Open source release of the geospatial-rf library provides a series of functions and wrappers to assist with random forest applications in a spatial context.
This code-base is associated with Williams et al. (submitted), where its underlying development is discussed for the purposes of predicting of exposed bedrock in uipland catchments of Great Britain. This incorporated manually created rock observation datasets with terrain model derivatives (slope, aspect, curvature etc.) as input variables. The same code-base was also applied for the purposes of mineral predictive mapping (Josso et al., 2023)[https://www.sciencedirect.com/science/article/pii/S0169136823003876].
The code provided here contains the core random forest model training functionality with examples of how it can be applied to a synthetic "rock -at-surface" training dataset, for which users are advised to consider the methodology detailed in Williams et al., (submitted). This will provide users with enough information to start applying it to their datasets.
This release aligns with the submission of the associated methodology paper for review.