This forum is shutting down! Please post new discussions at

feeling misunderstand: Slope Length Factor in SDR intermediate results

Dear All,

I have been running the SDR Model and get good results, but this time I am challenged with Governmental agency the credibility of InVEST model to estimate soil loss. The two projects were running in parallel using almost the same input but both models (GIS) are predicting different the soil loss (here I am defending InVEST). Thus we are aligning inputs and see how final results are generated in each model ( Both model quoted LS calculated based on Desmet 1996). As InVEST generated intermediate results, I am looking at the Slope length raster from intermediate results. I found a raster with LS, which I am assuming to be the LS factor. But when compared with the one they calculated in ArcGIS model, It seems to be too different and InVEST seems to over-estimate this LS factor. I am doubting my knowledge on intermediate results of SDR whether this raster (LS) is representing real LS in RUSLE equation of this LS (raster) is only Slope length not combined with steepness. 

I have tried to calculated LS using their approach (GIS) and I have found almost similar results (LS) with their LS so, your assistance is highly needed to be able to defend InVEST model algorithm behind the SDR model.



  • RichRich Administrator, NatCap Staff
    Hi Evariste,

    Set an m value via a lookup table as described in Oliveira, A.H., Silva, M.A. da, Silva, M.L.N., Curi, N., Neto, G.K.,
    Freitas, D.A.F. de, 2013. Development of Topographic Factor Modeling for
    Application in Soil Erosion Models, in: Intechopen (Ed.), Soil
    Processes and Current Trends in Quality Assessment. p. 28.

    And threshold the l factor to no more than 333m as suggested in Renard, K., Foster, G., Weesies, G., McCool, D., Yoder, D., 1997.
    Predicting Soil Erosion by Water: A Guide to Conservation Planning with
    the revised soil loss equation.

    I'm not familiar with the ArcGIS implementation, but it's possible that it may differ in some ways to our specific modifications. I'm curious what you think! And especially if you think we're doing something wrong.

  • RichRich Administrator, NatCap Staff
    Oh one other note, we use multiple flow direction to calculate the flow accumulation raster. It's possible that ArcGIS is using D8 which could have significant differences between the implementations. We believe a MFD implementation to be more physically correct than D8.
  • Dear Rich, 

    Thanks indeed for your guidance to clarify my mind on SDR model. 
    I have checked the SDR code and papers  you shared and now  confusion is no longer there. 
    I have also calculated LS using ArcGIS model and rerun the model using the same inputs as in InVEST model. Interesting my results are much similar  only ArcGIS model has increased 6% of InVEST result at National scale which I hope to be linked to your last comment of D8 in ArcGIS. 

    I am happy for your contribution 
Sign In or Register to comment.