A Method for Gamma-Ray Energy Spectrum Inversion and Correction

TL;DR

The paper tackles distortions in gamma-ray energy spectra arising from high-count-rate observations (pile-up, dead time, trailing) by coupling physics-based Monte Carlo simulations with a model-independent spectral inversion. It introduces a dual framework: (i) a data-acquisition correction driven by MC simulations that yields a correction function $C(E)$ to produce a corrected RMF$'$ and mitigate distortion, and (ii) a CNN-based inverse energy response method that learns an inverse mapping and provides an explicit inverse response matrix $R^{-1}_{175\times30}$ (with an extended variant $R^{-1}_{\mathbf{N_{ext}\times30}}$) for robust spectral deconvolution. Validation includes self-consistency and cross-validation across 27 spectral models, quantified by KS and AD tests and residual analyses, demonstrating high fidelity for most cases and a conservative systematic error bound via MRV. The approach is applied to GRB 221009A data from HEPP-H, showing consistency with independent GECAM-C measurements and improved spectral recovery via inversion, thereby enabling precise high-rate GRB spectral analysis. Overall, this framework provides a practical, model-lean path to accurate gamma-ray spectra in high-rate regimes, with broad applicability to X-ray, gamma-ray, and particle detectors facing complex instrumental responses.

Abstract

Accurate spectral analysis of high-energy astrophysical sources often relies on comparing observed data to incident spectral models convolved with the instrument response. However, for Gamma-Ray Bursts and other high-energy transient events observed at high count rates, significant distortions (e.g., pile-up, dead time, and large signal trailing) are introduced, complicating this analysis. We present a method framework to address the model dependence problem, especially to solve the problem of energy spectrum distortion caused by instrument signal pile-up due to high counting rate and high-rate effects, applicable to X-ray, gamma-ray, and particle detectors. Our approach combines physics-based Monte Carlo (MC) simulations with a model-independent spectral inversion technique. The MC simulations quantify instrumental effects and enable correction of the distorted spectrum. Subsequently, the inversion step reconstructs the incident spectrum using an inverse response matrix approach, conceptually equivalent to deconvolving the detector response. The inversion employs a Convolutional Neural Network, selected for its numerical stability and effective handling of complex detector responses. Validation using simulations across diverse input spectra demonstrates high fidelity. Specifically, for 27 different parameter sets of the brightest gamma-ray bursts, goodness-of-fit tests confirm the reconstructed spectra are in excellent statistical agreement with the input spectra, and residuals are typically within $\pm 2σ$. This method enables precise analysis of intense transients and other high-flux events, overcoming limitations imposed by instrumental effects in traditional analyses.

Figures (15)

• Figure 1: Light curves from HEPP-H (black) and GECAM-C (red) in the 2--5 MeV energy range. The discrepancies between the two curves highlight the instrumental effects present in the HEPP-H data. The deviations observed in the HEPP-H light curve are attributed to comprehensive instrumental effects, including particle incidence properties, detector energy response, large signal trailing, pile-up, and data acquisition dead time. Such instrumental effects not only distort light curve morphology but also affect spectral measurements, underscoring the importance of their characterization for accurate data analysis.
• Figure 2: Left Panel: Distribution of photon arrival time intervals, modeled with an exponential distribution (mean interval of 30 $\mu$s, equivalent to a count rate of 33000 s$^{-1}$). Right Panel: Cumulative distribution function of the detector energy response (1.7–600 MeV), calculated using the RMF. This RMF was generated through simulations sampling a uniform incident photon energy spectrum and characterizes the expected detector performance.
• Figure 3: Illustrative examples of pulse waveform modeling and pileup simulation. Left Panel: Single-event pulse waveform modeled with a Landau distribution, representing a 1 keV electron-equivalent energy deposition, exhibiting a characteristic rise time of $<$1 $\mu$s and a decay time of 3 $\mu$s. Right Panel: Simulated pulse pile-up generated from the linear superposition of individual waveforms within a [-1, +100] $\mu$s time window around each event arrival.
• Figure 4: Depiction of the peak search algorithm and dead time implementation. Left Panel: Peak detection process, illustrating the 1 $\mu$s detection window (shaded region) initiated when the waveform exceeds the 10 keV threshold. The ADC sampling rate is 54 ns. Right Panel: Illustration of the non-paralyzable dead time. Red and blue vertical lines indicate the onset and termination, respectively, of the dead time period of duration $T_{\mathrm{dead}}$ following an event trigger detection.
• Figure 5: Monte Carlo simulation of spectral generation for $10^5$ events. (a): The simulated incident photon spectrum, with photon counts as a function of energy. (b): The same incident spectrum binned into 175 logarithmically spaced energy channels. (c): Comparison of the deposited energy spectrum (after applying the Energy Response Matrix to the incident spectrum) and the final Data Acquisition (DAQ) output spectrum (after simulating subsequent instrumental effects). Counts observed between discrete energy channels are attributed to signal pile-up or false triggers. (d): The deposited and DAQ output spectra binned into 30 logarithmically spaced energy channels. All simulations used a $1.7$ MeV trigger threshold, a $1\,\mu$s shaping time, and a $240\,\mu$s dead time, with the energy channels covering the 1.37--600 MeV range.