The goal of this exercise is to utilize “machine learning” approaches to determine the potential natural vegetation (PNV) in Europe. Predictions are made for six different vegetation classes following the MAES Habitat Classification system.
This repository contains the analytical scripts from data preparation to prediction.
Description of scripts:
-
00_Functions.R = Helper functions for the analysis
-
01_ParameterSettings.R = General parameter script to support predictions on different machines and settings
-
02_PrepareOccurrenceData.R = The preparation and harmonization of input training data for the prediction.
-
03_variablePrep.R = Preparation of input covariates for current and future states.
-
04_runIBISpnvHabitats.R = The primary inference and prediction script using the ibis.iSDM package.
-
05_postprocessPNV.R = Posthoc correction of predictions.
-
07_XXX = Scripts to create output figures.
-
08_preparePublicRelease.R = Script to prepare all predictions for upload in Zenodo.
The data has been uploaded here:
Martin, J. (2024). Current and future European potential vegetation
types (1.0) [Data set]. Zenodo.
https://doi.org/10.5281/zenodo.13686776
A methodological description of the work can be found here:
Martin, J. (2024) Mapping current and future European potential
vegetation to support restoration planning
https://doi.org/10.31223/X59H71