Event types in H.E.S.S.: a combined analysis for different telescope types and energy ranges

TL;DR

The paper introduces an event-type framework for H.E.S.S. that jointly leverages monoscopic CT5 and stereoscopic CT1-4 data by labeling events as Type M, B, or A based on Hillas parameters. Through MC simulations and Crab Nebula validation, the approach achieves unified energy coverage with 25–45% sensitivity improvements across the range and a lower energy threshold, thanks to time-based cleaning, CT5 integration, and energy-dependent gamma/hadron separation cuts. A joint likelihood fit per type, rather than a simple area sum, combines the distinct reconstruction qualities, while run-by-run corrections for observation conditions are demonstrated as essential to control systematics. The results suggest a robust, full-energy analysis framework suitable for CTAO and future IACTs, capable of improving spectral reconstructions and reducing statistical uncertainties across energies.

Abstract

Imaging atmospheric Cherenkov telescopes (IACTs) are the main technique for detecting gamma rays with energies between tens of GeV and hundreds of TeV. Amongst them, the High Energy Stereoscopic System (H.E.S.S.) has pioneered the use of different telescope types to achieve an energy range as broad as possible. A large, 28 m diameter telescope is used in monoscopic mode to access the lowest energies ($E \gtrsim 30$ GeV), while the four smaller, 12 m diameter telescopes are used in stereoscopic mode to study energies between 150 GeV and 100 TeV. Nevertheless, a combination of both telescope types and trigger strategies has proven to be challenging. In this work, we propose for the first time an analysis based on event types capable of exploiting both telescope types, trigger strategies, and the whole energy range of the experiment. Due to the large differences between monoscopic and stereoscopic reconstructions, the types are defined based on Hillas parameters of individual events, resulting in three types (Type M, Type B, and Type A), each dominating over a different energy range. The performances of the new analysis configurations are compared to the standard configurations in the H.E.S.S. Analysis Package (HAP), Mono and Stereo. The proposed analysis provides optimal sensitivity over the whole energy range, in contrast to Mono and Stereo, which focus on smaller energy ranges. On top of that, improvements in sensitivity of 25-45% are found for most of the energy range. The analysis is validated using real data from the Crab Nebula, showing the application to data of an IACT analysis capable of combining significantly different telescope types with significantly different energy ranges. Larger energy coverage, lower energy threshold, smaller statistical uncertainty, and more robustness are observed. The need for a run-by-run correction for the observation conditions is also highlighted.

Figures (10)

• Figure 1: Top panel: effective area before gamma/hadron separation for a representative zenith angle of $20^{\circ}$. The standard configurations of HAP, Mono (full gray line) and Stereo (full orange line), are compared to the types, Type M (dashed green line), Type B (dashed blue line), and Type A (dashed magenta line), and to the combined Event types (full purple line). Bottom panel: energy ranges at which each type contributes to at least 10% of the total effective area for different zenith angles. The lower limit of Type M is defined as the energy for which its effective area drops below 10% of the maximum of the combined one.
• Figure 2: Effective area after gamma/hadron separation for a representative angle of $20^{\circ}$. In the top panel, Event types (full purple line), is compared to the standard configurations of HAP, Mono (full gray line) and Stereo (full orange line), and to the intermediate configurations with the old definition and new improvements, Mono++ (dashed gray line) and Stereo++ (dashed orange line). In the bottom panel, Event types is compared to the individual types: Type M (dashed green line), Type B (dashed blue line), and Type A (dashed magenta line).
• Figure 3: Relative contribution of each type to the total number of events (full lines) and the test statistic value for a gamma-ray source with flux on the sensitivity level (dashed lines) for a representative angle of $20^{\circ}$. Type M, Type B, and Type A are represented by green, blue, and magenta lines, respectively.
• Figure 4: Separation efficiencies as a function of energy for a representative zenith angle of $20^{\circ}$. The left panels show the gamma efficiency, $\gamma_{\rm{eff}}$, as a function of true energy. The right panels show the q-factor, $\gamma_{\rm{eff}}/\sqrt{\rm{bkg}_{\rm{eff}}}$, as a function of reconstructed energy (as real data is used for the background, true energy is not defined). In the top panels, Event types (full purple line), is compared to the standard configurations of HAP, Mono (full gray line) and Stereo (full orange line), and to the intermediate configurations with the old definition and new improvements, Mono++ (dashed gray line) and Stereo++ (dashed orange line). In the bottom panels, Event types is compared to the individual types: Type M (full green line), Type B (full blue line), and Type A (full magenta line). The shaded areas in the bottom panel show the energy range in which each type dominates the joint analysis. The combined efficiency for Event types was calculated as discussed in Section \ref{['sec:combined']}.
• Figure 5: Angular resolution, defined as the 68% containment radius, as a function of true energy for a representative zenith angle of $20^{\circ}$. In the top panel, Event types (full purple line), is compared to the standard configurations of HAP, Mono (full gray line) and Stereo (full orange line), and to the intermediate configurations with the old definition and new improvements, Mono++ (dashed gray line) and Stereo++ (dashed orange line). In the bottom panel, Event types is compared to the individual types: Type M (full green line), Type B (full blue line), and Type A (full magenta line). The shaded areas in the bottom panel show the energy range in which each type dominates the joint analysis. The combined resolution for Event types was calculated as discussed in Section \ref{['sec:combined']}.