PROCess‐based sEamless development of Earth system prediction over lanD
The modelling community achieved steady progress in dynamical predictions of weather and climate using Earth System models and the last decade has seen an accelerated development for the land surface component. This has led to predictions that are now considered useful for some societal applications over “hot-spot” land areas. However, forecasts performance over land is still substantially weaker compared with ocean, due to the lack of observations, which has hampered the development of well-constrained land processes models. For the short time-scales, while benefiting from daily verification, the models that are developed for intraseasonal-to-seasonal prediction purposes include only that part of the surface variability for which observations are available and that can be modeled/initialized to positively contribute to the forecasts (verificationbased approach). As a consequence, they lack some processes such as those related to ecosystems and their variability. On the other hand,longer time-scales models used for climate variability/change research contain comprehensive vegetation and soil schemes that are intended to represent as many processes as possible, even those that are still poorly constrained or understood. Through the synergy between the climate (process-based) approach and the short-term prediction approach (verification-based), the ambitious objective of this project is to obtain a practicable seamless development of the land modelling applied to Earth System prediction. The main goal will be to obtain a process-basedverifiable land surface model to enhance the performance and usefulness of the predictions. A fundamental contribution to fill in the gap between short-term climate predictions and the Earth System Models used for longer-term climate variability/change analysis will come from the increasing availability of remotely sensed satellite campaigns, which are now providing reliable global land surface products covering the last two decades.