SDR models

Good afternoon everyone

I am running a SDR model on INVEST 3.30 for watershed in the Amazon region in Peru. I have three question about my outputs.

First of all, Why in the outputs sed_export and sed_retention raster (output folder) there several gaps (no data)?. I have revised my DEM (and it is filled) and the other imputs looks fine. This gaps has influenced in the resuts shapefile?

Second, what is the influence of threshold flow accumulation if it were changed (1000 as a default to 10000)?, I showed for example that sed_export of a watershed changed between 1367.43 to 2289.4

Third, Why my results (in the shapefile) when is compared with a scenario actual and a scenario future (with a Dam), looks the sediment retent more and expor less? if I changed the soil for water (areas that will be flooded)

Thank you for you help.



  • RichRich Administrator, NatCap Staff
    Hi Anita, usually gaps in the result output are due to flow routing errors in the DEM due to hydrological pits.  Usually this is obvious if you look at the generated flow accumulation raster.  If that looks okay it may be an issue with other nodata gaps in the raster stack.  Could you take a look and/or post a screenshot here?

    The flow accumulation threshold determines how the model classifies a pixel as a stream.  You can also take a look at your flow accumulation raster to determine what you think is a good value for your region.  A larger value will result in shorter streams which influences the upstream/downstream ratio in the SDR calculation in very non-linear ways.
  • Hi Rich

    The flow direction raster is ok, but I think the problem is in the Ls tiff raster, lookup down my Sed Retention Index, has a hole and correlates with the LS raster (black part in the up picture). Thank you for you time...

    Imágenes integradas 1

    Imágenes integradas 2

    Imágenes integradas 3

  • RichRich Administrator, NatCap Staff
    Hi Anita, for some reason I'm unable to see your screenshots.  Is there anyway you could Dropbox me your entire datastack to  I'd be happy to take a look at the whole thing directly.
Sign In or Register to comment.