Errors with Habitat Suitability

JordaneauJordaneau Member
edited June 2014 in Terrestrial Models
Hi Everyone!

My name is Jordan Neau and I am a geographer with the USGS. A colleague and I recently got a paper published with Biological Conservation entitled "Modeling effects of conservation grassland losses on amphibian habitat". For this paper I used the InVEST biodiversity model in InVEST version 2.4 and 2.5 (now Habitat Suitability). In the paper I modeled LULC data from 2007-2012. I have just recently received LULC data for 2013 and am trying to continue our modeling efforts following the same protocols and design that we established in our efforts for the paper. I used ArcMap 10.2 to create all of the Input LULC, threat masks, and tables (ESRI GRID, and DBF formats). Now when I attempt to run any version of the InVEST models through ArcMap I get an error warning.

Error processing crop Executing:
Start Time: Wed Jun 25 16:32:01 2014
Failed to execute. Parameters are not valid.
ERROR 010333: Invalid Weight neighborhood mask:
Failed to execute (FocalStatistics).

I have gotten this error with InVEST versions 2.3-2.5 and with the 3.0.1 standalone version when using the GRID and DBF files. I have also tried using older versions of ArcMap (ex ArcMap 10.0) I am starting to think that the files I created using ArcMap10.2 are not supported by any version of InVEST models as I have tried mixing versions of the models and ArcMap program, knowing that the files themselves, no matter what version of ArcMap the model is run in will still have been created in a 10.2 format if such a thing exists. Is anyone aware of any changes from ArcMap10.0-10.2 for these file formats or why I was able to previously run files created in an ArcMap10x (I think 10.0 or 10.1), yet can no longer run any models having not changed any protocols? I may be off the mark with this assumption of versioning problems, however I have tried pretty much everything that I can think of to solve this problem and have yet to find a solution.

I was able to get the standalone 3.0.1 version to run successfully by converting my files to tiffs and the tables to csv files. I had to do a little reformatting of the tables to get the model to work (Delete OID and ensure that field names were identical in capitalization of names). However, upon review of the output dataset I did not get the desired outcome. I had similar results to other people where the decay function does not adequately fit the needs of our data and yields improper values. For our paper we used the linear decay function found with the Arc Based models. For the paper we did a sensitivity analysis where we ran the data using the exponential decay function but found that the results did not erode habitat values enough, perhaps a function of the distances we were evaluating at (<1km). We found the linear decay function to be the only function that actually showed the habitat being degraded by the threats. I know this issue has been addressed before in these forums but I would like to bring it up again. To stay consistent with the data we have already published, and to better depict threat influences at the scale we are modeling, our future work requires that there is the ability to select the use of the linear decay function for any new versions of InVEST Habitat Suitability Models.

I would also ask for the end Resampling output cell size function be included in future versions as well. I know it is an extra step that can be easily done outside of the model interface after the model has been run. However the convenience of having included in the model was greatly appreciated by myself and others as we were using input LULC maps with 40mx40m pixels and used outputs of 50mx50m for easy hectare conversions. After processing large datasets through multiple steps to create the required inputs, it was nice to not have to do one more process in the end. It also removed one more opportunity for human errors to be introduced.

In closing, If anyone has a suggestion as to how to get the older models running I would appreciate suggestions. We are trying to use the older models to stay consistent with processing the data with the same products that were used for the data in the paper. This also allows us to continue to use the linear decay function. Any help would be greatly appreciated in solving this issue. I would also ask the consideration of the InVEST programmers and developers for the inclusion of the linear decay function and output cell size function for the standalone version of the Habitat Suitability model in the future.
We would like to continue to use the InVEST modeling suit products to advance the modeling and understanding of ecosystem services here in the Prairie Pothole Region of North America as we see a great benefit in it but we can only do this with some changes and support going forward.

Thank you and I await and welcome all responses.

Jordan Neau

Comments

  • RichRich Administrator, NatCap Staff
    Hi Jordan, thanks for your detailed post. We're aware of the decay function issue. I can see how it's useful, but I can't promise we'll add the resampling feature anytime soon. We're also transitioning away from the ArcGIS platform models toward our standalones and have officially dropped support for all ArcGIS InVEST models except for the coastal protection model. It's ultimately an issue of limited engineering resources on our end.

    We expect the next release of InVEST we'll have the decay issue fixed for this model. Hard to say the timeline for that, but in the meanwhile I'll ping our GIS analysts Stacie and/or Brad who have the most experience tweaking the ArcGIS versions to see if there is something easy they can help you with.
Sign In or Register to comment.