This release is associated with the paper:
Mason, J. P., Charles N. Arge, Raphael Attie, Barbara J. Thompson, Thomas N. Woods, "The SDO/EVE Solar Irradiance Coronal Dimming Index Catalog. I. Methods and Algorithms", 2019, The Astrophysical Journal Supplemental (probably), in prep
Beyond the small improvements, it includes a big resolution to an error: the csv file format of version 1 apparently cannot support the number of columns present in JEDI, or at least the to_csv() function of pandas cannot. So JEDI v2.0.0 now uses HDF5 format instead. It is still possible to load the dataset easily with pandas:
jedi = pd.read_hdf('jedi_v2.h5', key='jedi')
The issue with the old csv files was that after a few thousand columns, they became out of sync so that the values under a column heading did not actually correspond to what that column was supposed to be.
This version of the catalog also includes significantly more vetting. You can see most of that in this repository under explore/explore_jedi_v2_spot_checks.ipynb. It includes some comparison with prior work and a bunch of statistics, much of which is also covered in the paper.
Here are a few more changes:
- Now includes maximum dimming depth in addition to what was provided before, first identified depth (whose column names are now
<wavelength in nm> Depth Max [%]) - Now includes flare latitude, longitude, and those values converted to a position angle for comparison with CME position angle
- Resolves an issue where flare times were being erroneously reported (maximum offset of < 1 second)
- Pre-flare irradiances are now pre-computed and saved to disk after identifying which clusters of flares are independent in time — this independence of events now allows
determine_preflare_irradianceto be run in parallel — Thanks Raphael Attie (@WaaallEEE)!