In the present work, different aspects of the measurement and reconstruction of extensive air showers generated by cosmic rays in the energy range between 10 16 eV and 10 18 eV are investigated. KASCADE-Grande detects charged particles of extensive air showers at ground level. From the energy deposits and the arrival times of the particles in the detector stations, the main parameters of the extensive air showers are reconstructed: the impact point, the direction of the shower axis, the total number of electrons (N e ), and the total number of muons (Nμ) of the shower at observation level. The nurnbers of electrons and muons are related to the mass and energy of the primary particle and are the basis for further analysis. The shape of the electron number spectrum reflects the shape of the cosmic ray energy spectrum. Knowing the reconstruction accuracies, the N e spectrum of the KASCADE- Grande data is studied in order to identify possible spectral features. It is found, that observed structures are consistent with the detector resolution and the reconstruction uncertainties. On the basis of the correlation between the total number of muons in the showers and the total number of charged particles, a composition estimation is carried out. A k Nearest Neighbours (kNN) classification procedure is developed: a measured air shower is classified as proton-like or iron-like by finding its nine nearest neighbours in the parameter space of the selected mass sensitive observables. The total number of charged particles (N ch ) and the total number of muons (N μ ) are found to be suitable observables for this purpose. The parameter space is populated by a set of simulated proton and iron induced air showers and an air shower is assigned to the class most common among its neighbours. The misclassification errors, as obtained in the training phase, are taken into account. An increase of the relative number of iron-like induced air showers is observed between 10 16 eV and 3.2·10 16 eV, which could be associated with the decrease of the cosmic ray light component (He, CNO group) due to the break in its energy spectrum. Above 3.2·10 16 eV, about 70% of the events result to be iron-like and ≈ 30% proton-like. At the highest observed energies an increase of light primaries is indicated. The results presented in this work constitute a first attempt to carry out composition analysis with KASCADE-Grande data.
Read full abstract