Open jaycedowell opened 10 months ago
After talking with @dentalfloss1 I realized that we are only saving two channels for the archival images from LWA-NA. It looks like that is from a combination of me not fully adjusting orville_imager.py
for the bandwidth from NDP and the imager being told to only save 30 to 50 MHz. That first part means that Orville is dividing the band into six 12.25 MHz chunks. That second part means that you only get the two channels centered at ~33.7 and ~45.9 MHz.
For sensitivity reasons, we might not want to make the channels much narrower than 12.25 MHz. And if we want to have a decent amount of spectral information, we may want to expand the range beyond 30 to 50 MHz.
Things that work out by averaging over N channels:
1024 leads to 3 @ 24.5 MHz
If you look at what satisfies >30 MHz and <50 MHz you get:
If you want to save >20 MHz and <50 MHz you get:
Maybe this argues for N=512 (what we currently have) or N=384 but expanding what is saved to 20-50 MHz? Either way I think adding a command line flag is the way to go.
From talking with Ken today it sounds like what we should do is double the pixel size (assume a station size of 50 m and set minimum image size of 64 pixels) and try to match what channels are at Sevilleta (3.3 MHz wide, first channel typically centered on 29.5 MHz, last channel typically centered on 46.3 MHz).
928696a95bc83104627d80a9a20aa445c5712ad7 changes the image size/resolution 83655639014bfdcc6cc85016afd1fef6037f8480 and 1c2e600ff3483cd7051b476a974d7e91156bb9af try to match what Orville typically provides at LWA-SV
For archival images we are currently averaging down to 3.3 MHz wide channels. Why? Because ADP has six servers that each process 3.3 MHz. With LWA-NA we don't have that kind of division again (it's only two ~36 MHz wide bands) and the 3.3 MHz doesn't make much sense.
This plus changes for #15 mean that we may want to be more flexible and have the archival channel width be configurable.