Vegetation and water resource variability within the K?ppen-Geiger global climate classification scheme: a probabilistic interpretation

Worldwide, climate classification schemes are useful for characterizing the biological potentials of terrestrial ecosystems. Among these schemes, the Koppen-Geiger global climate classification has increasingly been used in a wide range of environmental fields. However, the resulting climate information is insufficient to fully portend the dynamics of ecosystem components such as vegetation gradients and water resources. To enhance the interpretability of this climate information, research-driven frameworks are needed to connect terrestrial vegetation and water resource signals to the Koppen-Geiger climate classes. Hence, this study developed a probabilistic framework for characterizing vegetation and water resources variability within the Koppen-Geiger climate classification system. The framework combines an application of entropy theory with multivariate logistic models, and it uses variables including half-degree gridded precipitation, surface temperature, leaf area index, and liquid water equivalence anomalies. Explicitly, entropy-based disorder index (DI) values are quantified for individual variables and thresholds of DI percentiles are used to discretize vegetation and water resources variability zones at the global scale. Multivariate logistic models are later applied to grid-level DI zone attributes and long-term average values to predict Koppen-Geiger climate classes. The statistical evaluation sustained variable models' likelihoods (0.04 <= McFadden's pseudo-R2 <= 0.92) but consistent estimates (0.87 <= Count R2 <= 0.99) within the global climate classes. The developed framework could be an avenue to improve the interpretability of Koppen-Geiger climate classes by providing a probabilistic insight into vegetation and water resources change at regional, continental, or global levels.