This forum is shutting down! Please post new discussions at

Error in Habitat Quality model --InVEST 3.3.1

When running the habitat quality model (InVEST 3.3.1), I am getting an error after execution that "the datasets' intersection is empty, i.e. not all datasets touch each other". I am attaching an image of the error and the log file. Any ideas what may be happening?


  • RichRich Administrator, NatCap Staff
    Hi, from the log it looks like your access shapefile and landcover raster are in different projections.  One seems to be in UTM 48N and the other in "Pseudo-Mercator"; that probably means the points and raster are in different coordinate systems and thus don't align giving you the error above.

    Can you project your Peudo-Mercator data into UTM 48N and try again?
  • @Rich you are right there was one threat file in UTM 48N. Everything seems to work in that sense now, but I get a strange result in the degradation raster. Specifically, instead of having a 0-1 range like it should (if I understood the model correctly), it has some very high negative values...did you ever have a similar bug in the past? I can attach an image of the raster or the raster files themselves if you would like to see it...
  • RichRich Administrator, NatCap Staff
    Hi @ftoni, I'm not an expert on the model, but I can read the user's guide for what it's worth.  :)  Seems to me from Equation 4 you shouldn't get a negative value unless Hj is negative, which seems like it's user data that shouldn't be negative?

    Anyway, for what it's worth you can dropbox me your datastack to and at least I can figure out the mechanism that's driving them negative.
  • RichRich Administrator, NatCap Staff
    Hi @ftoni, I'm posting a similar message here to what I emailed you privately: 

    Looks like that negative value is from a single pixel or two I can't find easily by inspection on the raster.  If you clamp your visualization from 0 to the max value in the raster, you should get an accurate picture of the degradation values on your map and that stray value should otherwise have no effect on the rest of your results.

    Offhand, I'd guess a minor issue with the data overlap or model implementation, but if you're okay with the visual clamp, I think I'll let it stand until we refactor that model on our end.

    Sound okay for now?
Sign In or Register to comment.