CORDEX-WRF v1.3: development of a module for the Weather Research and Forecasting (WRF) model to support the CORDEX community

GIANNAROS, THEODORE M.; SOFIADIS, GIANNIS; FITA, LLUÍS; LORENZ, TORGE; KATRAGKOU, ELENI; POLCHER, JAN; MILOVAC, JOSIPA; BASTIN, SOPHIE

Geoscientific Model Development

Copernicus Publications

Año: 2019 vol. 12 p. 1029 - 1066

The Coordinated Regional Climate Downscaling Experiment(CORDEX) is a scientific effort of the World Climate Research Program (WRCP)for the coordination of regional climate initiatives. In order to accept anexperiment, CORDEX provides experiment guidelines, specifications of regionaldomains, and data access and archiving. CORDEX experiments are important to studyclimate at the regional scale, and at the same time, they also have a veryprominent role in providing regional climate data of high quality. Datarequirements are intended to cover all the possible needs of stakeholdersand scientists working on climate change mitigation and adaptation policiesin various scientific communities. The required data and diagnostics aregrouped into different levels of frequency and priority, and some of them evenhave to be provided as statistics (minimum, maximum, mean) over differenttime periods. Most commonly, scientists need to post-process the raw outputof regional climate models, since the latter was not originally designed tomeet the specific CORDEX data requirements. This post-processing procedureincludes the computation of diagnostics, statistics, and final homogenizationof the data, which is often computationally costly and time-consuming.Therefore, the development of specialized software and/or code is required.The current paper presents the development of a specialized module (version 1.3) for the Weather Researchand Forecasting (WRF) model capable ofoutputting the required CORDEX variables. Additional diagnostic variables notrequired by CORDEX, but of potential interest to the regional climatemodeling community, are also included in the module. ?Generic? definitions ofvariables are adopted in order to overcome the model and/or physicsparameterization dependence of certain diagnostics and variables, thusfacilitating a robust comparison among simulations. The module iscomputationally optimized, and the output is divided into different prioritylevels following CORDEX specifications (Core, Tier 1, and additional) byselecting pre-compilation flags. This implementation of the module does notadd a significant extra cost when running the model; for example, the additionof the Core variables slows the model time step by less than a 5 %. The useof the module reduces the requirements of disk storage by about a 50 %. Themodule performs neither additional statistics over different periods of timenor homogenization of the output data.