TODO.txt

Rerun nbs except grandmesa for new repo structure
Change backscatter plots to dB
Sentinel-2, use box instead of point for searching scenes so we have full images at each timestamp
For ripening function, add option for 2-3dB threshold drop
.melt_onset_elevation_line() levation-meltday curve for each year and each mountain
.aggregate_by_doy()
.plot_gif()
Add Mt. Shasta, others? Glacier Peak, etc
predict runoff with NDSI (correlate melt with landsat and calculate snowmelt timing for historical landsat catalog)
break out strength of MLS corrs by year
stream gauge data (comment from david on gage and validation sentence of my abstract: Worth trying gage data.  USGS has an API to access, but you can just manually identify and download csv.  Biggest issue is going to be dams and other control - need to select watersheds carefully. Nooksack may be good place to start. Honestly, met station data might be the most useful in situ data to consider, as you should be able to tell when melting starts from air temps.)
SNOTEL temperatures
given geojson,give correct MGRS tile
clean up env
trim down demonstration notebook
make sure demonstration notebooks are calculating runoff map correctly--which time domain to use