Quantifying energy fluence and its uncertainty for radio emission from particle cascades in the presence of noise

Sara Martinelli,Tim Huege,Diego Ravignani,Harm Schoorlemmer

2024-07-29

Abstract:Measurements of radio signals induced by an astroparticle generating a cascade present a challenge because they are always superposed with an irreducible noise contribution. Quantifying these signals constitutes a non-trivial task, especially at low signal-to-noise ratios (SNR). Because of the randomness of the noise phase, the measurements can be either a constructive or a destructive superposition of signal and noise. To recover the electromagnetic energy of the cascade from the radio measurements, the energy fluence, i.e. the time integral of the Poynting vector, has to be estimated. Conventionally, noise subtraction in the time domain has been employed for energy fluence reconstruction, yielding significant biases, including even non-physical and negative values. To mitigate the effect of this bias, usually an SNR threshold cut is imposed, at the expense of excluding valuable data from the analyses. Additionally, the uncertainties derived from the conventional method are underestimated, even for large SNR values. This work tackles these challenges by detailing a method to correctly estimate the uncertainties and lower the reconstruction bias in quantifying radio signals, thereby, ideally, eliminating the need for an SNR cut. The development of the method is based on a robust theoretical and statistical background, and the estimation of the fluence is performed in the frequency domain, allowing for the improvement of further analyses by providing access to frequency-dependent fluence estimation.

Instrumentation and Methods for Astrophysics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to accurately quantify the energy flux of radio emissions generated by particle cascades and its uncertainty in the presence of noise. Specifically:

1. **Superposition of Signal and Noise**: When measuring radio signals generated by cascades produced by astroparticles, these signals are always superposed with ineradicable noise. Due to the randomness of the noise phase, the measurement results may be a constructive or destructive superposition of the signal and noise.

2. **Energy Flux Estimation Bias at Low Signal - to - Noise Ratio (SNR)**: Traditional time - domain noise subtraction will lead to significant biases at low SNR, including the generation of unphysical negative values. To reduce this bias, an SNR threshold is usually set, but this will lead to the exclusion of some valuable data.

3. **Underestimation of Uncertainty**: Traditional methods underestimate uncertainty, even at high SNR.

To solve these problems, the paper proposes a frequency - domain - based method, which uses Rice distribution statistical background to correctly estimate the energy flux and its uncertainty, thereby reducing the reconstruction bias and ideally eliminating the need for an SNR threshold. This method allows access to frequency - dependent energy flux estimates, which is helpful for further analysis.

### Formula Summary

- **Total Energy Flux \( f_{\text{tot}} \)**:
\[
f_{\text{tot}}=\epsilon_0 c \Delta t \sum_{\text{pol}}\left(\sum_j E_{\text{pol}}^2(t_j)\right)
\]
where \(\epsilon_0\) is the vacuum permittivity, \(c\) is the speed of light, \(\Delta t\) is the sampling interval, and \(E_{\text{pol}}(t)\) is the component of the observed three - dimensional electric field vector.

- **Energy Flux of Polarization Component \( f_{\text{pol}} \)**:
\[
f_{\text{pol}}=\epsilon_0 c \Delta t \sum_j E_{\text{pol}}^2(t_j)\to f_{\text{tot}}=\sum_{\text{pol}} f_{\text{pol}}
\]

- **Energy Flux Estimation in the Frequency Domain**:
\[
f_{\text{pol}} = 2\epsilon_0 c\Delta\nu\Delta t^2\sum_{j = 0}^{M - 1}|D_{\text{pol}}(v_j)|^2
\]
where \(N\) is the number of samples in the time series, \(D_{\text{pol}}\) is the corresponding complex - valued discrete Fourier transform (DFT), and \(M=\frac{N}{2}\) is the number of positive frequencies.

- **Signal Energy Flux Estimation \( \hat{f}_s \)**:
\[
\hat{f}_s(\nu_j)=
\begin{cases}
\hat{f}_a(\nu_j)-\hat{f}_n(\nu_j)&\text{if }\hat{f}_a(\nu_j)\geq\hat{f}_n(\nu_j)\\
0&\text{if }\hat{f}_a(\nu_j)<\hat{f}_n(\nu_j)
\end{cases}
\]

- **Uncertainty Estimation**:
\[
\delta(\hat{f}_s(\nu_j))=\sqrt{\hat{f}_n(\nu_j)(\hat{f}_n(\nu_j)+2\hat{f}_s(\nu_j))}
\]

Through these formulas and methods, the paper aims to provide a more accurate and reliable way to estimate the energy flux and reduce the bias and underestimation of uncertainty in traditional methods.