GRB 230812B: Multi-band GRB–SN Observations

• GRB 230812B is a bright, nearby burst associated with a broad-lined Type Ic supernova (SN 2023pel), serving as a key test case for prompt-emission physics and jet structure.
• Extensive multi-band follow-up—from X-rays to radio—enabled precise measurements of redshift, afterglow evolution, and jet geometry using advanced modeling techniques.
• The event challenges traditional GRB classifications with its short duration yet collapsar origin, highlighting a decoupled central engine and supernova mechanism.

GRB 230812B is a bright, relatively nearby gamma-ray burst at $z = 0.36$ that is securely associated with the broad-lined Type Ic supernova SN 2023pel. Although its observed duration is close to the traditional long/short boundary, the event is now treated as a collapsar-origin burst because the supernova was established both photometrically and spectroscopically. The burst attracted extensive follow-up across X-ray, ultraviolet, optical, infrared, sub-millimeter, radio, GeV, and prompt keV–MeV bands, making it one of the best observed GRB–SN systems and a test case for prompt-emission physics, structured-jet inference, and the limits of duration-based GRB classification (Hussenot-Desenonges et al., 2023).

1. Discovery, duration, and classification

Fermi-GBM triggered on GRB 230812B at $T_0 = 2023$-08-12 18:58:12 UT. The burst was also observed by Fermi-LAT, GECAM, Konus-Wind, and later by Swift/XRT and Swift/UVOT, while ZTF, KAIT, MASTER, and other facilities identified the optical counterpart. Fermi-LAT detected high-energy photons up to 72 GeV at approximately $T_0+30$ s, and Swift/XRT localized the X-ray afterglow at approximately $25$ ks. Reported prompt durations depend on instrument and energy band: $T_{90} = 3.264 \pm 0.091$ s in 50–300 keV from GBM, $T_{90} \approx 4$ s from GECAM, $T_{90} = 2.95 \pm 1.02$ s in 10–1000 keV in one GBM-based analysis, and $T_{90} = 2.97^{+0.03}_{-0.02}$ s in 10–1000 keV from GRID-05B; Konus-Wind reported a total time of approximately 20 s in 20–1200 keV (Hussenot-Desenonges et al., 2023, Srinivasaragavan et al., 2023, Wang et al., 2024).

Spectroscopy established a redshift near $z \approx 0.36$. GTC spectroscopy gave $z = 0.3602 \pm 0.0006$, while a Keck/DEIMOS analysis reported $T_0 = 2023$0 from host emission lines. The supernova association, together with positive spectral lags, Amati-relation consistency, $T_0 = 2023$1, and $T_0 = 2023$2, places the burst in the Type II or collapsar class despite its $T_0 = 2023$3 s prompt duration. This made GRB 230812B a prominent example of a short-duration supernova-associated burst and therefore a counterexample to the strict identification of short-duration GRBs with compact-object mergers (Wang et al., 2024).

2. Multi-band follow-up and analysis workflow

The follow-up campaign was unusually broad. The GRANDMA network began observations at $T_0 = 2023$4 days and continued for 38 days, coordinating more than 20 professional and several amateur telescopes. The published data set includes over 80 observations in X-ray, ultraviolet, optical, near-infrared, and sub-millimeter bands, supplemented by radio and sub-mm measurements from AMI-LA, VLA, uGMRT, and NOEMA. The event was therefore sampled from Swift/XRT through UVOT, multiple optical filters, $T_0 = 2023$5 and $T_0 = 2023$6 bands, and low-frequency radio/sub-mm observations (Hussenot-Desenonges et al., 2023).

Optical and near-infrared imaging was bias/dark-subtracted and flat-fielded, with astrometry refined using Astrometry.net and coaddition performed with SWarp when appropriate. Forced-aperture photometry used STDPipe/STDWeb with apertures equal to the mean FWHM. Calibration employed Pan-STARRS DR1 for Sloan-like bands, fitting spatially variable zeropoints with a second-order polynomial and retaining color terms when they were significant. UVOT white-band photometry used uvotmaghist with a 7.5″ aperture and a 10–22″ background annulus, and Vega magnitudes were converted to AB by adding 0.8 mag in white. Near-infrared reduction used ESO-eclipse jitter, with astrometry from 2MASS and differential photometry against UKIDSS. Host-galaxy flux at the transient position was estimated from archival imaging and late-time data and then subtracted, after which Milky Way foreground extinction corrections were applied (Hussenot-Desenonges et al., 2023).

Two X-ray-to-NIR spectral energy distributions, at $T_0 = 2023$7 d and $T_0 = 2023$8 d, were fit with standard afterglow theory following Sari, Piran, and Narayan. Both epochs were best fit by a single power law, with negligible host extinction: at $T_0 = 2023$9 d, $T_0+30$0 mag and $T_0+30$1; at $T_0+30$2 d, $T_0+30$3 mag and $T_0+30$4. The adopted extinction law was SMC-like with $T_0+30$5, but because $T_0+30$6, no host-extinction correction beyond the Milky Way foreground was applied. Over the first 5 days, before the supernova contribution became important, the optical/NIR afterglow was fit with

$T_0+30$7

yielding $T_0+30$8 and $T_0+30$9 (Hussenot-Desenonges et al., 2023).

3. Afterglow structure, jet geometry, and circumburst medium

The early afterglow is broadly consistent with a forward-shock synchrotron origin, but the detailed dynamical interpretation is model-dependent. Using the empirical indices $25$0 and $25$1, the standard closure relations do not yield a single value of $25$2 that satisfies both temporal and spectral constraints simultaneously: slow-cooling spectral closure gives $25$3, whereas the temporal closures give $25$4 and $25$5. This inconsistency was taken as evidence for angular jet structure and/or deviations from a simple homogeneous circumburst medium (Hussenot-Desenonges et al., 2023).

A full Bayesian analysis within NMMA combined afterglowpy for the afterglow and sncosmo’s nugent-hyper template for the supernova. The sampled posterior was written as

$25$6

and model selection used Bayes factors. Among top-hat, Gaussian, and power-law structured jets, the preferred model was a power-law structured jet plus supernova, with $25$7, $25$8, and $25$9. The inferred geometry was narrow and viewed close to on-axis: $T_{90} = 3.264 \pm 0.091$0 deg, $T_{90} = 3.264 \pm 0.091$1 deg, $T_{90} = 3.264 \pm 0.091$2 deg, and $T_{90} = 3.264 \pm 0.091$3. The model comparison was decisive, with $T_{90} = 3.264 \pm 0.091$4 and $T_{90} = 3.264 \pm 0.091$5 (Hussenot-Desenonges et al., 2023).

The same study noted substantial early-time X-ray residuals relative to the best-fitting joint afterglow+SN model. Restricted fits gave $T_{90} = 3.264 \pm 0.091$6 from UVOIR data alone and $T_{90} = 3.264 \pm 0.091$7 from X-rays alone. The stated causes were the greater statistical weight of the UVOIR data, the absence in afterglowpy of reverse-shock, inverse-Compton, or early coasting components, and the fact that the supernova template does not contribute in X-rays (Hussenot-Desenonges et al., 2023).

Other analyses reached different environmental conclusions. One prompt-and-afterglow study described the low-energy afterglow as consistent with a slow-cooling ISM forward shock with $T_{90} = 3.264 \pm 0.091$8 and $T_{90} = 3.264 \pm 0.091$9–$T_{90} \approx 4$0, whereas a later broadband keV–VHE modeling study fit the hard-X-ray, GeV, optical, X-ray, and radio evolution with an external forward shock in a wind-like medium, adopting $T_{90} \approx 4$1 erg, $T_{90} \approx 4$2, $T_{90} \approx 4$3, $T_{90} \approx 4$4, $T_{90} \approx 4$5, and $T_{90} \approx 4$6 (Ror et al., 2024, Mohnani et al., 24 Jul 2025). In that wind-based picture, the implied mass-loss rate is $T_{90} \approx 4$7 for $T_{90} \approx 4$8 (Mohnani et al., 24 Jul 2025). The coexistence of ISM-based and wind-based solutions is a central feature of the published literature on this burst.

4. Associated supernova SN 2023pel

The supernova component emerged clearly in the optical light curves. One analysis identified a late-time flattening from $T_{90} \approx 4$9 days onward in both $T_{90} = 2.95 \pm 1.02$0 and $T_{90} = 2.95 \pm 1.02$1 bands, while another placed the onset of the supernova bump in optical/NIR at approximately 5 days after trigger. Spectroscopic confirmation came from OSIRIS+ spectra at 12.12 and 15.12 days in the observer frame, which showed broad Type Ic features typical of GRB-SNe, and from a Keck/DEIMOS spectrum obtained 27.394 days after trigger, corresponding to 15.5 days after the observed $T_{90} = 2.95 \pm 1.02$2-band peak, whose host- and afterglow-subtracted spectrum was classified as broad-lined Type Ic with broad Fe II and Si II absorption troughs (Hussenot-Desenonges et al., 2023, Srinivasaragavan et al., 2023).

Template matching and light-curve decomposition showed that SN 2023pel was roughly comparable in brightness to SN 1998bw but evolved faster. In the GRANDMA-led analysis, the nugent-hyper template fit gave $T_{90} = 2.95 \pm 1.02$3 and $T_{90} = 2.95 \pm 1.02$4 relative to SN 1998bw. In the independent optical study, a two-component PyMultiNest fit to $T_{90} = 2.95 \pm 1.02$5 and $T_{90} = 2.95 \pm 1.02$6 bands, with the afterglow represented by $T_{90} = 2.95 \pm 1.02$7 and $T_{90} = 2.95 \pm 1.02$8 fixed, yielded $T_{90} = 2.95 \pm 1.02$9 and $T_{90} = 2.97^{+0.03}_{-0.02}$0. Both analyses therefore found a supernova about as bright as SN 1998bw but on shorter timescales (Hussenot-Desenonges et al., 2023, Srinivasaragavan et al., 2023).

The photometric properties depend on methodology and wavelength integration. Using SNooPy bolometry in the 450–1050 nm range and subtracting the GRB component, the pseudo-bolometric peak luminosity was $T_{90} = 2.97^{+0.03}_{-0.02}$1 erg s$T_{90} = 2.97^{+0.03}_{-0.02}$2, reached at $T_{90} = 2.97^{+0.03}_{-0.02}$3 days, with an observer-frame half-max width of approximately 22.0 days. The corresponding rest-frame values are $T_{90} = 2.97^{+0.03}_{-0.02}$4 days and a half-max width of approximately 16.2 days. In the same work, SN 2023pel was found to be comparable in peak luminosity to SN 1998bw, to peak sooner than SN 1998bw but later than SN 2006aj, and to decline faster than SN 1998bw, SN 2006aj, and SN 2013dx (Hussenot-Desenonges et al., 2023). A separate study reported an absolute peak $T_{90} = 2.97^{+0.03}_{-0.02}$5-band magnitude of $T_{90} = 2.97^{+0.03}_{-0.02}$6 mag, an observed $T_{90} = 2.97^{+0.03}_{-0.02}$7-band peak at approximately $T_{90} = 2.97^{+0.03}_{-0.02}$8 days, and a bolometric peak luminosity of $T_{90} = 2.97^{+0.03}_{-0.02}$9 from an Arnett-type radioactive-heating model (Srinivasaragavan et al., 2023).

Velocity estimates also depend on phase and method. From the Fe II 5169 Å absorption proxy, one study measured $z \approx 0.36$0 on 2023-08-24 and $z \approx 0.36$1 on 2023-08-27, consistent with the GRB-SN population (Hussenot-Desenonges et al., 2023). Another analysis measured $z \approx 0.36$2 at $z \approx 0.36$3 days relative to the $z \approx 0.36$4-band peak and extrapolated to $z \approx 0.36$5 at maximum light (Srinivasaragavan et al., 2023).

The radioactive-heating analysis derived $z \approx 0.36$6, $z \approx 0.36$7 days, $z \approx 0.36$8, and $z \approx 0.36$9 erg, assuming full gamma-ray trapping, spherical symmetry, and $z = 0.3602 \pm 0.0006$0 (Srinivasaragavan et al., 2023). The same study independently tested a Cano et al. (2016) magnetar model against the X-ray and optical light curves and found inconsistent magnetar parameters between bands and a required optical flux-stretch factor $z = 0.3602 \pm 0.0006$1, which disfavored a magnetar-dominated explanation for the supernova luminosity under the assumed dipolar, unevolving-field model (Srinivasaragavan et al., 2023).

5. Prompt emission, GeV photons, and emission-mechanism debates

The prompt-emission literature on GRB 230812B is unusually diverse because different instruments provided different leverage on saturation, spectral evolution, and model testing. In a GBM-based prompt analysis, the time-integrated spectrum from $z = 0.3602 \pm 0.0006$2 to 5 s over 0.01–40 MeV was best fit by Band+BB with $z = 0.3602 \pm 0.0006$3, $z = 0.3602 \pm 0.0006$4 keV, $z = 0.3602 \pm 0.0006$5, and $z = 0.3602 \pm 0.0006$6 keV. The Deviance Information Criterion preferred Band+BB with DIC $z = 0.3602 \pm 0.0006$7, improving over physical synchrotron with DIC $z = 0.3602 \pm 0.0006$8 and Band alone with DIC $z = 0.3602 \pm 0.0006$9. Time-resolved fitting showed that the rising phase preferred CPL+BB, the decay phase preferred Band+BB, and that $T_0 = 2023$00 and $T_0 = 2023$01 tracked the prompt flux, while $T_0 = 2023$02 tracked the flux up to approximately 3 s and then remained roughly constant near 7 keV with larger uncertainties. That analysis interpreted the thermal component as evidence for a hybrid jet composition (Ror et al., 2024).

A separate GRID-based analysis, using unsaturated prompt data over 10 keV–2 MeV, instead found that a cutoff power law was preferred throughout the time-resolved spectra. Its time-integrated prompt fit over 0.01–4.97 s gave $T_0 = 2023$03 and $T_0 = 2023$04 keV, while time-resolved fits showed hard-to-soft evolution from $T_0 = 2023$05 MeV down to $T_0 = 2023$06 keV and positive spectral lags across energy bands. Within a time-evolving synchrotron model with a decaying magnetic field and a single episode of electron injection, the best-fit parameters were $T_0 = 2023$07, $T_0 = 2023$08, $T_0 = 2023$09, $T_0 = 2023$10, $T_0 = 2023$11, $T_0 = 2023$12 s, and $T_0 = 2023$13. That study argued that the short observed duration could arise from $T_0 = 2023$14, with jet breakout time nearly matching the engine activity time (Wang et al., 2024).

Later work explicitly challenged single-zone prompt models. A width-based study defined the spectral width in $T_0 = 2023$15 space as

$T_0 = 2023$16

with $T_0 = 2023$17, and found that $T_0 = 2023$18 increased steadily across 16 time bins, from $T_0 = 2023$19 in the earliest bin to $T_0 = 2023$20 by 4.36–4.92 s. Because the same work argued that both single-zone thermal fireball models and single-zone synchrotron cooling models predict decreasing $T_0 = 2023$21, it interpreted the observed broadening as evidence for multiple prompt-emission zones whose relative contributions change with time (Gupta et al., 12 May 2026).

The GeV component added a separate constraint. Fermi-LAT recorded a 72 GeV photon at about 32 s. One analysis found a time-integrated LAT photon index $T_0 = 2023$22 and a light-curve decay $T_0 = 2023$23, and argued that the 72 GeV photon exceeded the standard forward-shock synchrotron limit, estimating $T_0 = 2023$24 GeV for $T_0 = 2023$25 and $T_0 = 2023$26; it therefore invoked SSC and/or EIC for the highest-energy LAT photons (Ror et al., 2024). A later broadband keV–VHE analysis similarly identified a synchrotron-to-SSC turnover between hard X-rays and GeV energies, with the GBM afterglow representing synchrotron near its high-energy cutoff and the LAT emission during 25–250 s dominated by SSC. In that framework, the 72 GeV photon cannot be produced by standard forward-shock synchrotron and is naturally assigned to the SSC component (Mohnani et al., 24 Jul 2025).

6. Host galaxy, population context, and broader significance

The host galaxy spectrum contains emission lines [O II], [O III], H$T_0 = 2023$27, and in one analysis H$T_0 = 2023$28, together with absorption lines Mg II, Mg I, Ca II, and Ca I. The $T_0 = 2023$29 d spectrum yielded a line strength parameter $T_0 = 2023$30, slightly above average, with relatively strong Mg I indicating a low-ionization interstellar environment along the line of sight (Hussenot-Desenonges et al., 2023). CIGALE modeling of host photometry within the transient aperture gave a stellar mass $T_0 = 2023$31, star-formation rate over the last 10 Myr of $T_0 = 2023$32, and global attenuation $T_0 = 2023$33 mag, with the explicit caveat that these values may be biased low because only part of the galaxy was covered and UV constraints on the SFR were limited (Hussenot-Desenonges et al., 2023).

In the GRB–SN population, GRB 230812B is notable for combining high prompt energetics with a supernova that is photometrically ordinary by GRB-SN standards. Depending on analysis choices, published prompt-energy estimates span $T_0 = 2023$34 erg, $T_0 = 2023$35 erg, $T_0 = 2023$36 erg, and $T_0 = 2023$37 erg, reflecting different instruments, fluence measurements, energy bands, and cosmological assumptions (Hussenot-Desenonges et al., 2023, Srinivasaragavan et al., 2023, Ror et al., 2024, Wang et al., 2024). One population study nevertheless emphasized that, within the observed GRB–SN sample, SN 2023pel has brightness and nickel mass typical of GRB-SNe even though GRB 230812B has among the highest prompt energetics in that subset (Srinivasaragavan et al., 2023).

The same population analysis reported no significant correlation between supernova brightness and GRB prompt energy. Pearson tests gave $T_0 = 2023$38 ($T_0 = 2023$39) for $T_0 = 2023$40 versus $T_0 = 2023$41 among high-energy jets and $T_0 = 2023$42 ($T_0 = 2023$43) for the full GRB–SN sample; for $T_0 = 2023$44 versus $T_0 = 2023$45, the corresponding values were $T_0 = 2023$46 ($T_0 = 2023$47) and $T_0 = 2023$48 ($T_0 = 2023$49). The published conclusion was that the central engine powering the relativistic jet is not coupled to the supernova powering mechanism in GRB–SN systems (Srinivasaragavan et al., 2023).

GRB 230812B therefore occupies several boundary regions simultaneously: it lies near the canonical duration division but is securely a collapsar; it shows a bright prompt phase but a fainter afterglow than some benchmark low-redshift LAT bursts; it admits both ISM-based and wind-based afterglow interpretations in different analyses; and its prompt emission has been modeled as Band+BB, time-evolving synchrotron, and multi-zone broadening. What is not in dispute is the core observational picture: a nearby, extremely well observed GRB–SN event, viewed close to on-axis, accompanied by a rapidly evolving Type Ic-BL supernova, and valuable for testing how prompt gamma-ray emission, external-shock afterglow physics, and core-collapse supernovae are linked in relativistic stellar explosions (Hussenot-Desenonges et al., 2023, Ror et al., 2024).