SDR inputs and interpretation of outputs

Dear NatCap team,

I am running the SDR model for two scenarios so far, current
conditions and green economy. The dataset I am changing is the Land Use raster.
I would like to address you the following questions:

About the inputs:

1. If I understood well the definition for the “threshold
flow accumulation”: “The default is 1 over the pixel area (in km2), i.e. ~1000
for 30m resolution”, when my DEM has 250m resolution, my threshold flow
accumulation is 16.
Could you please
confirm this?

About  the outputs:

1.       I am getting the sediment export and
sediment retention areas somehow overlapping, but at a pretty much different
scale, i.e. the sediment retention capacity is much bigger than the sediment export,
so presenting the results as a ratio [sediment export/sediment retention]
brings at least a more meaningful picture. Is this similar distribution
of export and retention areas normal/acceptable?

2.       When running the model with the green
economy scenario, the results I get present relatively “negligible” variations
in the subwatersheds where land use changed with respect to the baseline
scenario. What can explain this?


Sedim. retention (ton)

Sedim. export (ton)

USLE (ton)













Variation (%)




My river
basin area is 1.3 million km2

3.       Comparing both scenarios, I get a
sediment retention increase equal to the decrease in sediment export with the
adoption of the green economy scenario. Are the results supposed
to be like that?

4.       With regards to the results given by
the model, the assumptions done by the model can sufficiently explain a
great mismatch between modeled and observed results?

 Thank you so much for your help!!

Best wishes,



  • swolnyswolny Member, NatCap Staff
    Hi Elena -

    Regarding inputs: You should select the Threshold Flow Accumulation value so that the resulting "stream.tif" layer shows streams that represent your actual known stream network as closely as possible. This will be different in each place - the guidance in the User Guide is just to help you get started. We strongly advise trying to find a river map for your area, which could be digital, or even a paper map, anything that you feel is accurate and you can use to visually compare. Run the model with different TFA values until the stream.tif result looks close. You can do this process more quickly by using the InVEST RouteDEM tool, which will create multiple stream layers for you, based on a range of TFAs, to make it easier/quicker to find the right value.

    Regarding sediment retention/export values. First, it's important to note that the sediment retention result is not intended to be used as an absolute value (as tons/year), it is really an index of where current vegetation is providing sediment retention ability. It does not include information about sediment being routed down the flow path, and does not represent the actual amount of sediment that is retained by the landscape. It only says "if we removed this vegetation, and turned it all to bare ground, how much erosion would there be from each pixel?" Sediment export is, however, the result of routing sediment down the landscape, and does represent an absolute value (if you calibrate) of erosion that makes it to the stream. That all said, it is common to have much higher values of retention than export.

    And about the "green economy" scenario. I'd have to see the different LULC maps to start understanding why the difference is small. But I've seen these kind of surprisingly low results before, which is often the result of where and how much the different LULC types change between scenarios. If, for example, agriculture is currently done right up to the edge of streams, but in the green economy scenario that changes so that riparian buffers are now used, that is likely to have a larger impact on the results. But if agriculture is currently done right up to the streams, and in the scenario agriculture expands in the area far away from streams, that is likely to have a much smaller impact.

    Also, sometimes you think you're modeling a large change in land cover, but really it only translates into a small number of pixels being changed in the LULC map, so check for that. And finally, look at your USLE C values and how they differ in the places where LULC changes between current and scenario - if the C values have only a very small difference, you will get a very small difference in results.

    ~ Stacie
  • RGriffRGriff Member
    That first part about flow accumulation was really helpful and timely for something I am working on too - thanks Stacie! 
  • elenaelena Member
    Hi Stacie!

    Thank you so much for your reply and helpful explanations!

    It has indeed been a relief to see how the stream.tif matches the river network. Thanks a lot for this remark!

    And thanks a lot also for the other two comments! They indeed help to look at and understand the results

    Best wishes!
Sign In or Register to comment.