Improvement in the gamma-ray energy reconstruction of MAGIC and impact on the spectral analysis of the first Gamma Ray Burst detected at TeV energies
Ludwig-Maximilians-Universität München & Max-Planck-Institut für Physik
External url: https://edoc.ub.uni-muenchen.de/27600/
This thesis is about the development of a novel methodology, the Random Forest based Energy reconstruction (RF-Erec), to determine the energy of the very high energy (VHE, energy larger than 50 GeV) gamma-ray events detected with the MAGIC telescopes. RF-Erec improves the energy reconstruction of gamma-rays, and thereby extends the capabilities of the MAGIC telescopes, compared to the previous methodology for energy reconstruction, which is based on Look-Up-Tables (LUTs-Erec), and has been used over the last decade. When the energy reconstruction is evaluated in the energy resolution, which is the width of Gaussian fit to the distribution of estimation error normalised with energy, RF-Erec is better by a factor more than around 2 in a very wide range of the energies and pointing Zenith distances (Zd). Such improvement is even larger for high Zd observations. Moreover, the standard deviation of the error distribution is substantially smaller, as the long tail seen in the LUTs-Erec disappears in the RF-Erec. This means the energy migration matrix becomes tighter, and the energy estimation of each event becomes more robust. Consequently, RF-Erec enables a reliable spectral measurement even in situations with poor statistics, an event-wise analysis like for Lorentz Invariance Violation (LIV) studies, and a search for anomalies in the spectral shape. The benefit is not only a better accuracy, but also a wider applicability, such as for observations at high Zenith distance, and morphological together with spectral studies. As a side-product of my studies, I also identified the major source of systematic uncertainties in the LUTs-Erec, clarified its mechanism, and confirmed that it is insignificant in the RF-Erec. I evaluated the actual performance improvement in the spectrum reconstruction for different realistic scenarios. One of the cases with the biggest improvement is on a high Zd observation of a gamma-ray source with very steep spectrum. In such spectrum, the energy mis-reconstruction error, namely the spillover to higher energies, complicates substantially the spectral analysis and reduces its reliability. While spillover extends to at most factor of a few in the RF-Erec, it extends to more than one order of magnitude in the LUTs-Erec. Therefore the high energy events estimated using LUTs-Erec are dominated by spillover events, but the RF-Erec keeps the fraction of genuine events to be more than half. I show that the RF-Erec has better ability than LUTs-Erec in estimating the slope and amplitude, as well as more reliable and consistent results among the available strategies for spectral analysis. I have implemented this novel methodology into the standard MAGIC Analysis and Reconstruction Software (MARS). It is now available to the MAGIC collaboration and, starting from year 2020, regarded as part of the standard data analysis framework. The first scientific application of the novel energy estimation was on the data from the MAGIC observations of the gamma-ray burst (GRB) GRB 190114C. It was the first GRB detected significantly at VHE gamma-rays, after more than 15 years of intense searches with the MAGIC telescopes. The spectrum has the steepest shape over one decade in energy (from 0.2 TeV to 2 TeV) that has been ever measured with MAGIC, and with any VHE gamma-ray instrument to date. The steep spectrum is due to the absorption by the Extragalactic Background Light (EBL), that reduces the gamma-ray flux by factors of several hundreds at the highest energies. Moreover the observation was performed at high Zd. Under these observing conditions, the previous method for energy reconstruction, LUTs-Erec, would not have been able to provide a reliable characterization of the VHE gamma-ray spectral shape from GRB 190114C. However, based on my novel methodology, the MAGIC GRB data were analyzed successfully, leading to two Nature papers reporting this historical discovery (Nature, vol.575, p455-458 and p459-463). The rich photon statistics enabled the characterization of the VHE spectra on timescales as short as 1 minute. The analysis revealed the existence of a new emission component extending to about 2 TeV. This new component could be explained as SSC from the external forward shock of the GRB outflow, which has been long predicted by several theorists. The data reveal that the SSC component has approximately the same power-law temporal behavior as the synchrotron component that decreases as the shock decelerates, and that it accounts for substantial amount of the kinetic energy deposited in the outflow from the GRB. Despite the technical difficulties in detecting TeV gamma-rays from GRBs, these results indicate that the SSC emission may be a common process among GRBs, which implies the need to substantially update our knowledge about these extreme phenomena.