Evaluation of Native Earth System Model Output with ESMValTool

Projections from Earth system models (ESMs) are essential to allow for targeted mitigation andadaptation strategies for climate change. ESMs are state-of-the art numerical climate modelsused to simulate the vastly complex Earth system including physical, chemical, and biologicalprocesses in the atmosphere, ocean, and on land. Progress in climate science and an increase inavailable computing resources over the last decades has led to a massive increase in thecomplexity of ESMs and the amount of data they provide. For this reason, innovative tools fora frequent and comprehensive model evaluation are required more than ever. One of these toolsis the Earth System Model Evaluation Tool (ESMValTool), an open-source communitydiagnostic and performance metrics tool.

Originally designed to assess output from ESMs participating in the Coupled ModelIntercomparison Project (CMIP), ESMValTool expects input data to be formatted according tothe CMOR (Climate Model Output Rewriter) standard. While this CMORization of modeloutput is a quasi-standard for large model intercomparison projects like CMIP, this complicatesthe application of ESMValTool to non-CMOR-compliant data like native climate model output(i.e., the raw output directly provided by a climate model). In this work, we describe anextension of ESMValTool that allows reading and processing native climate model output. Thisis implemented via a CMOR-like reformatting of the input data during runtime. For modelsusing unstructured grids, data can optionally be regridded to a regular latitude-longitude grid tofacilitate comparisons with other data sets.

This extension opens up the large collection of diagnostics provided by ESMValTool for thesemodels. This includes assessing the models' performance against observations, reanalyses, orother simulations, the evaluation of new model setups against predecessor versions, theCMORization of native model data for contributions to model intercomparison projects, andmonitoring of running climate model simulations. Currently, five ESMs are supported: CESM2,EC-Earth3, EMAC, ICON, and IPSL-CM6. Support for other climate models can be easilyadded. ESMValTool is an open-source community-developed tool and contributions from othergroups are very welcome.