Searching for quasi-periodicities in short transients: the curious case of GRB 230307A

TL;DR

The paper presents a comprehensive search for quasi-periodic oscillations in the prompt emission of GRB 230307A using Fourier analysis, wavelets, and Gaussian Processes across INTEGRAL/SPI-ACS and Fermi/GBM data. Across methods and instruments, a robust QPO near 1.2 Hz is identified during the burst’s peak, with a second, less robust candidate near 0.34–0.35 s. The authors interpret the 1.2 Hz signal as a jet-rotation/plasma-vorticity signature imprinted during the jet launch and early coasting phase, offering a window into the central engine and jet structure in a neutron-star merger scenario. They emphasize the limitations of non-stationary data analyses for GRBs and advocate for physically motivated models to improve robustness of QPO detections in short transients. Overall, the work highlights how multi-method, multi-instrument QPO analyses can constrain jet physics while illustrating the need for improved statistical frameworks for non-stationary GRB signals.

Abstract

Gamma-ray bursts (GRBs) are the most powerful explosions in the Universe; their energy release reache s us from the end of the re-ionization era, making them invaluable cosmological probes. GRB 230307A i s the second-brightest GRB ever observed in the 56 years of observations since the discovery of the phenomenon in 1967. Follow-up observations of the event at longer wavelengths revealed a lanthanide-ri ch kilonova with long-lasting X-ray emission immediately following the prompt gamma-rays. Moreover, t he gamma-ray light curve of GRB 230307A collected with INTEGRAL's SPectrometer of INTEGRAL AntiCoincidence Shield (SPI-ACS) and Fermi's Gamma-Ray Burst Monitor (GBM). We use Fourier analysis, wavelets and Gaussian Processes to search for periodic and quasi-periodic oscillations (QPOs) in the prompt gamma-ray emission of GRB 230307A. We critically assess all three methods in terms of their robustness for detections of QPOs in fast transients such as GRBs. Our analyses reveal QPOs at a frequency of $\sim 1.2$ Hz (0.82s period) near the burst's peak emission phase, consistent across instruments and detection methods. We also identify a second, less significant QPO at $\sim 2.9$ Hz (0.34s) nearly simultaneously. We hypothesise that the two QPOs originate from the transition epoch at the end of the jet acceleration phase. These QPOs re present plasma circulation periods in vorticity about the jet axis carried outwards to the prompt radiation zone at much larger radii. They are sampled by colliding structures (e.g., shocks) in the spinning jet, possibly marking the evolution of plasma rotation during the final stages of the progenitor neutron star coalescence event.

Figures (17)

• Figure 1: Top: Light curves of GRB 230307A in 50 ms temporal resolution as seen with SPI-ACS (left panel), and the brightest GBM NaI and BGO detectors; NaI 10 (middle panel) and BGO 1 (right panel). The time is in seconds since the trigger time. The orange area within the lower plots marks the time interval of 2.5 to 7.5 s, for which the GBM team issued a warning for possible data problems. Bottom: Fourier periodograms corresponding to the GRB light curves on the top. We show both the unbinned periodogram (black) and the log-binned periodogram (blue). Note that for the Fermi/GBM data, these do include the segment for which a warning was issued. All three periodograms contain strong variability above the instrumental noise limit at all frequencies considered here, and show peaks on top of the broadband variability present across all frequencies in the periodogram.
• Figure 2: Left: Fourier periodogram of the INTEGRAL data with posterior draws from the three models compared via likelihood ratio tests: in green, the power law model; in blue, a power law model with a Lorentzian component for a single QPO; in orange, a model comprising a power law and two Lorentzians. Middle: distribution of the likelihood ratios from 1000 simulated periodograms: the likelihood ratio for the observed periodogram is a clear outlier. Right: same as middle panel, but for the model with two QPOs. Again, the observed likelihood ratio is a clear outlier compared with the null hypothesis (a single QPO).
• Figure 3: Fractional rms amplitude as a function of photon energy for the two Fermi/GBM detectors.
• Figure 4: Left: INTEGRAL light curve of GRB230307A (black), with three random light curves generated from the stochastic process used in the QPO detection methods outlined in this section (orange). The GRB has a well-defined beginning and an end, in between which there exists rapidly changing variability. The simulated light curves also contain variability at a high amplitude, but the overall process does not change throughout the light curve. This is expected for a stationary stochastic process. Right: periodogram of the INTEGRAL data (black) and of the simulated light curves (orange). While the periodogram of the GRB exhibits peaks formed by excess power in correlated neighbouring frequency bins, the periodograms of the stochastic process contain--by design and construction--powers that are statistically independent.
• Figure 5: Left: 2D wavelet transform (spectrogram) of the INTEGRAL observation of GRB230307A. The transform shows transient power at low frequencies in the first $\sim$20 seconds or so of the burst, where the variability is particularly strong. It also shows a short, transient signal between $0-10\,\mathrm{s}$ at 1.2 Hz, similarly to what was identified in the Fourier periodogram. The candidate signal at 0.32 Hz identified in the Fourier periodogram is less apparent here. Right: Fourier periodogram (black) and wavelet periodogram (blue) with the candidate signals found in the Fourier analysis noted as orange dashed lines. The Fourier and wavelet periodograms largely match.