WIM Updates for ArcGIS Pro 3
WIM Updates
Preprocess Ground Truth Data
This tool formats ground truth features into a WIM-conformant raster. Every cell in the output raster is treated as a true occurrence of some classification (e.g., wetland or nonwetland) during training and/or testing. Previously, Preprocess Ground Truth Data required that background area (area outside of ground truth features and within ground truth extents) belong to a known class. The updated version of this tool will work with unknown background areas, converting these to NoData cells in the output. While this ultimately leads to fewer data cells for training and testing, as there is less known area to rasterize, the updates allow for more streamlined application of point ground truth data and incomplete wetland delineations.
See three use-cases below.
Train Test Split
Curvature, TWI, and DTW using Surface Parameters
The predictor variables tools have been updated to leverage optimized core tools: Surface Parameters (all three tools) and Derive Continuous Flow (Calculate TWI).
Surface Parameters applies an optimal window size over which curvature or slope are calculated, i.e., adaptive neighborhoods. Smaller window sizes are used where more topographic variation occurs locally, and larger sizes are used where there is little variation and it is appropriate to generalize the geomorphology of an area. Leveraging this functionality within the WIM tools allows users to model hydrologic patterns at varying scales. Calculate Curvature uses the output of Surface Parameters directly. TWI and DTW use the Slope output of Surface Parameters as an input to their underlying equations.
Derive Continuous Flow offers an alternative to Fill as a way to hydrocondition a DEM with sinks. This tool addresses erroneous flow direction and accumulation outputs caused by DEM sinks, not by modifying the input DEM data, but through the application of the least-cost path algorithm. This algorithm routes flow into depressions along the steepest downhill path and back out of the depressions along the least steep path. The updated TWI tool allows users to implement this flow routing algorithm when creating the flow accumulation input to the TWI equation. Using this method in appropriate landscapes has been shown to improve wetland modeling accuracy as more hydrologic detail is captured in the TWI that may have been otherwise generalized by flat, filled areas.
See below outputs from older versions of the Curvature, TWI, and DTW tools versus outputs created with the updated tools.
 Traditional Curvature |  Surface Parameters Curvature |
 DTW with Traditional Slope |  DTW with Surface Parameters Slope |
 TWI with Traditional Slope and Fill |  TWI with Surface Parameters Slope and Least-Cost Flow Accumulation |
Postprocess Wetland Predictions
Postprocess Wetland Predictions is a new tool that applies simple, geometry-based cleaning to binary raster wetland predictions. Postprocessing executes the following:
- Majority Filter on the binary predictions raster
- Converts predicted raster cells to polygons
- Deletes polygons smaller than a given minimum wetland size
- Uses Eliminate Polygon Part to fill holes smaller than the minimum wetland size
Users should determine if postprocessing is best for their application.
 Raster wetland predictions (yellow), NWI reference wetlands (blue) |  Postprocessed polygon wetland predictions (red outline), NWI reference wetlands (blue) |
On the Horizon
Additional learning materials showcasing these changes are on the horizon! Be on the lookout for:
- updated WIM User Guide and use cases
- WIM code snippets
- WIM training