RFI toolkit in development on Green Bank Observatory computing system.
OFFLINE RFI INSTRUCTIONS:
In OfflineRFI/:
use detectRFI_VPM to calculate SK on an entire GUPPI/VPM raw file (coarse channelized, not accumulated) and replace data. The original datafile is preserved.
Supporting functions are found in RFI_detection.py (RFI detection methods) and RFI_support.py (misc support functions)
Examples:
Perform spectral kurtosis the entire pipeline on the guppi raw file 'guppi_58626_J1125+7819_0014.0001.raw' in the 'in_dir' directory, with m=512 in the SK equation and statistical noise data replacement:
$ python guppi_SK_fromraw.py -i guppi_58626_J1125+7819_0014.0001.raw -m 512 -r stats -rfi SKurtosis
The output SK, flags, and spectrogram npy filenames are derived from the input filename. Use -h for help on other optional arguments.
The raw data in each block is NOT written to npy files that would be saved in 'my_dir'. SK is run on 512 samples at a time, and Spectrogram is the result from averaging 512 spectra at once Flagged data is replaced with statistical noise generated by unflagged data in the same channel (or adjacent channels if there is no unflagged data) 3-sigma thresholding is the default for flagging RFI. Corresponds to a probability of false alarm (PFA) of 0.0013499.
For the first two input filename, you may put the full path to a directory (start with '/' to trigger this)
OfflineRFI/utils/ contains other misc python programs for random tasks OfflineRFI/dev/ contains programs that are incomplete / in development