Currently calculate_water does a pretty good job of showing where water can be seen, whether through clear skies or thin cloud. The idea with clearsky_water is that it also brings in the swir2 band to refine the results. Unfortunately, it currently does a completely awful job of doing that, because the logistic function's threshold value is unchanged from what was being used for the boolean mask.
We need to experiment with the values in here in a variety of scenarios and see the results.
Currently calculate_water does a pretty good job of showing where water can be seen, whether through clear skies or thin cloud. The idea with clearsky_water is that it also brings in the swir2 band to refine the results. Unfortunately, it currently does a completely awful job of doing that, because the logistic function's threshold value is unchanged from what was being used for the boolean mask.
We need to experiment with the values in here in a variety of scenarios and see the results.