J/ApJS/284/9 CNN catalog of solar flares (Farhang+, 2026)
A Convolutional Neural Network-derived Catalog of Solar Flares from Soft
X-Ray Observations.
Farhang N., Wheatland M.S., Melatos A.
<Astrophys. J. Suppl. Ser. 284, 9 (2026)>
=2026ApJS..284....9F 2026ApJS..284....9F (SIMBAD/NED BibCode)
ADC_Keywords: Sun
Keywords: Solar flares - Catalogs - Convolutional neural networks
Abstract:
A convolutional neural network (CNN) is used to construct a new
catalog for solar flares based on high-resolution (1 s cadence)
Geostationary Operational Environmental Satellites (GOES) soft X-ray
data. The CNN is trained to identify flare rise episodes. From 2018
January 1 to 2025 August 22, the algorithm detects 111,580 flare
candidates, compared with 14,612 events in the corresponding GOES
catalog. For each candidate, the probability of being a true positive
is quantified by Bayesian inference based on the peak flux, rise time,
and temporal coincidence with cataloged events where available. The
flare size and waiting-time distributions are studied and compared
with the GOES catalog. The CNN catalog shows a steeper power-law index
for raw peak fluxes (-2.59±0.02) than GOES (-2.25±0.04),
indicating the CNN's higher sensitivity to small events. After
background correction, the indices are -1.97±0.02 (CNN) and
-2.05±0.04 (GOES). The CNN catalog extends the power-law
distribution of flare peak fluxes by 1 order of magnitude at the
small-flux end compared with the GOES background-subtracted catalog. A
Bayesian block analysis of the waiting-time distributions from the
GOES and CNN catalogs indicates broad consistency with a piecewise
Poisson process. We find that previously reported correlations between
flare sizes and waiting times are significantly influenced by
obscuration, that is, undercounting weaker or overlapping flares
during periods of elevated flux. The new CNN catalog provides a
foundation for complete and consistent studies of solar flare
statistics.
Description:
A new solar flare catalog is constructed using a convolutional neural
network (CNN)-based detection framework. The method operates on
high-resolution (1s cadence) Geostationary Operational Environmental
Satellites (GOES) soft X-ray data and identifies flare rise episodes
(rather than the complete rise-decay profiles). By focusing on the
rise episode, the approach partially relaxes the slow-driving
(non-overlap) constraints inherent to many conventional flare
detection algorithms, thereby allowing consecutive and overlapping
events, particularly those initiating during the decay of a preceding
flare, to be recognized. For each candidate, the probability of being
a true positive is quantified by a Bayesian inference based on the
peak flux, rise time, and temporal coincidence with known cataloged
events where available.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
flarecat.dat 111 110177 CNN-based solar flare catalog
refdat.dat 48 7700 Reference catalog
--------------------------------------------------------------------------------
Byte-by-byte Description of file: flarecat.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 10 A10 "date" Date Event date (MM/DD/YYYY)
12- 27 A16 --- StartTime Flare start time (MM/DD/YYYY hh:mm)
29- 39 E11.9 W/m2 StartFlux Flux at flare start
41- 56 A16 --- PeakTime Peak time (MM/DD/YYYY hh:mm)
58- 69 E12.9 W/m2 PeakFlux Flux at flare peak
71- 74 I4 s RiseTime Rise time
76- 86 E11.9 W/m2 BgSubFlux Background-subtracted peak flux
88- 99 E12.10 --- TrueProb Probability of true positive
101-111 E11.9 --- FalseProb Probability of false positive
--------------------------------------------------------------------------------
Byte-by-byte Description of file: refdat.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 10 A10 "date" Date Event date (MM/DD/YYYY)
12- 19 A8 "h:m:s" StartTime Flare start time (hh:mm:ss)
21- 28 E8.3 W/m2 StartFlux Flux at flare start
30- 37 A8 "h:m:s" PeakTime Peak time (hh:mm:ss)
39- 49 E11.4 W/m2 PeakFlux Flux at flare peak
--------------------------------------------------------------------------------
Acknowledgements:
From Nastaran Farhang, nastaran.farhang(at)sydney.edu.au
The authors acknowledge the use of GOES SXR data provided by the NOAA.
This work was supported by an Australian Research Council Project (ARC
DP220102201).
License: CC-BY-4.0
(End) Patricia Vannier [CDS] 28-May-2026