PREPRINT

Background rejection using image residuals from large telescopes in imaging atmospheric Cherenkov telescope arrays

Laura Olivera-Nieto, Helena X. Ren, Alison M. W. Mitchell, Vincent Marandon, Jim Hinton

Submitted on 23 November 2022

Figures are fetched from the INSPIRE database at: https://inspirehep.net/literature/2514252

Large telescope images of rejected events that would be labelled as gamma-ray candidates in the small telescope reconstruction. In both cases, the right panel shows the ImPACT prediction associated with that event, whereas the left panel shows the actual recorded event image. The event image shown in the panel labelled as "a" contains a clear muon arc, whereas the additional feature in the image in panel "b" is much smaller and has a simpler shape. However, the maximum pixel intensity in this feature is more than double what is expected from NSB noise. Note that the colorbar has been restricted so that the fainter features are visible, since the main shower is much brighter in both cases.
Figure 1: Large telescope images of rejected events that would be labelled as gamma-ray candidates in the small telescope reconstruction. In both cases, the right panel shows the ImPACT prediction associated with that event, whereas the left panel shows the actual recorded event image. The event image shown in the panel labelled as "a" contains a clear muon arc, whereas the additional feature in the image in panel "b" is much smaller and has a simpler shape. However, the maximum pixel intensity in this feature is more than double what is expected from NSB noise. Note that the colorbar has been restricted so that the fainter features are visible, since the main shower is much brighter in both cases.
Schematic diagram representing the algorithm structure.
Figure 2: Schematic diagram representing the algorithm structure.
Fraction of events kept by the ABRIR cut applied after the \hess \textit{standard} cuts for simulated gamma-rays (blue dots), background data from off-runs (red squares) and events taken from a radius of 0.2\degree~from bright gamma-ray sources (green stars), in particular the blazar PKS 2155-304 at zenith angles of 20\degree~and the Crab Nebula for the 40\degree~zenith range. Note that since PKS 2155-304 is an extragalactic source, no gamma-rays are expected to arrive from it above a few TeV due to absorption on the extragalactic background light. When zero events survive the cut, the 68\% containment limit is drawn assuming Poissonian statistics.
Figure 3: Fraction of events kept by the ABRIR cut applied after the \hess \textit{standard} cuts for simulated gamma-rays (blue dots), background data from off-runs (red squares) and events taken from a radius of 0.2\degree~from bright gamma-ray sources (green stars), in particular the blazar PKS 2155-304 at zenith angles of 20\degree~and the Crab Nebula for the 40\degree~zenith range. Note that since PKS 2155-304 is an extragalactic source, no gamma-rays are expected to arrive from it above a few TeV due to absorption on the extragalactic background light. When zero events survive the cut, the 68\% containment limit is drawn assuming Poissonian statistics.
Fraction of events kept by the ABRIR cut applied after the \hess \textit{hard} cuts. Meaning of different panels and colors is the same as in~\ref{fig:rejection_power}. When zero events survive the cut, the 68\% containment limit is drawn assuming Poissonian statistics.
Figure 4: Fraction of events kept by the ABRIR cut applied after the \hess \textit{hard} cuts. Meaning of different panels and colors is the same as in~\ref{fig:rejection_power}. When zero events survive the cut, the 68\% containment limit is drawn assuming Poissonian statistics.
Verification of the performance using data from the Crab Nebula. All ratios shown in this figure are computed as the quantity after applying the ABRIR cut divided by the same quantity before the cut. \textit{Left}: Ratio of background counts with and without the use of ABRIR for both the \textit{standard} (dark blue) and \textit{hard} (orange) cuts. \textit{Middle}: Gamma-ray efficiency computed as the ratio of the resulting effective area for the datasets with and  without ABRIR (solid lines) and as the ratio of the measured excess (data points), again for both sets of cuts using the same color scheme as in the left panel. \textit{Right}: Ratio of the measured flux from the Crab Nebula as a function of energy with and without the use of ABRIR for both sets of initial cuts.
Figure 5: Verification of the performance using data from the Crab Nebula. All ratios shown in this figure are computed as the quantity after applying the ABRIR cut divided by the same quantity before the cut. \textit{Left}: Ratio of background counts with and without the use of ABRIR for both the \textit{standard} (dark blue) and \textit{hard} (orange) cuts. \textit{Middle}: Gamma-ray efficiency computed as the ratio of the resulting effective area for the datasets with and without ABRIR (solid lines) and as the ratio of the measured excess (data points), again for both sets of cuts using the same color scheme as in the left panel. \textit{Right}: Ratio of the measured flux from the Crab Nebula as a function of energy with and without the use of ABRIR for both sets of initial cuts.
Cut efficiencies of the \textit{stereo} reconstruction for the \textit{standard} and \textit{hard} cuts.
Figure 6: Cut efficiencies of the \textit{stereo} reconstruction for the \textit{standard} and \textit{hard} cuts.
Cut efficiencies of the \textit{hybrid} reconstruction using the \textit{standard-hybrid} cuts.
Figure 7: Cut efficiencies of the \textit{hybrid} reconstruction using the \textit{standard-hybrid} cuts.