R code to:
- Run a weigthed random-forest (meta-forest) model selection analysis to identify the best descriptors of the CO2 fertilization effect in the dataset.
- Design a regression model to predict CO2 effects in the dataset.
- Upscale CO2 effects globally
- Compare results with independent approaches
- Zonal statistics for tabuled results
- Plot resulting maps (Fig. 2)
Full details of the implemented methods are available in Terrer et al. 2019