Calculated LS factor seems too high in SDR model

Hello all

I have seen some past posts relating to this issue but I wanted to add my own experience that the LS estimation in the USLE calculation of the SDR model seems to be coming out too high and driving up the erosion estimates, even when using the value of 122 as the maximum L value when setting up the model.

I am currently using the model in Rwanda with the 30 m SRTM DEM and looking at the LS factor intermediate output. This is giving a mean LS of 24 for the country, which seems too high based on the literature and manual calculation of the LS factor (see below). In the hilly parts of the country it often exceeds 100.

I have been comparing my results to a published study of erosion rates in East Africa which used the same Desmet and Govers equation to calculate the LS factor, though using a coarser 90 m SRTM DEM. Their estimated mean LS factor for Rwanda was much lower at around 5.

I also tried to calculate the LS factor manually using the same 30 m SRTM DEM with QGIS and the Desment and Govers method. Again, I am getting much lower values than what the InVEST LS output is showing, with an average LS value of around 4. This is 6X lower than the average from the InVEST LS output and much closer to the published estimate I am comparing to.

Perhaps this has already been done before after this concern was raised, but I wonder if it would be possible to look again at the LS calculation being done by the model, to see if there is anything strange that could be causing values to come out too high?

Thanks for your consideration of this issue.


@lukezw ,
Please see this thread for a summary of the known differences betwen InVEST’s LS factor and other implementations. Some modifications will likely be released in version 3.14.

Hello @lukezw ,

To add on to what @esoth mentioned, a summary of the changes that we are making to InVEST in an upcoming release can be found in this document:

Furthermore, if you would like to try a preview of the latest changes, feel free to take a look at this development build:



Thanks James and Emily for the helpful information and I’m glad to hear it is something you have been working on.

I think I might give the experimental build a try and see how the results come out. Always good to see how the models are continually evolving and improving!

Just a follow up, I re-ran using the development build with identical parameters and data inputs. This resulted in a marked sixfold reduction in my total erosion estimate for Rwanda! I do feel this result is a lot more in line with reality and what other studies have estimated.

1 Like

Fantastic, thanks for the update and I’m glad the updated version is a better reflection of reality!

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.