Seasonal Variation of Polar Ice: Implications for Ultrahigh Energy Neutrino Detectors

TL;DR

This work addresses how seasonal firn-density fluctuations in polar ice affect in-ice radio propagation used for ultrahigh-energy neutrino detection. Using the Community Firn Model forced by MERRA-2 data, the authors generate time- and depth-dependent refractive-index profiles and simulate RF propagation with MEEP (FDTD), paraProp (PE), and NuRadioMC ray tracing, across multiple viewing-angle offsets. They quantify how shallow firn fluctuations induce significant year-to-year variations in the fluence and arrival times of direct and refracted/reflected RF signals, with typical effects on the order of 10% in fluence and sub-ns to ns-scale timing variations, depending on geometry. The results imply irreducible uncertainties in neutrino energy and arrival-direction reconstruction for detectors using ice as a medium, particularly for events traversing the shallow firn, and highlight the need for up-to-date, site-specific ice models to mitigate systematic errors in vertex, energy, and direction measurements. These findings have practical implications for current and future radio-based neutrino observatories and may also inform radar echo detection strategies in polar firn.

Abstract

The upper $100 \, \mathrm{m}$ to $150 \, \mathrm{m}$ of the polar ice sheet, called the firn, has a time-dependent density due to seasonal variations in the surface temperature and snow accumulation. We present RF simulations of an in-ice neutrino-induced radio source that show that these density anomalies create variations in the amplitude and propagation times of radio signals propagating through polar firn at an altitude of ${\sim}3000 \, \mathrm{m}$ above sea level. The received power from signals generated in the ice that refract within the upper ${\sim} 15 \, \mathrm{m}$ firn are subject to a seasonal variation on the order of 10\%. These variations result in an irreducible background uncertainty on the reconstructed neutrino energy and arrival direction for detectors using ice as a detection medium.

Figures (23)

• Figure 1: Radio signals that traverse the upper $15 \, \mathrm{m}$ of the ice exhibit a seasonal fluctuation of $\mathcal{O}(0.1)$ in received fluence $\phi^{E}$. Shaded regions highlight the portion of the detection region subject to this seasonal effect for the associated neutrino vertex (shown as triangles at $x=0$, 3 different example vertices shown).
• Figure 2: Top: Two measured density profiles at Summit; $\rho_{RET}(z)$Kyriacou:2025tj & $\rho_{NSP}(z)$Hawley_Morris_McConnell_2008, alongside a density profile $\rho_{AK}(z)$ made at Site A Alley_Koci_1988, located 222 km south of Summit station. A split asymptotic exponential function $\rho_{fit}(z)$ is fit to the NSP (Neutron Scattering Probe). The output of the Herron-Langway $\rho_{HL}(z)$ model is also shown. The left plot zooms in to the shallow firn $z < 15 \, \mathrm{m}$, where the fluctuation in density is most pronounced, and shows the measurement uncertainties for the RET data. The right displays density profiles down to the glacial ice. Bottom: The residuals of the measured density profiles to the best-fit function.
• Figure 3: The CFM-modeled density profile at Summit for June for the decades 1980-1990 and 2010-2020, with the residuals displayed in profile and in histogram (\ref{['fig:rho_prof_cfm_hist']}).
• Figure 4: Above: A color-map of the time-dependent density profile of the shallow firn from 1980 through to 2020. The coloured lines indicate the CFM refractive index model outputs $n_{x,7}(z)$ for month 7 of 2012, 2015 & 2018, which are some of those used to quantify the RF signal variation. Below: shows the discrete density profiles of the aforementioned dates.
• Figure 5: The test pulses (top) & spectra (bottom) injected into the source in the RF simulations, with the former initialized such that the amplitude maximum occurs at $t_{0} = 50 \, \mathrm{ns}$. The spectra are modeled using the AMZ parameterization of a hadronic shower observed from viewing-angle offsets of $\Delta \theta_{VC} = |\theta_{V} - \theta_{C}| = 3.5^{\circ}, \, 5.0^{\circ} \, \& \, 7.5^{\circ}$Alvarez_Mu_iz_2000. Parameters such as the bandwidth and time spread of the pulses are summarized in Table \ref{['tab:signal_properties']}.