Evaluating Climate Model Bias and Uncertainty for Regional Climate Change Risk Assessment

Climate hazards impose inevitable pressures on natural and managed ecosystems, human infrastructures, natural resources, socio-economic functioning, and security within the water-energy-food-ecosystem nexus. In this context, Climate Risk Assessment (CRA) plays a vital role in providing a better perspective on hazard, vulnerability, exposure, and essential responses for developing adaptation strategies. Considering different modelling assumptions, parameterization and initialization, climate projections entail specific geographical biases and uncertainties. Although bias-adjusting of climate projections before conducting CRA is recommended, a critical question emerges to delineate which climate models/scenarios should be chosen not only to provide the least bias/uncertainty but also to improve reliability for such regional CRA.

The CLIMAAX project provides an open-access toolbox for conducting CRA for a variety of climate hazards (e.g., river and coastal flooding, heavy precipitation, drought, heatwaves, wind, snow, and wildfire). Among them a specific workflow is designed to address how to select climate model projections/scenarios for each NUTS2 region throughout Europe. The workflow is hosted on an interactive platform characterizing (i) bias and (ii) uncertainty currently for precipitation and surface temperature. It relies on a consolidated precalculated dataset and provides rapid calculation and graphical display of bias and uncertainty of climate projections with an easy-to-use GUI to facilitate understanding of climate model skills, their optimal use and exploitation to conduct CRA at regional scales for a wide community of users and practitioners. The first part of the workflow identifies which climate model has the lowest biases in precipitation and temperature. It starts by selecting the region of interest and EURO-CORDEX models. The biases of precipitation and temperature are calculated in percentages and degrees Celsius, respectively, for historical records simulated by climate models in comparison to either Eobs or ERA5 reanalysis datasets. The second part of the workflow provides an absolute or a relative range of precipitation and temperature accumulated in five 20-year windows in the period of 1986 - 2100. As an example, this study demonstrates the application of the workflow to a specific region to showcase its usefulness in the CRA process. Finally, this workflow plays a key role in evaluating biases, understanding uncertainties, reducing uncertainty ranges, which can result in more valuable and credible CRAs.

Link to the current version of the workflow:

https://handbook.climaax.eu/dashboards/bias-uncertainty/

Acknowledgments: This research work was carried out as part of the CLIMAAX project with funding received from the European Union’s Horizon Europe – the Framework Programme for Research and Innovation (2021-2027) under grant agreement No. 101093864.