An optical transient candidate of $\lesssim$ 2-second duration captured by wide-field video observations

TL;DR

The study targets optical transients on timescales of seconds or shorter, a relatively unexplored regime, by performing a 1 fps wide-field video survey of the Earth’s shadow with Tomo-e Gozen. It develops a two-stage transient-detection pipeline with stack-based references and per-frame PSF modeling to identify short-lived events, discovering one candidate, TMG20200322, with a duration of about $2$ s and an elongated PSF in the second frame. The authors find that conventional explanations such as atmospheric seeing, NEA impact flashes, or head-on meteors cannot fully account for the event and provide a sky-projected rate $R_{\mathrm{trans}}$ along with a stringent upper limit on shorter second-timescale transients; they also explore FRB-related implications via a fluence ratio $\eta(\nu_{c})$ that aligns with magnetar giant-flare models. The work demonstrates that Earth’s shadow monitoring with wide-field, high-temporal-resolution instruments is a promising path to uncover a new population of ultra-fast optical transients and highlights the potential for complementary science with upcoming Rubin/LSST data.

Abstract

Recent time-domain surveys have revealed rapid transients that evolve on timescales of $\lesssim 10$ days, expanding the transient population into the short-duration regime. The transient search on even shorter timescales, particularly those lasting only seconds or less, remains a largely unexplored frontier. Very short-duration optical transients could serve as potential counterparts to millisecond-duration fast radio bursts (FRBs), providing clues to their origins. However, the optical search for transients on such short timescales has been limited primarily by instrumental constraints. Here we report the discovery of an optical transient candidate (TMG20200322) with a duration of $\lesssim 2$~s by wide-field video observations in the direction of the Earth's shadow. TMG20200322 was detected in just two consecutive images of 1-second exposure time, with its shape becoming elongated in the second frame. PSF shape variability analysis of field stars reveals that such an elongated PSF cannot be explained by atmospheric fluctuations. We investigate the potential origins of TMG20200322 in two scenarios: meteoroid impact flashes on near-Earth asteroids (NEAs) and head-on meteors in the Earth's atmosphere. None of the scenarios provides a satisfactory explanation for this transient. We derive a sky-projected rate of the TMG20200322 event of $R_{\mathrm{trans}} = (3.4 \times 10^{-2})^{+0.13}_{-0.028}$~deg$^{-2}$~day$^{-1}$ and an upper limit on second-timescale transients with durations of $1~\mathrm{s} \leq τ\lesssim 15~\mathrm{s}$ of $R_{\mathrm{trans}} \lesssim 0.10$~deg$^{-2}$~day$^{-1}$ for the non-detection case. We highlight that continuous monitoring observations in the direction of the Earth's shadow could be a key strategy to unveil a new population of optical transients on timescales of seconds or less.

Figures (5)

• Figure 1: Survey footprint of our 1 fps video observations targeting Earth’s shadow regions at GEO. Each pointing area is shown as a circle with a diameter of $\phi = 9 \deg$, corresponding to the FoV of Tomo-e Gozen. The cumulative observation time in hours is color-coded. The blue solid curve denotes the Galactic plane, while the red dotted curve marks the Ecliptic plane. Alt text: All-sky map in equatorial coordinates showing the coverage of our survey.
• Figure 2: Histogram of the limiting magnitudes for our 1 fps video observations. The limiting magnitude is derived from each data cube consisting of 120 frames, using CMOS sensor ID 133, which has median sensitivity among the 84 sensors. The red vertical line indicates the median limiting magnitude ($\rm S/N = 5$), which is 17.48 mag. The black curve represents the cumulative distribution of the histogram, corresponding to the right vertical axis. For the transient search analysis described below, we set a limiting magnitude cut at $m_{\rm lim} = 16.0$ mag and exclude frames with shallower limiting magnitudes. The binning interval is 0.2 mag. Alt text: A graph of histograms with a reference line.
• Figure 3: Flowchart of our transient detection pipeline. The input data cube is processed through two analytical flows: the stacking process (core functions are make_stackimg, make_stacksegmap and make_psfmodel shown as blue in the brackets) and per-frame processing (search_allframe). Objects detected in individual frames are compared against a binary image called a segmentation map (segmap), which contains positional information of detected objects in the stacked image. Unmasked objects are those that are detected only in some frames but are not detected in the stacked image. These unmasked objects can be transient candidates, which are filtered based on their PSF profile and ellipticity parameters to remove false positives (e.g., cosmic rays) and meteors. In the right-hand figure, the schematic diagram illustrates the classification method for unmasked objects (search_transcand process). Unmasked objects located in the blue regions are classified as false positives, whereas those appearing as red circles along the stellar sequence are identified as transient candidates.
• Figure 4: A schematic picture of how to assign masked/unmasked objects with a segmap. Each object detected in each frame (marked as red and blue circles) is assigned segmap IDs by referring to the segmap. Finally two output catalogs of masked objects and unmasked objects including transient candidates are created. We then search for transients from the unmasked objects. Alt text: Diagram showing how objects detected in time-series frames are compared with a segmentation map and sorted into masked and unmasked object lists.
• Figure 14: Comparisons of the fluence ratio $\eta$ derived from several transient searches which are summarized by Chen2020 (Wevers2018Andreoni2020Richmond2020) as a function of the temporal resolution of each study. We include results from the simultaneous observations of FRBs with optical and radio bands, and model predictions of FRB optical emissions. The error bar of the point (detection case) indicates the $3\sigma$ of the power law index of the radio fulence CDF model James2019. Alt text: Fluence ratio vs temporal resolution with study markers, upper-limit triangles, a red star for this work, and model labels.