[Paper Review] HESS II Data Analysis with ImPACT

R. D. Parsons, M. Gajdus+1 more|arXiv (Cornell University)|Sep 17, 2015

Astrophysics and Cosmic Phenomena1 references4 citations

TL;DR

This paper presents an adapted version of the ImPACT event reconstruction algorithm for the H.E.S.S. II array, enabling high-performance stereo and mono reconstruction of very high energy gamma-ray showers using mixed telescope designs. The method achieves sub-100 GeV sensitivity with improved angular and energy resolution, demonstrated by a 96σ detection of the Crab Nebula down to 100 GeV and a log-parabolic spectral fit consistent with prior results.

ABSTRACT

The High Energy Stereoscopic System (H.E.S.S.) very high energy gamma-ray telescope array has added a fifth telescope of 600 m$^{2}$ mirror area to the centre of the 4 existing telescopes, lowering its energy threshold to the sub-100 GeV range and becoming the first operational IACT array using multiple telescope designs. In order to properly access this low-energy range however, some adaptation must be made to the existing event analysis. We present an adaptation of the high-performance event reconstruction algorithm, Image Pixelwise fit for Atmospheric Cherenkov Telescopes (ImPACT), for performing mono and stereo event reconstruction with the H.E.S.S. II array. The reconstruction algorithm is based around the likelihood fitting of camera pixel amplitudes to an expected image template, directly generated from Monte Carlo simulations. This advanced reconstruction is combined with a multi variate analysis based background rejection scheme to provide a sensitive and stable analysis scheme in the sub-100 GeV gamma-ray energy range. We will present the latest results of the ImPACT analysis on both simulated and real H.E.S.S. II data, demonstrating the behaviour of the ImPACT analysis at the lowest energies.

Motivation & Objective

• To enable sensitive and stable analysis of H.E.S.S. II data in the sub-100 GeV energy range, which is critical for probing low-energy gamma-ray sources.
• To adapt the high-performance ImPACT reconstruction algorithm for use with the new 600 m² central telescope (CT5), which lowers the energy threshold below 50 GeV.
• To address the challenges of mono events (detected only in CT5) and stereo events (detected in CT5 and at least one other telescope) through tailored reconstruction and background rejection techniques.
• To validate the new analysis chain using both simulated data and real observations of the Crab Nebula, confirming performance and sensitivity at low energies.

Proposed method

• The ImPACT algorithm performs maximum likelihood fitting of camera pixel amplitudes to pre-computed image templates generated from Monte Carlo simulations for both CT5 and the original H.E.S.S. I telescopes.
• For stereo events, a multi-dimensional fit using MINUIT optimizes source position, impact point, energy, and shower maximum depth, with templates interpolated from a library based on fixed primary particle parameters.
• For mono events, a two-step fitting procedure is used: first a free fit with Xmax fixed at expected value, followed by a fit fixing the source position to resolve degeneracies due to limited information.
• Background rejection is implemented via multivariate analysis (MVA) and neural networks trained on Hillas parameters, with optimized cut levels (safe, std, loose) for mono events.
• The analysis uses a likelihood function based on pixel amplitude comparison to templates, enabling precise reconstruction even in low-signal conditions.
• A box cut on Mean Scaled Width/Length is applied as a preliminary background rejection step, with future improvements planned via advanced MVA techniques.

Experimental results

Research questions

• RQ1 Can the ImPACT algorithm be successfully adapted to reconstruct events in the H.E.S.S. II array, which uses two distinct telescope designs (CT5 and H.E.S.S. I telescopes)?
• RQ2 How does the performance of mono and stereo event reconstruction compare in terms of angular and energy resolution, especially at energies below 100 GeV?
• RQ3 To what extent does the new analysis chain improve sensitivity and energy threshold compared to previous H.E.S.S. I-based methods?
• RQ4 How well does the ImPACT reconstruction perform on real data, particularly for strong, well-known sources like the Crab Nebula?

Key findings

• The ImPACT analysis achieved a 96σ significance detection of the Crab Nebula using 7.2 hours of H.E.S.S. II data, with over 7,700 excess γ-like events and a signal-to-background ratio of 2.3.
• The analysis successfully detected gamma-ray emission down to 100 GeV, representing a significant improvement over the previous 300–400 GeV energy threshold.
• The spectral fit to the Crab Nebula data using a log-parabolic model yielded best-fit parameters: φ₀ = 3.951 ± 0.09 × 10⁻¹¹ cm⁻² s⁻¹ TeV⁻¹, α = 2.49 ± 0.03, and β = 0.203 ± 0.02 at E₀ = 1 TeV.
• Stereo reconstruction provided significantly better angular and energy resolution than mono reconstruction across the full energy range, especially above 300 GeV, due to the use of multiple telescope images.
• Mono event reconstruction, while less constrained, achieved stable performance with a 0.2-second fit time per event and effective background suppression via neural network cuts.
• The effective area of stereo analysis is substantially improved above 300 GeV due to the inclusion of the 4 H.E.S.S. I telescopes, confirming the advantage of multi-telescope reconstruction.