hide
Free keywords:
-
MPINP:
HESS - Abteilung Hinton
Abstract:
Gamma-rays of astrophysical origin having energy Eγ ≳ 10 GeV produce a cascade of secondary particles called an extensive air-shower upon their interaction with the particles in the atmosphere. Detecting these gamma rays is crucial to understanding the nature of high-energy particles and non-thermal emissions in the universe. The High Energy Spectroscopic System (H.E.S.S) uses an array of Cherenkov telescopes to detect the emission from the air-shower particles. H.E.S.S has been successful in detecting TeV gamma rays from various sources like supernovae, AGNs and pulsars. This thesis involves the use of monoscopic observations of the Vela pulsar from the central CT5 telescope of H.E.S.S to detect pulsations in the sub-100 GeV energy range.
The analysis process requires separating the gamma-ray induced showers from the background-dominated cosmic-ray induced showers. This process of separation is carried out in H.E.S.S using a Boosted Decision Tree classifier (BDT) by training on point-source Monte-Carlo simulations of gamma-ray showers. This separation is comparatively difficult for low energies using single telescope observations due to high similarities in the observed properties of the shower. Having additional features that are different for the two classes would be beneficial for this process. This thesis presents the improvement in the performance of the classifier by adding new input parameters to the BDT training process. The parameter among them that is the most helpful in the discrimination is discussed. The results from training the classifier on two different ranges of shower intensities are also presented.
The selection cuts that define the decision boundary between what is considered signal or background are optimized to get the best significance of signal over background. Further, a consistency check using diffuse Monte-Carlo simulations of gamma-ray showers is performed to investigate the effect the trained model could have on showers recorded at different quadrants of the camera.
An improved background separation at low energies would benefit in better distinguishing the pulsar signal from the background. This thesis also presents the application of the improved performance of the classifier to the Vela pulsar to get a spectral energy distribution in the 30 to 100 GeV energy range. Finally, the results are compared with the previous detections of Vela with H.E.S.S.
