NDR result error?

Hi all
I am running 3.4.2 version of the NDR.The NDR model is running to completion for me,but when i check the model results,i find a problem of stream.tif,the lake in the layer is fault.According to user guide,0s representing land pixels,and 1s representing stream pixels,but the lake in stream.tif is not like the user guide.The picture of the fault is below
In addition,i use the 30m DEM for the NDR model,which coordinate system is WGS_1984_World_Mercator,the DEM of lake is 1045 to 1083.

No more questions for now, Can't Wait to hear from You all !
Thanks!image

Comments

  • RichRich Administrator, NatCap Staff
    Hi, thanks for the screenshot, that's very helpful. Offhand, that stream layer is what I would expect if routing were occurring across a large flat surface... such as a lake. So long as your landcover map classifies this as a lake and has appropriate biophysical parameters, I wouldn't expect any significant issues. Is this your concern?
  • Hi Rich

    Thanks for your reply.

    After reading your analysis,I check the landcover map and 30m DEM,there are other classes which scale are small in the lake area.I gather the reference material,i'm sure the situation is no problem,so how to deal with my question. In additon,if the biophysical parameters is related to stream.tif, which parameter is it?
  • swolnyswolny Member, NatCap Staff
    Hi @luoyangxu -

    The biophysical parameters are related to the land use/land cover (LULC) map. If your LULC gives a value of "lake" to this area, then you will provide values for load_n/eff_n and/or load_p/eff_p in the biophysical table that are appropriate for lakes. If you do this, then I think Rich is saying that the model should give you good results, even if the stream.tif looks funny in the lake areas.

    ~ Stacie
  • Hi swolny

    Thanks for your detailed explanation.I know what it means.If the result is correct, how do i explain this in my conclusion to others?
  • swolnyswolny Member, NatCap Staff
    Hi @luoyangxu -

    There is another thing that you can do if those flow lines in lakes bother you. You can use a shapefile of lakes (or extract them from your LULC or some other source) and use the lakes to alter your DEM. If you set the DEM to NoData where there are lakes, the model will route nutrient until it hits the lake, then stop, which seems realistic. (It does not do any modeling inside streams or lakes, only until it hits a stream or lake.) Your resulting stream layer will also have NoData where there are lakes. 

    When thinking about why the model creates those different streams in lake areas, there are a few things to consider. It is really hard to model the flow of water in a very flat area and create a single stream network from it. You will see similar results if you make stream networks for your area using different hydrology tools as well. In the case of lakes, where there is a lake there actually isn't a stream with tributaries at all, it is simply covered in water. But the model does not know that this is a lake, so it does the best that it can, and you end up with a stream map that looks somewhere between a lake and a stream. InVEST uses a good water routing algorithm, but still, these are intentionally simple models. If it is important for your work that you model the flow within lakes accurately, then you might want to look at a more complex model that is designed for that purpose.

    ~ Stacie
  • Hi swolny

    I'll take your advice and try again.Thanks for your patient answer.
Sign In or Register to comment.