I used the NDR model and the result is abnormal output. The raster file has a lot of null values. Where these nulls are located is not a body of water, but parts of farmland.
Hi @qiaoxiao1997 ,
Thanks for writing in to the forum.
In order for us to assist you, please first upload the logfile (ending in extension
.txt) from your model run to this thread here.
My nitrogen output raster distribution turned out to be similar to the stream distribution. But the distribution of this stream is very different from my actual river distribution. I want to know where is the problem.
Hi @qiaoxiao1997 ,
Thank you for uploading the logfile.
My first recommendation is to update to the latest version of InVEST and try re-running the model there to see if the results are any different. You can download version 3.10.2 from our website here. Several changes have been made to the model since the version you’re using.
As for the modeled stream network being “…very different from [your] actual river distribution”, this is not uncommon. This is largely controlled by the DEM and the threshold flow accumulation (TFA) value you enter. It’s possible that the DEM needs to be resampled to a finer resolution or that you need to source one that is already finer. Here are some sources for global elevation data. But, we always recommend trying different TFA values and comparing the results against a real perennial stream/river network before putting great confidence in other model results. This is an important step that deserves some more of your time to iterate on before proceeding. Of course, a higher TFA will create a coarse network and a lower one will create a relatively finer network. Please refer to step 5 in our “Working with the DEM” section of the User Guide. Does your study area happen to be particularly flat? Our water models do not model stream/river networks very precisely in very flat areas.
After updating InVEST to version 3.10.2, please spend some time playing with the DEM and TFA value to see if you can get a better match with the real stream/river network before moving on to modeling nutrient flows. If you’re still having trouble, you could try burning the stream network into the DEM.