Optimizing the potential of KM3NeT in detecting core-collapse supernovae

TL;DR

Core-collapse supernovae emit a burst of MeV neutrinos whose detection can illuminate explosion mechanisms. The paper develops a new CCSN search strategy for KM3NeT that leverages intra-DOM multi-PMT coincidences, introducing single-DOM observables and machine-learning discrimination to improve background separation, aided by a data-driven background estimation. The approach yields a 25–46% improvement in the 5σ detection horizon across configurations and models, enabling KM3NeT to reach full Galactic sensitivity upon completion and to contribute robustly to the Supernova Early Warning System (SNEWS). The work demonstrates the power of the multi-PMT DOM design for low-energy neutrino searches in a dynamic seawater environment and provides a practical path to real-time CCSN detection with KM3NeT.

Abstract

Core-collapse supernovae mark the end of life of massive stars. However, despite their importance in astrophysics, their underlying mechanisms remain unclear. Neutrinos that emerge from the dense core of the star offer a promising way to study supernova dynamics. A strategy is presented to improve the potential of the KM3NeT neutrino telescope to detect core-collapse supernovae in our Galaxy or the Large Magellanic Cloud by further exploiting the properties of its optical modules equipped with multiple photomultipliers. A supernova burst is expected to produce a sudden hit rate increase in the KM3NeT detectors. New observables have been defined for individual optical modules that exploit the geometry and time distribution of the detected hits, enabling a better discrimination between signal and background signatures. In addition, a thorough investigation of the related systematic uncertainties is presented for the first time. When implemented, this new methodology allowed KM3NeT to probe 46% more Galactic core-collapse supernova candidates than with the previous trigger strategy, reaching the dense Galactic bulge. It is now expected that, once completed, KM3NeT will achieve full Galactic sensitivity to core-collapse supernovae.

Figures (11)

• Figure 1: Picture representing a single-DOM event with multiplicity $4$, with the hit PMTs highlighted in blue. The $\vec{R}$ vector from Equation \ref{['eq:Rvec']} is shown in red. Its magnitude $|R|$ is correlated with the spatial concentration of the PMT hits. In this particular case, since the PMT hits are close together, $|R|$, $\cos\alpha_{i,i+1}$ and $\cos\alpha_{i,i+2}$ will be close to $1$, while $\sigma(\cos\alpha_{i,i+1})$ and $\sigma(\cos\alpha_{i,i+2})$ will be close to $0$. Here, $\cos\theta \sim -0.7$ as the hits are located in the lower hemisphere of the DOM.
• Figure 2: Multiplicity distributions for simulated CCSN signal events (orange) and background events from a $60$-day period of ORCA6 data (blue). The areas under the distributions are normalized. Only events passing the triggered event veto introduced in van2021km3net are included. For the signal, the $11~M_\odot$ progenitor from tamborra2014neutrino is used.
• Figure 3: Distributions of the total ToT (first row), hit concentration $|R|$ (second row), $\cos\theta$ (third row), and $\langle\Delta t\rangle$ (fourth row) for signal (orange) and background (blue) considering only events with multiplicity $6$ (left), $8$ (center), and $9$ (right). The simulated signal is based on the $11 M_\odot$ model from tamborra2014neutrino. The background is taken from ORCA6 data and is composed of atmospheric muons and radioactive decays. The areas under all the distributions are normalized to unity. Only events that passed the triggered event veto are included.
• Figure 4: Two-dimensional $(|R|, \cos\theta)$ distributions for data and for simulated ORCA6 background events of multiplicity $8$ passing the CCSN preselection. Radioactivity and muon events, simulated as described in Appendix \ref{['sec:bkgmodels']}, are shown in the left and middle panels while the data are shown in the right panel. In the region on the left of the dashed red lines (on the right of the solid red lines) $99\%$ of the background is due to atmospheric muons (radioactivity).
• Figure 5: Radioactivity rates predicted by the simulations described in Appendix \ref{['subsec:radioactivity']} (dashed line) and rescaled for each multiplicity using the procedure described in Section \ref{['subsec:rescale']} (blue triangles). The uncertainty bars on the rescaled rates account for both the finite statistics in the control regions and the modeling uncertainties on the background fractions. The red dots indicate the radioactivity rate estimations obtained using simulations tools, as in the previous CCSN analysis van2021km3net.