The following process steps were performed as part of an automated geoprocessing script written in the R statistical language, which utilizes the open-source SAGA-GIS and GRASS-GIS software:

1. The Permafrost Classification Model of Northern Alberta (AGS DIG 2018-0008) which is a binary classification map (1 = permafrost present) was generalized by using 35 iterations of a morphological closing filter with a structuring element radius of 1 grid cell and circularly structuring element shape. This step was performed at the scale of the original classification model (15 m horizontal resolution) in order to group areas of discontinuous permafrost into contiguous zones.

2. The morphologically-filtered raster dataset was resampled to a 500 m grid cell size using majority resampling.

3. The up-scaled raster dataset was filtered using a majority filter with a radius of 4 grid cells in order to smooth and slightly simplify the generalized zones of permafrost.

4. The raster dataset from step 3 was then polygonized using the open-source GRASS GIS, whereby the polygon edges were smoothed during the polygonization process while maintaining topological integrity. The polygons were cleaned by removing small areas that are less than 10 km2 in area, and the boundaries were further smoothed using a Chaiken filter with a 500 m smoothing threshold.

5. The number of grid cells that were classified as containing permafrost in the original 15 m resolution model were counted for each polygon ('Permpx_n' attribute). This was divided by the total number of grid cells in each polygon ('Modelpx_n' attribute) in order to estimate the percentage area of permafrost ( 'Perm_pct' attribute) that occurs within each polygon.