New insights from old cosmic rays: A novel analysis of archival KASCADE data

Abstract

Cosmic ray data collected by the KASCADE air shower experiment are competitive in terms of quality and statistics with those of modern observatories. We present a novel mass composition analysis based on archival data acquired from 1998 to 2013 provided by the KASCADE Cosmic ray Data Center (KCDC). The analysis is based on modern machine learning techniques trained on simulation data provided by KCDC. We present spectra for individual groups of primary nuclei, the results of a search for anisotropies in the event arrival directions taking mass composition into account, and search for gamma-ray candidates in the PeV energy domain.

Scaling violation in interaction of cosmic ray hadrons and the nature of the 3 PeV knee in the spectrum of primary cosmic rays

“…It is easy to see that the discussed result of the high-altitude HADRON experiment about domination of the light primaries between (3 − 20) PeV is quite opposite to what was later concluded on composition of cosmic rays in some publications of the KASCADE collabo-ration, e. g. [58,59]. According to their analysis, it is the light component of cosmic rays that starts to die away at E 0 (3 − 5) PeV, whereas the spectrum of heavy primaries (Fe) lasts essentially further and prevails at a few tens of PeV.…”

Scaling violation in interaction of cosmic ray hadrons and the nature of the 3 PeV knee in the spectrum of primary cosmic rays

“…It is easy to see that the discussed result of the high-altitude HADRON experiment about domination of the light primaries between (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) PeV is quite opposite to what was later concluded on composition of cosmic rays in some publications of the KASCADE collaboration, e.g. [58,59]. According to their analysis, it is the light component of cosmic rays that starts to die away at E 0 (3-5) PeV, whereas the spectrum of heavy primaries (Fe) lasts essentially further and prevails at a few tens of PeV.…”

Using Modern Machine Learning Methods on KASCADE Data for Outreach and Education

“…In this work we discussed sharing knowledge on machine learning for astroparticle physics with a broad audience. In particular, we looked into sharing our results achieved in neural network-based analysis of KASCADE particle mass composition [32] within an interactive dedicated software for data exploration and visualization as well as for comparing predictions of neural networks, fit to different hadron mass compositions models.…”

“…The machine-learning models used in the application are described in detail in the work [32] devoted to mass reconstruction of primaries and determining spectra of individual mass-groups composition.…”