This forum is shutting down! Please post new discussions at

HRA model InVEST 3.3.3. log file errors

ftoniniftonini Member
edited June 2017 in Marine Models

I just ran hra at InVEST 3.3.3 using the sample data provided. Can anyone possibly help me understand what causes the errors highlighted in the attached image (and log file)? Is this a concern that may change final output numbers? Thanks!

Post edited by ftonini on


  • DaveDave Member, Administrator, NatCap Staff
    I haven't verified this, but it may just be that those particular habitat - stressor combinations do not overlap in space. The rasters raising those errors are supposed to represent the overlap between two input shapefiles. So as long as the results indicate there is 0 risk to Hardbottom from Finfish Aquaculture, then I think it's all good!
  • Thanks @Dave! I will keep this in mind. A couple more clarifications, if you do not mind:

    1) Looking at the output from sample data model run, am I understanding correctly that the streaky/linear pattern of higher risk along certain boundaries is an effect of the habitat overlap, but not necessarily because that streak has so much higher risk than the immediate surroundings? Or should I interpret this differently? 

    2) What would be the difference in the input shapefiles provided in the case of Habitat Risk versus Species Risk? Specifically, what would the species shapefile represent? E.g. occurrence areas envelopes?

  • DaveDave Member, Administrator, NatCap Staff
    Not at all!

    1) Yes, I believe that's exactly it. After the model rasterized the input shapefiles, there is overlap in the rasters around the edges, so both habitats are at risk there and that's reflected in the 'ecosystem risk' output, which sounds like the one you're looking at. You could also look at the risk results for the individual habitats and those edge effects would not show up.

    2) Yes, I think you can use the same model to assess risk to species rather than habitat if your input represents a species distribution (positive occurrence area like you suggest) rather than a habitat distribution (environment suitable for one or many species). Other bona-fide ecologists may wish to weigh-in!
  • @Dave and everyone,

    Thanks for the info! By looking at the code and running the model, it seems to me that the hra_preprocessor returns a ratings CSV with labels starting with "HABITAT" (e.g. "HABITAT NAME") even if I choose the species directory. Is this intended behavior? Also, running the hra main model using species layers instead of habitat, produces the same intermediate directory names, like "Habitat_Rasters", instead of "Species_Raster" this also intended behavior?

    Thank you!
  • DaveDave Member, Administrator, NatCap Staff
    I'm actually not sure what is supposed to be different under the hood for 'habitat' vs 'species' risk assessment. @Rich? @James?
  • jdouglassjdouglass Administrator, NatCap Staff
    The preprocessor's implementation appears to regard 'habitats' and 'species' the same way, so all of the output CSVs would have HABITAT even if it's actually from a species vector.

    That being said, the HRA preprocessor currently only handles either habitats or species, and not both.
  • RichRich Administrator, NatCap Staff
    That sounds like something I've heard before too.  I forwarded to Katie to see if she can comment on why that is or if it matters.
Sign In or Register to comment.