Hi @Tookie -
While I’m also not sure about the difference between sed_retention.tif and sed_retention_index.tif (and would love clarification as well, especially since I’d think that an “index” would have values between, say, 0 and 1, but sed_retention_index.tif does not), I’ll say a few things about some of your other points.
Using rkls.tif minus usle.tif only considers the erosion produced on a particular pixel, it does not take into account retention of sediment coming to that pixel from upslope. So it depends what you’re trying to model, soil loss by pixel, or amount of soil loss reaching streams. Using the SDR-related results will also take into account what happens upslope/downslope of each pixel, and help evaluate soil loss to streams.
The only thing I’ve ever (rarely) used sed_retention.tif for is looking at the patterns (not quantity) of how soil loss would be affected if the natural vegetation was removed and turned to bare ground. Really, it’s just creating a quick scenario for us by comparing our input landscape with bare ground.
Usually, we evaluate sediment retention by comparing a “current” or “baseline” with some user-created scenario of change, and use the difference in sed_export.tif to show how the scenario changes the amount of erosion reaching streams, and where those changes occur. If you don’t have a scenario, and you’re looking at the watershed or sub-watershed scale, you can look at the ratio between sed_export and USLE in the watersheds output shapefile to see what percentage of erosion is being retained in the watershed.
To your last point, when the C factor is larger, it means that more erosion is being created, which also means that there is more erosion to retain. So it’s not surprising that sed_retention would also increase in that case.
Once all of this gets worked out, I’ll update the User Guide, hopefully clarifying retention a bit.