Reconstructing the arrival direction of neutrinos in deep in-ice radio detectors

Abstract

In-ice radio detectors are a promising tool for the discovery of EeV neutrinos. For astrophysics, the implications of such a discovery will rely on the reconstruction of the neutrino arrival direction. This paper describes a first complete neutrino arrival direction reconstruction for detectors employing deep antennas such as RNO-G or planning to employ them like IceCube-Gen2. We will didactically introduce the challenges of neutrino direction reconstruction using radio emission in ice, elaborate on the detail of the algorithm used, and describe the obtainable performance based on a simulation study and discuss its implication for astrophysics.

Deep Learning Based Event Reconstruction for the IceCube-Gen2 Radio Detector

“…The direction reconstruction of neutral-current events showed an average resolution of 4 • for the shallow components and 7.5 • for the deep components. This discrepancy very likely comes from the less sensitive horizontally polarized antennas used by the deep station components and is compatible with previous findings using traditional reconstruction techniques [18,19,23]. The improved methods laid out for the deep components in this contribution will also be applied to the shallow components and further improvements can be expected.…”

“…The detector and trigger simulation was performed using NuRadioReco [14]. The dataset for the shallow components (from [12]) was comprised of 40 million events of charged-and neutral-current events but along a non-uniform energy spectrum ranging from 10 16 eV to 10 19 eV with an underrepresentation of low-energy events. The limited and non-uniform energy spectrum had a negative effect on the reconstruction [12] which is why it was improved here for the deep components.…”

“…Before deep learning was used, the direction reconstruction was done using a forward folding technique [14,[18][19][20]. The deep-learning technique, including normalizing flows as presented in this contribution, allows for event-by-event predictions of the PDF and therefore also uncertainty contours.…”

Status and recent results from the Radio Neutrino Observatory in Greenland (RNO-G)

“…A reconstruction method using forward folding has been presented, that achieves a median direction resolution of 7 • on simulations [19]. Waiting for the final antenna position calibration, a preliminary verification of the pointing is obtained by reconstructing the origin of the impulsive solar flare events occurring during bright flares.…”

Direction reconstruction performance for IceCube-Gen2 Radio

“…The reconstruction algorithm used in these proceedings is the forward-folding algorithm [2][3][4][5]. The forward-folding technique has been successfully used to reconstruct cosmic rays with the shallow upward-facing antennas [4] and has been applied to the shallow detector component of IceCube-Gen2 finding an average resolution of the neutrino direction of σ 68% = 3 • [3].…”