KASCADE

The Kertzman And Sembroski Cherenkov Airshower and Detector Emulation (KASCADE) (Kertzman & Sembroski, 1994) is a set of detailed, three-dimensional computer simulations, which (i) generates the particles (ksKascade) and subsequent Cherenkov photons (ksLite) produced by VHE gamma-ray and cosmic-ray air show- ers, and (ii) simulates the response of the optics (ksAomega) and triggers (ksTrigger and ksArrayTrigger) of the telescopes. It has the ability to simulate a wide range of primaries, including gamma rays, all ions from proton to iron, electrons and positions. It has been developed and maintained by Glenn Sembroski from Purdue University and Mary Kertzman from DePauw University since 1989. The KASCADE system has been designed as a general tool in investigating a variety of air-shower Cherenkov telescope designs with the goal of maximizing their gamma ray detection sensitivity. It is relatively easy to change telescope configurations as well as detector models.

The current settings of KASCADE simulates particles with 45 discrete energies evenly distributed (in steps of 0.1 log10 GeV) from 20 GeV to 52.265 TeV for each combi-

nation of azimuth, zenith angle, noise and offset. The number of simulated showers decreases roughly following a power law from a total of 1382 showers generated at 20 GeV, to 10 showers at 350 GeV, a constant number of 10 showers are simu- lated from 350 GeV to 25.56 TeV, and 5 showers are simulated from 30.565 TeV to 52.265 TeV.

The particle interactions (described in section 2.1 and section 2.2) in the shower is treated in ksKascade according to a QCD Monte Carlo algorithm proposed by Gaisser & Stanev (1989). This algorithm is also used in neutrino and particle accelerator experiments. The interaction or decay channel for different particles, as well as the “thickness” (in the unit of g cm−2) that a particle can travel before interaction or

decay, are considered. Each particle is tracked in segments of lengths of 0.2 radiation length, until it further interacts, or decays, or hits the ground, or loses enough energy and becomes sub-luminal. The effect of the geomagnetic field and the density profile of the atmosphere is taken into account. Cherenkov photons are generated in ksLite for each segment and traced to the ground. The atmosphere extinction of the Cherenkov light is considered. The location where the photon hits the ground, the time, and the direction of this photon are recorded. The photons are then sorted by their location and arrival time. ksTrigger and ksArrayTrigger divide the ground into grids, each of which roughly correspond to the dimension of a telescope. For each shower, a virtual telescope array is put at different locations (corresponding to different impact distances) on the grids, and trigger decision is made according to the number of Cherenkov photons within the telescope grid and a detailed detector model including the QE of PMTs and the jitter from the mirrors. Random photons are added to represent the NSB noise light, with the total amount reaching the desired noise level (characterized by the pedestal variance value).

The simulation results are processed through VEGAS (see below in section 2.4.2) to produce look-up tables and effective areas. A lookup table has a value of the mean energy/width/length of simulated gamma-ray events with a particular combination of size and impact distance. One sub look-up table is made for each combination of:

1. zenith angle of 1, 10, 20, 30, 40, 50, 60, and 70 deg,

3. offset angle (the angle between the incoming direction of the simulated particle and the optical pointing direction of the telescope) of 0, 0.25, 0.5, 0.75, 1.0, 1.25, 1.5, 1.75, and 2.0 deg, and

4. pedestal variance of 4.73, 5.55, 6.51, 7.64, 8.97, 10.52, 12.35, 14.49, and 17.00. Each all-offset KASCADE look-up table file for upgrade array configuration consists of these 5184 sub look-up tables above. Different look-up tables are made for different array configurations (before and after the relocation of T1 in 2009 and the PMT upgrade in 2012) and season (Winter ATM21 and Summer ATM22).

For all simulated events with a combination of the above parameters at each energy, an “effective area” is calculated based on the simulated triggered rate and the number of simulated events. With the help of lookup table and effective area, standard VERITAS analysis can be performed. One can reconstruct each shower and get the gamma ray map, light curve and spectrum of a source. I have participated in the KASCADE detector modeling (model name “MDL15”) of the VERITAS new-array configuration. A good agreement between the effective area produced by KASCADE using this model and by CORSIKA simulation package is shown in Figure 2.11.

Electron simulation With the goal of study cosmic ray electrons, we generated a set of electron simulations with KASCADE using the detector model “MDL15” following the steps described above. Electrons produce electromagnetic air showers which are identical to gamma-ray showers. Thus it is very difficult, if not impossible, to separate CR electrons and gamma rays. The general strategy for studying electrons with IACTs is to take observations at a region free of any gamma-ray source, and assume that all gamma-ray like air showers are dominated by the diffuse CR electron emission. A comparison between the shape of the MSW distribution from simulated gamma rays, electrons, protons, and helium particles are shown in Figure 2.12. Different from gamma rays, CR electrons are diffuse. This introduces another major difficulty in background (hadronic background) rejection: the whole field of view is occupied

Figure 2.11.: A comparison between the KASCADE 7-sample (black), 12-sample (red), and CORSIKA 7-sample (green) effective areas for the VERITAS new-array configu- ration with medium cuts, using Winter atmosphere profile, at 20 deg zenith angle, 180 deg azimuth angle, 0.5 deg offset, and 5.5 σ above the pedestal variance.

htemp Entries 14712 Mean 0.9969 RMS 0.1087 S.fMSW 0.6 0.8 1 1.2 1.4 0 10000 20000 30000 40000 50000 60000 htemp Entries 14712 Mean 0.9969 RMS 0.1087 S.fMSW {(S.fMSL>0.05&&S.fMSL<1.25&&S.fShowerMaxHeight_KM>6)*50.}

Figure 2.12.: A comparison between the MSW from (i) a fake Crab-like cosmic elec- tron source (green), (ii) a simulated Crab-like gamma-ray source (magenta), (iii) simulated diffuse CR protons (red), and (iv) simulated CR helium cores (blue) using KASCADE simulations with the VERITAS new-array configuration, using Winter at- mosphere profile, at 20 deg zenith angle and 180 deg azimuth angle. Cuts are made to select events with MSL between 0.05 and 1.25, and shower height greater than 6 km. The normalization is arbitrary.

by the “source” (CR electrons) and there is no possible background region. Thus background rejection has to be performed on a event by event basis. This is difficult because although the majority of CR protons and Helium ions produce different shower images, there are still a portion of CR hadronic showers that have similar air shower images after the cuts. The feasibility of selecting electron events based on image parameters is being studied. If this cannot be achieved, we should consider boosted decision tree (BDT) method to perform particle classification.