Occurrence rate of stellar Type II radio bursts from a 100 star-year search for coronal mass ejections

TL;DR

We address whether Type II-like radio bursts occur on stars as signatures of CMEs by performing the largest unbiased search within 100 pc using LoTSS data, totaling ~107.21 stellar-years. Employing a drift-corrected, inverse-variance weighted analysis on Stokes V spectra and a Parker-wind based drift model, we detect two drifting bursts from M dwarfs with rates $a_1 = -0.801^{+0.003}_{-0.003}$ and $a_2 = -0.060^{+0.002}_{-0.002}$, plus 19 non-drifting bursts. A joint analysis yields a cumulative luminosity distribution with $\alpha = -0.732$ and $L_0 = 10^{3.528}$ erg s$^{-1}$ Hz$^{-1}$, and a Poisson-based rate of about 1 burst per year for $E > 6.8\times 10^{13}$ erg s$^{-1}$ Hz$^{-1}$, with solar-type bursts showing consistent slopes ($\alpha \,\approx\, -0.81$). The findings suggest the paucity of detections is primarily sensitivity-driven rather than intrinsic rarity and inform expectations for next-generation surveys like SKA1-LOW to probe CME-associated radio bursts across a broader stellar sample.

Abstract

Coronal mass ejections (CMEs) are major drivers of space weather in the Solar System, but their occurrence rate on other stars is unknown. A characteristic (deca-)metric radio burst with a time-frequency drift, known as a Type II radio burst, is a key observational signature of CMEs. We searched a total of 107 years of stellar data using time-frequency spectra that targeted all known stars within 100 parsecs in the LOFAR Two Metre Sky Survey (LoTSS) up to May 2023. This resulted in the largest unbiased search for circularly polarised stellar Type II metric radio bursts to date, with a typical 3$σ$ sensitivity of 2.5 mJy for an integration time of 1 minute. We detected two drifting stellar radio bursts: the published 2-minute burst from the M dwarf StKM 1-1262 and a new 13-minute burst from the M dwarf LP 215-56. The new burst is characterised by a drift rate of $-0.060^{+0.002}_{-0.002}$ MHz s$^{-1}$, an average Stokes V flux density of $-4.5^{+1.4}_{-1.3}$ mJy, and a temporal duration of $63^{+31}_{-11}$ seconds. We constrained the occurrence rate of drifting stellar bursts by calculating Poisson upper and lower limits based on the two drifting bursts. We also fitted a cumulative burst luminosity distribution to the data using the burst detections and the non-detections; this yielded a power law index ($α$) of $-0.7^{+0.9}_{-0.6}$ and a normalisation point (N) of one burst per year with $E>6.8\times10^{13}$ erg s$^{-1}$ Hz$^{-1}$. We find an agreement between this and the cumulative luminosity distribution of decametric SOHO/LASCO solar Type II data ($α= -0.81 \pm 0.06 \pm 0.02$), which suggests that the current scarcity of detected stellar Type II bursts is likely due to limited sensitivity rather than to the intrinsic rarity of these events. Additionally, we identify 19 circularly polarised stellar radio bursts without a time-frequency drift.

Figures (10)

• Figure 1: Colour-magnitude diagram of Gaia sources observed with LoTSS, applying the quality filters described in Sect. \ref{['sec:obs']}. The sources are categorised into two clusters based on their proximity in the colour-magnitude space.
• Figure 2: Panel a: Mock Stokes V dynamic spectrum of a radio burst due to plasma emission modelled using the Parker Solar wind model with solar parameters, assuming a shock speed of 1000 km s$^{-1}$ and a S/N of 30. Panel b: Frequency-integrated burst brightness as a function of the drift delay and time, modelled using the Parker Solar wind model with solar parameters, assuming a shock speed of 1000 km s$^{-1}$. The defining 'bow-tie'-shaped feature of a broadband swept astrophysical signal is clearly visible. The red cross indicates the brightest pixel and its corresponding burst drift.
• Figure 3: The 19 stellar radio bursts detected without drifting features. Each thumbnail shows a dynamic spectrum ($\Delta\nu$ = 0.4 MHz, $\Delta$t = the boxcar width). The interferometric images are shown as insets in the top left. The time series of the inverse variance weighted frequency-integrated spectra are shown at the top of each panel, with vertical dotted lines indicating the start and end times of the burst. The time-integrated spectra (shown to the right of each panel) are integrated over the time range indicated by the vertical dotted blue lines in the time series. Denoted in the top right of each thumbnail is the burst ID, which corresponds to Table \ref{['tab:hosttable']}. White areas in the dynamic spectra represent areas masked due to the presence of RFI, and the white ovals in the interferometric image represent the point spread function. For visual purposes, the colour maps of the dynamic spectra range from -20 to 20 mJy. Bursts B02, B04, B10, B11, B12, B14, B16, B17, and B20 are also presented in Tassesubmitted.
• Figure 4: Panel a: Dynamic spectrum of B20 normalised by the blank-sky variances. The inset in the bottom left displays an interferometric image indicating that the emission originates from a point source. The horizontal white bands are frequency channels masked due to the presence of RFI. Panel b: Dynamic spectrum of B20, with the highest likelihood drift rate (i.e. the overall drift rate) shown as a dashed red line. Panel c: Frequency-integrated burst brightness as a function of the modelled drift delay. The intersection of the dotted red lines indicates the location of the highest S/N at $\sim$1 minute; the horizontal line represents zero drift and corresponds to the original time series.
• Figure 5: Panel a: Dynamic spectrum of B21 convolved with a kernel designed to follow the burst’s drift pattern. The drift of the burst is overplotted as a dotted red line. Panel b: Raw dynamic spectrum of B21. The horizontal white bands are frequency channels masked due to the presence of RFI. The inset in the bottom left shows an interferometric image reconstructed along the dotted line in panel a, indicating that the emission originates from a point source. Panel c: Frequency-integrated burst brightness as a function of the modelled drift delay. The dotted cross indicates the location of the highest S/N at a drift of 13 minutes, where the burst aligns vertically and thus appears offset from the horizontal zero drift line, which corresponds to the original time series.