Singular Templates of Imaging Cherenkov Shower distribution (STOICS): A background estimation method for Very-High-Energy γ \gamma -ray observations
Ruo-Yu Shang, Karl Kosack, Brian Humensky · 2026
Read Singular Templates of Imaging Cherenkov Shower distribution (STOICS): A background estimation method for Very-High-Energy γ \gamma -ray observations by Ruo-Yu Shang, Karl Kosack, Brian Humensky (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-16525.txt.
Introduction
Shang, Kosack, and Humensky estimate backgrounds for very-high-energy gamma-ray telescopes. Wide sources leave too little empty sky in the field of view. Standard background methods then break down. Their STOICS method mines OFF-run cosmic-ray event patterns. Singular value decomposition compresses those patterns to low dimension. Cosmic-ray-like events in ON runs then predict the gamma-like background. VERITAS archive tests show the modeling holds across varied conditions.
Catalog record
| CREATOR | Ruo-Yu Shang, Karl Kosack, Brian Humensky |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45375 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.16525
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Singular Templates of Imaging Cherenkov Shower distribution (STOICS): A background estimation method for Very-High-Energy γ \gamma -ray observations Ruo-Yu Shang, Karl Kosack, Brian Humensky Abstract Analyses of Imaging Atmospheric Cherenkov Telescope (IACT) data for extended γ \gamma -ray sources face the issue that the field of view does not offer sufficient regions for background estimations. In cases where the source angular size exceeds or occupies a significant part of the field of view, an independent background estimation method is necessary to carry out IACT analyses and to have a better understanding of the systematic uncertainties. The proposed new method utilizes Singular Value Decomposition to extract the low-dimension representations of the distribution of cosmic-ray events in OFF runs and uses cosmic-ray-like events in the ON runs to estimate the background of γ \gamma -like events. Using VERITAS archival data, we demonstrate that the new method is capable of providing reliable background modeling for observations across a wide range of observing conditions. [orcid=0000-0002-9856-989X] [orcid=0000-0001-8424-3621] [orcid=0000-0002-1432-7771] keywords VHE \gamma rays ,Cosmic-ray ,Background ,TeV-halos ,Extended sources , † † corresponding: Ruo-Yu Shang 1 Introduction Extended \gamma -ray sources are ideal laboratories for studying particle acceleration mechanisms and transport. The extended sizes of these sources allow for detailed morphological analyses of these energetic objects. Evolved pulsar wind nebulae (PWNe), supernova remnants (SNRs), and TeV halos are examples of such objects. As an SNR-PWN system ages, the SNR reverse shock starts to interact with the PWN, creating an irregular nebula morphology, and the \gamma -ray emission is a sensitive probe to reveal the interactions that could accelerate the particles to the Tev-PeV regime. A middle-aged ( t\sim 50 kyr) SNR could have reached a size of d\sim 80 pc ( 19 ) . Imaging Atmospheric Cherenkov Telescope (IACT) observatories with excellent \gamma -ray angular resolution (better than 0.08^{\circ} at >1 TeV) are ideal instruments for resolving the morphology details within these objects. However, such a large angular size of an SNR exceeds or fills a significant part of the field-of-view of the current IACT observatories ( 3.5^{\circ}-5.0^{\circ} for the current generation instruments), making it difficult to estimate the background of the \gamma -ray events. These background events are events of cosmic-ray-induced air-showers mis-classified as \gamma -ray events. Although it is possible to remove the majority of cosmic-ray events recorded by the IACT camera, an irreducible background of \gamma -like background events remains. Conventionally, background estimation is achieved using outside sources in the field of view to normalize the background ( 12 ) , which is not available for extended source analysis. New techniques utilizing external information for background estimation need to be developed in the cases of significantly extended sources. Motivated by the need for background estimation methods for extended \gamma -ray sources, we present a novel technique, Singular Templates of Imaging Cherenkov Shower distribution (STOICS), which learns the correlations between the cosmic-ray-like events (i.e. events that failed the \gamma /hadron separation cut) and \gamma -ray-like background events in the OFF observations, and utilizes the cosmic-ray-like events from the ON observations to derive a background model for \gamma -ray-like background events of the ON observations. In this paper, we review the \gamma -hadron separation variables in Section 2 and the conventional methods in Section 3 . Then we introduce the new method in Section 4 , and demonstrate the application of the method using the published VERITAS data in Section 6 . 2 \gamma -hadron separation variables Imaging Atmospheric Cherenkov Telescopes (IACTs) detect particle cascades through Cherenkov radiation produced by charged particles traveling through the air. The \gamma -ray-initiated showers and the cosmic-ray-initiated showers present intrinsically different shower shapes in the atmosphere due to the different particle interactions with the atmosphere. The particles in \gamma -ray showers are produced in electromagnetic interactions, and their momenta are mostly aligned with the parent particles, resulting in narrow elliptic images of these showers when projected onto the camera of an IACT. The particles in cosmic-ray showers are produced in hadronic interactions, the momenta of which can have significant transverse components due to decays of massive pions. Therefore, the images of cosmic-ray showers appear to be more irregular compared to those of \gamma -ray showers. The Hillas parameters, which characterize the shower images as ellipses, are used for shower image parameterization ( 20 ; 17 ) . The primary particle energy E , the shower impact parameter R (the distance from the telescope to the shower axis), and the level of the night sky background (NSB) are important factors for the observed width and length of the ellipse. In order to minimize the dependency on these factors, the length and width of a shower image are compared against a set of \gamma -ray shower simulations at a given interval of shower size s (the summation of the charge of all pixels of the image) and the impact parameter R ( 8 ; 23 ) . The derived parameters are mean reduced scaled length ( MSCL ) MSCL=\frac{1}{N}\sum_{i=1}^{N}\frac{l_{\mathrm{obs},i}(s,R)-\bar{l}_{\mathrm{sim}}(s,R)}{\sigma_{l,\mathrm{sim}}(s,R)}, (1) and mean reduced scaled width ( MSCW ) MSCW=\frac{1}{N}\sum_{i=1}^{N}\frac{w_{\mathrm{obs},i}(s,R)-\bar{w}_{\mathrm{sim}}(s,R)}{\sigma_{w,\mathrm{sim}}(s,R)}, (2) where N is the number of bright images detected by the IACT array, l_{\mathrm{obs},i} ( w_{\mathrm{obs},i} ) is the shower length (width) observed by the i th telescope, \bar{l}_{sim} ( \bar{w}_{sim} ) is the expected length (width) of the simulated events, and \sigma_{l,\mathrm{sim}} ( \sigma_{w,\mathrm{sim}} ) is the 90% containment variation of the length and width of the simulated events. In this paper, we refer to events that pass the mean reduced scaled parameter cuts ( \gamma -ray-like region) as \gamma -like events, and events that fail the mean reduced scaled parameter cuts (cosmic-ray-like region) as CR-like events. Furthermore, modern IACT analyses often use multivariate machine-learning methods for \gamma /hadron separation. For example, the Random Forests (RF) and Boost-Decision Tree (BDT) methods take input of the Hillas parameters and stereoscopic parameters and improve \gamma /hadron separation performance by their better handling of parameter correlations ( 22 ) . In this paper, we will not consider advanced machine-learning classifiers, and we will use the classical Hillas parameters for \gamma /hadron separation for simplicity. In addition to the Hillas (or the multivariate) parameters of a shower image, the arrival direction of the shower, which is reconstructed by tracing the projected shower axis to the origin of the \gamma ray, is an effective parameter to discriminate against isotropic cosmic-ray events as the \gamma rays are charge-neutral and cannot be deflected by magnetic fields. 3 Review of conventional background methods The background to the \gamma -ray events is cosmic-ray (CR)-induced air showers, which can be largely rejected in the analyses using the shower-image parameters described in Section 2 . The background events for the IACT analyses are the remaining cosmic-ray events that pass \gamma /hadron separation cuts and fall into the \gamma -like region. Background events mimic gamma-ray events because they originate from \pi^{0} decays, which initiate electromagnetic cascades indistinguishable from those of primary \gamma rays. The irreducible background cannot be removed despite further improvements in \gamma /hadron separation ( 25 ; 29 ) . Given a number of counts N_{\mathrm{ON}} in the ON region, the expected signal count S is defined as S(\vec{r})=N_{\mathrm{ON}}(\vec{r})-\alpha N_{\mathrm{OFF}}(\vec{r}), (3) where \vec{r} is the sky coordinate of the region, N_{\mathrm{OFF}}(\vec{r})