In this thesis we present new results for the 99.9% confidence level flux upper limits on the pulsed VHE gamma ray signal coming from the Crab pulsar. In order to achieve optimum hadronic background suppression we implement a new neural network based selection technique and apply it to Cherenkov shower imaging data from the WHIPPLE 10m IACT telescope at Mount Hopkins Arizona. Special emphasis will be given to the fact that the neural network selector is trained with real data exclusively. An energy estimator for gamma ray induced extensive air shower events has been derived from Monte Carlo simulations using the Monte Carlo framework GrISU. This estimator, applied to the image data, serves as input to the neural set selector and is needed to determine the energy dependent flux upper limits. We compare our results to the results from previous studies and the performance of our neural network selection technique to the so-called Supercuts and Optimized Supercuts methods.The new flux upper limits and the new technique show the potential to settle the question about the production mechanism of pulsar radiation. However, the current analysis does not answer this question fully.

In der Astronomie zeigen viele Quellen ein veränderliches Verhalten, das in manchen Fällen sehr kurz sein und im Bereich von Sekunden bis Minuten liegen kann. Beispiele dafür sind Gammablitze (GB) und aktive Galaxienkerne (AGK). Um die zugrundeliegenden physikalischen Prozesse besser zu verstehen, ist eine Messung der Photonenflussänderung notwendig. Allerdings kann die Entdeckung sehr kurzzeitiger Ausbrüche im sehr hochenergetischen Bereich des Spektrums, für welche Cherenkov Teleskope, wie VERITAS, zum Einsatz kommen, schwierig sein. Die Ursache dafür ist der hohe Untergrund und die relativ niedrige Signalrate. Diese Arbeit behandelt die Implementierung fortschrittlicher und für die Entdeckung signifikanter Ratenänderung optimierter, statistischer Methoden (exp-test and Bayesian-Blocks) innerhalb der VERITAS-Analyse. Das Verhalten dieser Methoden wird anhand von Daten minutenlanger Ausbrüche, die mittels Monte-Carlo-Technik für zwei unterschiedliche VERITAS Stadien simuliert wurden (vor und nach der Hardware-Aufrüstung), bewertet und miteinander verglichen. Die fortschrittlichen Methoden verbessern die Empfindlichkeit bei der Entdeckung von kurzzeitigen Ausbrüchen, wenn deren Fluss den des Krebsnebels übersteigt (Krebsnebel-Einheit = K.E.). Es sind nun Ausbrüche von 2 K.E. selbst dann nachweisbar, wenn deren Dauer nur 100 Sek. beträgt. Dies ist mit der Standardmethode nicht möglich. Im darauffolgenden Schritt dieser Arbeit werden das Nachglühen von 6 GB sowie die 450, ebenfalls mit VERITAS erzeugten, Aufnahmen von AGK-Quellen mittels der fortschrittlichen Methoden analysiert. In keiner der AGK-Aufnahmen wird eine signifikante Entdeckung kurzzeitiger Variabilität gemacht, was mit den anerkannten AGK-Modellen übereinstimmt. Auch die Untersuchung des Nachglühens der 6 GB offenbart keine kurzen Ausbrüche im GeV-TeV Bereich. Jedoch ist es für 2 von ihnen möglich, ein oberes Flusslimit von 1.25 K.E. (3,7 · 10^(−10) erg/(cm^2*s) [0,1;10 TeV]) abzuschätzen. / In astronomy, many of the observed sources show a transient behavior. Examples are gamma-ray bursts (GRBs) and active galactic nuclei (AGN). For those source types the variability can be very short, in the order of seconds to minutes. Measuring the flux variations is necessary to understand the underlying physical processes responsible for the emission. However, the detection of very short flares can be difficult in the very high-energy range, in which imaging atmospheric Cherenkov telescopes like VERITAS are operating. This is due to the large background and the comparable low signal rates. This thesis discusses the implementation of advanced statistical methods (exp-test and Bayesian-Blocks) into the VERITAS analysis framework, that are optimized for the detection of significant variations in the event rate. The performance of these methods is evaluated and compared by using Monte Carlo simulations of minute scale flares for two different VERITAS states, pre- and post-hardware-upgrade. It is shown that the advanced methods can improve the detection sensitivity for short flares with high fluxes of more than the Crab flux (Crab unit = C.U.). For example, flares at 2 C.U. with short durations down to 100 sec are now detectable, which is not possible with the standard method. In the next step of this thesis, 6 GRB afterglows and 450 runs of AGN data, observed by VERITAS, are analyzed with the advanced methods. In none of the AGN runs a significant detection of short time variability is made, which is in consistence with the canonical AGN models. The investigation of the 6 GRB afterglows also did not reveal any short flares in the GeV-TeV range. However, for two of them it was possible to estimate an upper flux limit of 1.25 C.U. (3,7 * 10^(−10) erg/(s*cm^2) [0,1;10 TeV]).

In this thesis, we present the Monte Carlo study of two prototypes of tele- scopes for the Cherenkov Telescope Array (CTA) observatory, followed by the first data analysis partially using our reconstruction pipeline based on Random Forests. The Monte Carlo model of the SST-1M prototype is created and val- idated by comparison with data. Using the precise Monte Carlo models, we evaluate the performance of the SST-1M and LST-1 prototypes, working so-far in mono-regime as standalone telescopes, resulting in their energy and angular resolution, and the differential sensitivity. We also present an analysis of the data from the first two Crab Nebula observation campaigns conducted with the LST-1 telescope. In the last part of the thesis, we present a study of aerosol optical depth of the atmosphere above both future sites of the CTA observa- tory, retrieved from photometric measurements of Sun/Moon photometers. We focus on the photometer in-situ calibration for nocturnal measurements and introduce corrections to minimize systematic shifts between diurnal and noc- turnal measurements. Using the developed methods, we present the aerosol characterization of both CTA sites based on the photometric data. 1