0.2.t4n20

This is an enhanced version of 0.1.t4n20, released on July 11, 2021. It is based on more MKF parameters, 27 in total, including the KP index from the geomagnetic data. It is also a classification model that samples the MKF parameters every 4 seconds that classifies the background data into 20 basis spectra. The root-mean-squared performance in the background estimates is 1.7 counts/s vs 3.2 in the 3C50 model. If the 1% outlier background observations are discarded, the performance is 0.87 counts/s (vs 2.3 for the 3C50 model).

Version 0.2.t4n20 requires a geomagnetic data file. The latest file can be downloaded from https://heasarc.gsfc.nasa.gov/FTP/caldb/data/gen/pcf/geomag/kp_noaa.fits. The model was trained using the noaa data file, so it is suggested that it is the one used and not the Potsdam one.