgected by magnetic fjelds in the Galaxy. Circular motion around fjeld lines if they are regular. Turbulence causes the fjeld lines to scramble leading to a difgusive process. Isotropic if fjeld is random. Conditions in the ISM are likely somewhere in between. Lots of uncertainties Current theory far from complete and many details are missing. B-fjeld as spaghetti

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Propagation of CRs

CRs move at the speed of light, but are afgected by magnetic fjelds in the Galaxy. Circular motion around fjeld lines if they are regular. Turbulence causes the fjeld lines to scramble leading to a difgusive process. Isotropic if fjeld is random. Conditions in the ISM are likely somewhere in between. Lots of uncertainties Current theory far from complete and many details are missing. B-fjeld as spaghetti

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Propagation of CRs

CRs move at the speed of light, but are afgected by magnetic fjelds in the Galaxy. Circular motion around fjeld lines if they are regular. Turbulence causes the fjeld lines to scramble leading to a difgusive process. Isotropic if fjeld is random. Conditions in the ISM are likely somewhere in between. Lots of uncertainties Current theory far from complete and many details are missing. B-fjeld as spaghetti

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

GALPROP code for CR transport and difguse emission

Tool for modelling and interpreting CR and non-thermal emissions data for Milky Way and

• ther galaxies in a self consistent and realistic way.

GALPROP can be downloaded/installed locally, or run from a web-browser at the GALPROP website: http://galprop.stanford.edu Newly released v56 includes among other things

Spatial variation in difgusion coeffjcient and Alfvén speed (re-acceleration). Generalized source distributions (2D and 3D) and spectral models. 3D gas and ISRF models. Improved solvers for propagation – dramatic performance increase. New integrators for non-thermal intensity map calculations.

A little warning Note that there is no such thing as “the” GALPROP model.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

GALPROP code for CR transport and difguse emission

Tool for modelling and interpreting CR and non-thermal emissions data for Milky Way and

• ther galaxies in a self consistent and realistic way.

GALPROP can be downloaded/installed locally, or run from a web-browser at the GALPROP website: http://galprop.stanford.edu Newly released v56 includes among other things

Spatial variation in difgusion coeffjcient and Alfvén speed (re-acceleration). Generalized source distributions (2D and 3D) and spectral models. 3D gas and ISRF models. Improved solvers for propagation – dramatic performance increase. New integrators for non-thermal intensity map calculations.

A little warning Note that there is no such thing as “the” GALPROP model.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

3D models for interstellar emission

GALPROP v56 + 3D ISRF + 3D gas + 3D CR source density. 3 CR source density models: CR power injected according to ’Pulsars’ (2D), 50% Pulsars + 50% spiral arms, 100% spiral arms. Propagation parameters adjusted for each to reproduce measurements of CRs near Earth. The models are not tuned to γ-ray data. Reference case: 2D CR, 2D gas, 2D ISRF

⋆

SA0

−15 −10 −5 5 10 15 X (kpc) −15 −10 −5 5 10 15 Y (kpc) 0.1 1 Energy Density (eV cm−3)

⋆

SA50

−15 −10 −5 5 10 15 X (kpc) −15 −10 −5 5 10 15 Y (kpc) 0.1 1 Energy Density (eV cm−3)

⋆

SA100

−15 −10 −5 5 10 15 X (kpc) −15 −10 −5 5 10 15 Y (kpc) 0.1 1 Energy Density (eV cm−3)

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Interstellar Emission for SA100 + R12 + 2D gas

At 10.579315 MeV

−0.3 −0.2 −0.1 0.0 0.1 0.2 0.3

At 1184.057105 MeV

−0.3 −0.2 −0.1 0.0 0.1 0.2 0.3

Fractional residual maps (model/reference - 1) at 10 MeV (left) and 1 GeV (right). Most of the enhancement in the IC component. Squared efgect because spiral arms of CR sources and ISRF align. The ’hole’ at the GC is because the spiral arm cut ofg for R 4 kpc.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Recent developments – Time dependent calculations

CRs are most likely generated in individual sources over short periods of time and not continuously from a smooth distribution. Transition from a smooth “sea” of old propagated CRs to distribution of freshly accelerated sources caused by energy losses. Most notable in IC emission at 100 GeV energies.

Lots of photons collected by Fermi-LAT; HESS Galactic plane survey; HAWC; CTA in the near future. Very important to have a tool that can explore these features.

GALPROP now effjciently calculates full 3D interstellar emissions using time dependent CR injection and/or propagation. Implemented a discrete sampler that can use arbitray underlying source density. Number

• f sources, their duration, and their size are user defjned parameters.

Also allows for non-linear grid spacing to improve resolution where needed.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Time dependent CR source distribution

SA50 source density, propagation parameters determined from calculations using smooth

• distribution. Same average CR injected power.

Sources are 50 pc wide and are on with constant power for 100 kyr. Fractional residuals, comparing to smooth steady state calculations at several energies.

Electrons @ 12 GeV, 600 Myr

⋆

5 10 15 X (kpc) −5 5 Y (kpc) −0.5 −0.25 0.25 0.5 (TD - SS)/SS

Electrons @ 136 GeV, 600 Myr

⋆

5 10 15 X (kpc) −5 5 Y (kpc) −0.5 −0.25 0.25 0.5 (TD - SS)/SS

Electrons @ 1.6 TeV, 600 Myr

⋆

5 10 15 X (kpc) −5 5 Y (kpc) −0.5 −0.25 0.25 0.5 (TD - SS)/SS

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Fractional Residual Movies – IC

IC emission — Time dependent - steady state / steady state. Energy dependent efgects – strongest at the highest energies, but non-negligible over entire LAT energy range.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Fractional Residual Movies – π0-decay

π0-decay emission — Time dependent - steady state / steady state. Efgect not as large as for IC, but still signifjcant.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Time dependence – Summary

Even though source on time is a lot smaller than CR residence time, the resulting calculations show a signifjcant deviation from steady state calculations for both protons and electrons. Fluctutations in interstellar emission of the order of 10% at 1 GeV, up to 60% at 1 TeV for IC emission. Diffjcult to look for faint DM signal in all that noise. Must know the CR source history to make accurate predictions – revert to statistics

• therwise.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Indications of Non-Uniform Difgusion

Johannesson et al. 2016, ApJ, 824, 16 Results from difgerent secondary species indicate difgerent propagation parameters. Hints at non-uniform propagation.

D0 (1028 cm2 s!1)

5 10 15

zh (kpc)

2 4 6 8 10 12 14 16 18

vAlf (km/s)

10 20 30 40

zh (kpc)

2 4 6 8 10 12 14 16 18

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Indications of Non-Uniform Difgusion

Abeysekara et al. 2017, Science, 358, 911 HAWC observations of Geminga and PSR B0656+14 require difgusion coeffjcient that is two orders of magnitude smaller than estimated from CR secondaries. Observations of CR electrons at TeV energies not in agreement with such slow difgusion.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Two zone difgusion model for Geminga

Small difgusion coeffjcient within rd increasing to ISM difgusion at ro. Fixed fraction of spin-down energy converted to electrons and positrons. Signifjcantly afgects the predictions for CRs reaching earth and the predicted emission in the Fermi-LAT energy range Johannesson et al. (2019) / arXiv:1903.05509 Positrons at Earth

100 101 102 103 104 105 Ek [GeV] 0.1 1 10 JkE3

k [GeV2 m−2 s−1 sr−1]

SDZ(∞, ∞) SDZ(30, 50) SDZ(30, 70) SDZ(50, 70) AMS-02

GALPROP calculations Profjle at 8–40 TeV

2 4 6 8 10 Angular distance [deg] 0.0 0.2 0.4 0.6 0.8 1.0 SB [eV cm−2 s−1 deg−2]

SDZ(∞, ∞) SDZ(30, 50) SDZ(30, 70) SDZ(50, 70) HAWC

Spectrum within 15◦

101 102 103 104 Eγ [GeV] 101 102 Fγ [eV cm−2 s−1]

SDZ(∞, ∞) SDZ(30, 50) SDZ(30, 70) SDZ(50, 70) IC aniso IC iso HAWC Magic

Profjle at 3–10 GeV

2 4 6 8 10 Angular distance [deg] 0.0 0.2 0.4 0.6 0.8 1.0 SB [eV cm−2 s−1 deg−2]

SDZ(∞, ∞) SDZ(30, 50) SDZ(30, 70) SDZ(50, 70)

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

Summary

Structure of gas and its column density essential for accurate modeling of the high-energy interstallar emission

Optical depth correction for atomic hydrogen badly constrained (up to factor of few) Molecular hydrogen not perfectly traced by the CO line - XCO factor determined solely from data. Dust emission afgected by dust temperature corrections. Unknown systematics and non-linear efgects.

ISRF requires knowledge of all starlight in the Galaxy as well as distribution and properties

• f the dust.

Nearly an order of magnitude uncertainty in the inner Galaxy.

CR density profjle not necessarily as smooth as previously thought

Sources are generally confjned in space and time resulting in large fmuctuations.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky

However

Despite all of this, the Fermi-LAT collaboration can produce a model of the interstellar emission that is globally correct to within a few 10s of percent, and locally to within a few percent if allowed to be scaled. The trick here is of course that we don’t know how many sources are afgected by the unknown uncertainty.

Gulli Johannesson HI & NORDITA The Fermi-LAT Sky