3.12.0

This minor release includes the following fixes and features:

• General

  • Update python packaging settings to exclude a few config files
    and the workbench from source distributions and wheels
  • Updating SDR test values due to an update in GDAL's mode
    resampling algorithm. See
    #905
  • Updated our scipy requirement to fix a bug where invest
    crashed if a Windows user had a non-Latin character in their
    Windows username.

• Workbench

  • Fixed a bug where some model runs would not generate a new item
    in the list of recent runs.
  • Enhanced model input forms so that text boxes always show the
    rightmost end of the filepath when they overflow the box.

• Coastal Blue Carbon

  • Fixed a bug where using unaligned rasters in the preprocessor
    would cause an error. The preprocessor will now correctly align
    input landcover rasters and determine transitions from the
    aligned rasters.

• Habitat Quality

  • Removed a warning about an undefined nodata value in threat
    rasters because it is okay for a threat raster to have an
    undefined nodata value.

• HRA

  • Fixed an issue with risk calculations where risk values would be
    much lower than they should be. Risk values are now correctly
    calculated.
  • Fixed an issue with risk reclassifications where most pixels
    would end up classified as medium risk.
  • Added an input field to the model to indicate the number of
    overlapping stressors to use in risk reclassification
    calculations. This input affects the numerical boundaries
    between high, medium and low risk classifications.
  • Various improvements to the model have resulted in a modest
    (~33%) speedup in runtime.

• Coastal Vulnerability

  • Fixed a bug where redundant vertices in the landmass polygon
    could raise an error during shore point creation.

• NDR

  • Added parameters to the sample data to support nitrogen
    calculations.
  • Effective retention calculations have been reworked so that the
    source code more closely matches the mathematical definition in
    the NDR User's Guide. There should be no difference in outputs.

• SDR

  • We have made a significant update to the SDR model's outputs
    described here:
    https://github.com/natcap/peps/blob/main/pep-0010.md.
  • Specific changes include:
    • Legacy outputs sed_retention_index.tif and sed_retention.tif
      have been removed from the model.
    • Two new output rasters have been added, specifically
      • avoided_export.tif, indicating vegetation's contribution to
        reducing erosion on a pixel, as well as trapping of sediment
        originating upslope of the pixel, so that neither of these proceed
        downslope to enter a stream.
      • avoided_erosion.tif, vegetation’s contribution to reducing
        erosion from a pixel.
    • The summary watersheds vector no longer includes the sed_retent
      field and two fields have been added:
      • avoid_exp representing the sum of avoided export in the watershed.
      • avoid_eros representing the sum of avoided erosion in the watershed.
    • Sediment deposition, sed_deposition.tif, has been clarified to
      indicate the sediment that erodes from a pixel goes into the next
      downstream pixel(s) where it is either trapped or exported. This update
      removes a form of double-counting.

• Urban Flood Risk

  • Validation of the curve number table will now catch missing CN_*
    columns and warn the user about the missing column.