RESUMO
A promising approach to improve cloud parameterizations within climate models and thus climate projections is to use deep learning in combination with training data from storm-resolving model (SRM) simulations. The ICOsahedral Non-hydrostatic (ICON) modeling framework permits simulations ranging from numerical weather prediction to climate projections, making it an ideal target to develop neural network (NN) based parameterizations for sub-grid scale processes. Within the ICON framework, we train NN based cloud cover parameterizations with coarse-grained data based on realistic regional and global ICON SRM simulations. We set up three different types of NNs that differ in the degree of vertical locality they assume for diagnosing cloud cover from coarse-grained atmospheric state variables. The NNs accurately estimate sub-grid scale cloud cover from coarse-grained data that has similar geographical characteristics as their training data. Additionally, globally trained NNs can reproduce sub-grid scale cloud cover of the regional SRM simulation. Using the game-theory based interpretability library SHapley Additive exPlanations, we identify an overemphasis on specific humidity and cloud ice as the reason why our column-based NN cannot perfectly generalize from the global to the regional coarse-grained SRM data. The interpretability tool also helps visualize similarities and differences in feature importance between regionally and globally trained column-based NNs, and reveals a local relationship between their cloud cover predictions and the thermodynamic environment. Our results show the potential of deep learning to derive accurate yet interpretable cloud cover parameterizations from global SRMs, and suggest that neighborhood-based models may be a good compromise between accuracy and generalizability.
RESUMO
A new release of the Max Planck Institute for Meteorology Earth System Model version 1.2 (MPI-ESM1.2) is presented. The development focused on correcting errors in and improving the physical processes representation, as well as improving the computational performance, versatility, and overall user friendliness. In addition to new radiation and aerosol parameterizations of the atmosphere, several relatively large, but partly compensating, coding errors in the model's cloud, convection, and turbulence parameterizations were corrected. The representation of land processes was refined by introducing a multilayer soil hydrology scheme, extending the land biogeochemistry to include the nitrogen cycle, replacing the soil and litter decomposition model and improving the representation of wildfires. The ocean biogeochemistry now represents cyanobacteria prognostically in order to capture the response of nitrogen fixation to changing climate conditions and further includes improved detritus settling and numerous other refinements. As something new, in addition to limiting drift and minimizing certain biases, the instrumental record warming was explicitly taken into account during the tuning process. To this end, a very high climate sensitivity of around 7 K caused by low-level clouds in the tropics as found in an intermediate model version was addressed, as it was not deemed possible to match observed warming otherwise. As a result, the model has a climate sensitivity to a doubling of CO2 over preindustrial conditions of 2.77 K, maintaining the previously identified highly nonlinear global mean response to increasing CO2 forcing, which nonetheless can be represented by a simple two-layer model.