This dataset is a .csv (space separated) file listing all Starlink satellite detections in Grigg et al. 2025 (https://www.aanda.org/articles/aa/full_html/2025/07/aa54787-25/aa54787-25.html) using the Engineering Development Array 2. Each column is as follows: 0 - datetime as a timestamp. Use str(datetime.utcfromtimestamp(dt_float)) to read it into a string1 - image index - image number in the dataset2 - central frequency (Hz) of the coarse channel3 - norad id4 - fine channel frequency index. 0-30 are the fine channels and 31 is from the stacked image. the 5 - polarisation index. 0 - XX, 1 - YY6 - flux density (Jy/beam). if frequency index (col 4) is 0-30 then it's the pixel value, if frequency index is 31 then it's the fitted flux density7 - tle range (km)8 - right ascension from the fit (deg)9 - declination from the fit (deg)10 - right ascension from the tle predicted position (deg)11 - declination from the tle predicted position (deg)12 - azimuth from the fit (deg)13 - elevation from the fit (deg)14 - x pixel location of the peak of the fit15 - y pixel location of the peak of the fit16 - x pixel location of the tle predicted position17 - y pixel location of the tle predicted position18 - euclidean distance (in pixels) between the fitted and tle predicted locations19 - uncertainty in the x pixel location of the peak of the fit (in pixels) - calculated by doing the sqrt of the diagonal of the covariance of the fit20 - uncertainty in the y pixel location of the peak of the fit (in pixels) - calculated by doing the sqrt of the diagonal of the covariance of the fit Please note as mentioned above that a fine channel frequency index of 31 is the frequency stacked image (0.90 MHz bandwidth), and that the rest are from fine channel images (28.9 kHz bandwidth). Some columns like 12 and 13 don't really apply to the fine channels as a fit is only attempted in the stacked image, so the value is just copied from the stacked image at the same timestep. For more information, please refer to the paper (link will be provided when published), or feel free to drop me an email at dylan.grigg@icrar.org