Aggregating individual DMs into a wall-to-wall map.
there are five main parts to the code (01-05) where we
- first mask out redundant areas,
- then we assemble the maps using three methods
- we rasterize reference data and mask out evaluation areas
- we compute confusion matrices for the evaluation
- and last we evaluate using proportional dissimilarity (bray-curtis)
accompanying csv files (transl*.csv) contain the reclassification key for harmonizing the reference data used in step 3 (rasterize)
Excel files contain results in raw form, later imported into the final manuscript
Output raster datasets for the three methods are available through the ZENODO repository
- probability-based method
- performance-based method
- prevalence-based method