Sediment retention - model output values too high

kimfalinski

Hello all,

I've been working on trying to get my units straight for the sediment retention model for some time. I am working in watersheds in West Maui, which have extremely steep sloping hillsides and a very large ridge to reef rainfall gradient. My watersheds are small (300 - 2000 ha).

Here are some details about my input files. I have been working in 2.5.4 x64 Standalone, but plan on upgrading to 2.5.6 soon.

DEM - in meters
Erosivity - in MJ*mm/ha/yr/h --> I converted this from the NRCS contour maps (by multiplying by 17.3), and my values range from 1000~7800.
Erodibility - between 0 and 1
Biophysical - Using 0 to 1 for usle_c (eg. 0.002); 0-100 for sedret_eff (eg. 10)

I am not yet doing valuation, so still have the original dummy values there.

Just looking at the RUSLE2 outputs, I expect to have about 20 tons/ha/yr, but my values for everything are *very* high - about 1000 for usle_mn, and then 50000 (5e4) for sed_export (or about 600 tons/ha/yr).

I am off by two orders of magnitude.

Does anything look off to anyone? I really can't see my error.

Thanks,
Kim

kimfalinski

Ah! I think I might have found my order of magnitude error?

My K values are in the English system. To convert to the metric, you need to multiply by 0.1317. *Or*, don't convert the R from the English system to begin with (the x17.02 mentioned in the Handbook).

This is a convenient conversion sheet:
http://www.techtransfer.osmre.gov/nttmainsite/Library/hbmanual/rusle/ah703apa.pdf

Rich

Hi Kim, the standalones for sediment and nutrient are still in beta and we identified some issues recently that overestimate the amount of sediment retention, which is the opposite of what you're experiencing. I suspect it otherwise may be a calibration issue that a hydrologist would need to comment on.

In the meanwhile when you upgrade from 2.5.4, you can try our cutting edge nightly build if you're brave: http://ncp-dev.stanford.edu/~dataportal/nightly-build/InVEST_dev79_2_5_6 [1029bd49a77a]_x86_Setup.exe That has a fixed version of our flow algorithm and addresses those overestimation of retention issues.

Perrine

Hi Kim,
You're right, both the K and R values should be in SI units (so K is in roughly in the [0-0.1] range).