Sediment Delivery Ratio - very high values


I ran the Sediment Delivery Ratio model with InVEST 3.2.0
for a study area in the Ethiopian highlands obtaining very high values. For
example the usle ranges between 0 and almost 500000 with a mean of 3500. For
comparison I calculated the usle with GRASS GIS using the same data. In this
case the results are lower, between 0.6 and 18000 with a mean of 240. With
further investigation I realised that maps of length slope calculated by InVEST
and GRASS (using r.watershed module
are very different. In GRASS GIS the minimum slope length is 0.03, maximum = 24
and mean =  0.4, while in Invest minimum
= 0.05, maximum = 1174 and mean = 8.5.

I am wondering if this is expected behaviour of InVest ? In
this case how this can be explained?  Or
is there something that I am missing?

Thanks in advance for your help.




  • PerrinePerrine Moderator, NatCap Staff
    Hey Stefania, 

    Thanks for sharing this. There are indeed large uncertainties around the LS factor. The expression we use in InVEST is derived from Desmet and Govers(1996), and we observed a high sensitivity to the DEM resolution. Which resolution are you working at?

    I'm not sure which algorithm they use in GRASS, but if you have reasons to believe it's more accurate, you can substitute these values in the InVEST model. (Because the USLE is a simple product, you can do some raster calculation to divide by the InVEST LS factor and multiply by the GRASS LS factor. The SDR factor won't be affected.

    Please feel free to share any insight you'd gain from reading the literature on this issue.



    Desmet, P.J.J., Govers, G., 1996. A GIS procedure for automatically calculating the USLE LS factor on topographically complex landscape units. J. Soi 51, 427–433.
  • Hi Perrine,
    thanks for your reply.

    I'm working with a DEM of 92 m.
    Unfortunately I couldn't find the work used by GRASS GIS to calculate the LS factor (Weltz et al., 1987).


    Ref: Weltz M. A., K. G. Renard, J. R. Simanton (1987). Revised Universal Soil
    Loss Equation for Western Rangelands
    , U.S.A./Mexico Symposium of
    Strategies for Classification and Management of Native Vegetation for
    Food Production In Arid Zones

  • Hi Stefania,

    Coincidentally I'm also working on an area in the Ethiopian highlands (north of Debre Birhan) and finding impossibly high values (mean of 274 t/ha) although yours are much higher (mean of 4,135 t/ha if your cell size was 92 m) for the usle. 

    Some things: 

    - Why use a 92-m DEM? The 30-m SRTM DEM should be available.
    - A suggestion to check your units for K-factor. It should range from 0.001 to 0.06 (max). This was my prior problem.

  • Hello,

    sorry for the late reply.

    I have just started working with the 30m SRTM DEM. When I started running the
    models it was not available.

    My K-factor is far higher than yours. It ranges approximately from 0.3 to 0.4.

    How did you calculate it?

    I used the equations in Zhang et al.,


    REF: Zhang et al., 2008. Soil erodibility and its estimation for agricultural
    soils in China. Journal of Arid Environments, 72: 1002-1011.

  • Hi Stefania,

    Did you use Eq. 6 in Zhang, the one with geometric mean diameter? 

    As FYI, I used the nomograph method found in the Auerswald paper (see bottom). However, if you use this, make sure to also check out the corrigendum (see bottom), since there were a few typos in the original.

    0.001 to 0.06 was not my range. It was just the range of possible K-values you could have using the SI units. See Perrine's last comment here:

    My actual K-factor for my site ranged from 0.018 to 0.027. I'm guessing your 0.3-0.4 is fine, but it's probably in (t h)/(ha N), and needs to be divided by 10 to achieve (t h)/(MJ mm) which are needed for InVEST.

    PS: Not sure what I fixed, but my SDR results are now more sensible: 18 ton/ha average. 

    Auerswald, K., Fiener, P., Martin, W., & Elhaus, D. (2014). Use and misuse of the K factor equation in soil erosion modeling: An alternative equation for determining USLE nomograph soil erodibility values. Catena118, 220-225.

    Auerswald, K., Fiener, P., Martin, W., & Elhaus, D. (2015). Corrigendum to “Use and misuse of the K factor equation in soil erosion modeling”[Catena 118 (2014) 220–225]. Catena.

  • Hi James,
    thanks for your reply and for all the advices! 

    Sorry I forgot to indicate the equation that I used of Zhang et al. (2008). I didn't use the Eq. 6 but I used the Eq. 5, the one with silt, clay, sand and carbon %. According to this equation my values should be in SI units.

    I'll read your references soon, thanks.

  • Ok, yea I see. It's not so clear in the Zhang paper what the units are for Eq. 5. But anyway, to me, both (t h)/(ha N)  AND  (t h)/(MJ mm)  are both using SI units. Yet, they are different by a factor of 10. Just be careful on that one. 
  • PerrinePerrine Moderator, NatCap Staff
    Hi both, 

    Just checked the Zhang's paper and it is *possible* that the K values in Eq. 5 are in US customary units (but I agree, it's very unclear in the paper). Eq 5 comes just after the Wischmeier and Schmitt's reference where values *are* in US customary units. I couldn't find the original EPIC paper reference online so it's difficult to know for sure.

    If they are in US customary units, the conversion needs to be done according to the attached .pdf (from Renard et al. 1996, USDA Handbook 703).

    One way to check would be to compare your erodibility values with other studies in the region.

Sign In or Register to comment.