Simulation studies of the information content of muon arrival time observations of high energy extensive air showers

By extensive Monte Carlo calculations, using the air shower simulation code CORSIKA, EAS muon arrival time distributions and EAS time profiles up to 320 m distances from the shower centre have been generated, for proton, oxygen and iron induced showers using different hadronic interaction models as Monte Carlo generators. The model dependence and mass discriminating features have been scrutinised for three energies ranges, (1.0-1.78) 1015 eV, (1.0-1.78) 1016 eV and (1.78-3.16) 1016 eV. The present studies have been focussed to the exploration of the information carried by EAS time observables and their correlations in view of features discriminating the mass of the cosmic primary and different hadronic interaction models. Advanced non-parametric statistical methods based on Bayesian decision rules have been applied to scrutinise the EAS observables and to specify quantitatively the results. A first inspection allows some tentative conclusions: 1. The correlations of the local muon arrival time variables with the local muon density improves the true classification rate and discrimination features. It turns out that the correlation can be replaced by a single parameter: Δτq/ρμ. The classification gets improved by correlating the muon arrival times with the shower age, the shower size Ne and Nμtr; 2. Correlating the observation of Δτq/ρμ for two radial distances, the mass discrimination of the primaries get only slightly improved, different from our previous results analysing the (global) arrival times of the foremost muon correlated at two different radial distances; 3. Comparing the classification rates for different muon arrival time quantities: the first quartile, the median and the third quartile, by both considered models QGSJET and VENUS similar results have been obtained.The analysis of the median Δτ0.50 and of the slow component of the arrival time distribution, represented by Δτ0.75, improves the true-classification rate of the oxygen component. From the present results an enhancement of the discriminative features may be expected at larger distances from the shower core ( >150 m) and to higher primary energies (>1016 eV). This finding is of actual interest in view of the KASCADE collaboration to extend the detector array to a larger area, KASCADE GRANDE. (authors)