New method for evaluating Earth System Models

Until now, biomization techniques were only available for individual models or in form of biome models that were forced with simulated or observed climate distributions. These methods were neither consistently defined nor directly comparable with each other. Since most vegetation models differ in the definition and type of implementation of the vegetation, a direct comparison of simulated vegetation distributions between different models - if at all - has so far only been possible on a very general level.

The new method is based on the biome classification used for reconstructions and the established classification rules of conventional biome models. The method is simple and requires few input data, so it can be applied consistently to most models. The study shows that the new method can qualitatively compete with conventional biome models that have been tuned to match the data. Furthermore, the skill of the calculated biome distribution does not depend on the spatial resolution or the complexity of the vegetation description in the models. The new method works well for simple vegetation models as well as for complex coupled Earth System models with dynamic vegetation.

The new biomization method ensures the direct comparability of the vegetation distribution calculated in Earth System models with pollen-based biome reconstructions, which - due to the relatively good spatial coverage - provides a direct regional picture of the climatic conditions over the continents and their changes in the past. Thus, the method not only facilitates the comparison of vegetation simulations but also offers a powerful tool for the evaluation of Earth system models.