Improving the sensitivity of gamma-ray telescopes to dark matter annihilation in dwarf spheroidal galaxies

The Fermi-LAT Collaboration has studied the gamma-ray emission from a stacked population of dwarf spheroidal galaxies and used this information to set constraints on the dark matter annihilation cross section. Interestingly, their analysis uncovered an excess with a test statistic (TS) of 8.7. If interpreted naively, this constitutes a $2.95\sigma$ local excess ($p‐\text{value}=0.003$), relative to the expectations of their background model. In order to further test this interpretation, the Fermi-LAT team studied a large number of blank sky locations and found $\mathrm{TS}>8.7$ excesses to be more common than predicted by their background model, decreasing the significance of their dwarf excess to $2.2\sigma (p‐\text{value}=0.027)$. We argue that these $\mathrm{TS}>8.7$ blank sky locations are largely the result of unresolved blazars, radio galaxies, and star-forming galaxies, and show that multiwavelength information can be used to reduce the degree to which such sources contaminate the otherwise blank sky. In particular, we show that masking regions of the sky that lie within 1° of sources contained in the BZCAT or CRATES catalogs reduce the fraction of blank sky locations with $\mathrm{TS}>8.7$ by more than a factor of 2. Taking such multiwavelength information into account can enable experiments such as Fermi to better characterize their backgrounds and increase their sensitivity to dark matter in dwarf galaxies, the most important of which remain largely uncontaminated by unresolved point sources. We also note that for the range of dark matter masses and annihilation cross sections currently being tested by studies of dwarf spheroidal galaxies, simulations predict that Fermi should be able to detect a significant number of dark matter subhalos. These subhalos constitute a population of subthreshold gamma-ray point sources and represent an irreducible background for searches for dark matter annihilation in dwarf galaxies.

Figure 1

A set of empirically driven population models for a number of gamma-ray source classes, including blazars [21], radio galaxies [22], star-forming galaxies [23], and millisecond pulsars [24]. Also shown is an estimate for the distribution of dark matter subhalos, calculated as in Ref. [31] (using the mass-concentration relationship of Ref. [32]) for the case of $m_{\mathrm{DM}}=35\mathrm{GeV}$ and $\sigma v=2\times 10^{-26}\mathrm{c}{\mathrm{m}}^3/\mathrm{s}$ to $b\overline{b}$.

Figure 2

The distribution of TS values for a population of 5200 randomly selected sky locations constrained to lie at a galactic latitude $|b|>30°$ and at least 1° (5°) away from pointlike (extended) 2FGL sources. We show the distribution of TS values when no additional sky regions are excluded from our analysis (red solid line with shaded Poisson error bars), and when we make regions that lie within 0.5° of a Roma-BZCAT Multi-Frequency Catalog of Blazars (BZCAT) source (blue dotted line), within 1° of a BZCAT source (blue dotted line), within 1° of a Combined Radio All-Sky Targeted Eight-GHz Survey (CRATES) source (yellow solid line), and within 1° of either the BZCAT or CRATES sources (black solid line). In the case that sky locations are constrained to lie at least 1° away from any BZCAT or CRATES source, the density of $\mathrm{TS}>8.7$ locations is reduced by a factor of 2.1.

Figure 3

The ratio of the fraction of “blank sky” locations (at $|b|>30°$) with $\mathrm{TS}>\{4,8,12,16\}$ compared to that predicted by Fermi’s diffuse emission model, as a function of dark matter mass (assuming annihilations to $b\overline{b}$). For all curves, the sky locations are chosen to be at least 1° (5°) from 2FGL point (extended) sources. This ratio is reduced when we further mask around BZCAT and CRATES sources, as denoted in the key. Shaded regions represent Poisson errors. By masking regions near BZCAT and CRATES sources, we can significantly reduce the fraction of the sky with TS values larger than predicted by Fermi’s diffuse model. This conclusion is true for spectral shapes corresponding to a wide range of dark matter masses.