Burger
Journalpaper

Hindcast regional climate simulations within EURO-CORDEX: evaluation of a WRF multi-physics ensemble

Abstract

In the current work we present six hindcast Weather Research and Forecasting (WRF) simulations for the EURO-CORDEX domain with different configurations in microphysics, convection and radiation for the time period 1990–2008. All regional model simulations are forced by the ERA-Interim reanalysis and have the same spatial resolution (0.44). These simulations are evaluated for surface temperature, precipitation, short- and longwave downward radiation at the surface and total cloud cover. The analysis of the WRF ensemble indicates systematic biases in both temperature and precipitation linked to different physical mechanisms for the summer and winter season. Overestimation of total cloud cover and underestimation of downward shortwave radiation at the surface, mostly when using Grell–Devenyi convection and the CAM radiation scheme, intensifies the negative summer temperature bias in northern Europe (max−2.5 C). Conversely, a strong positive downward shortwave summer bias in central (40–60 %) and southern Europe mitigates the systematic cold bias in WRF over these regions, signifying a typical case of error compensation. Maximum winter cold bias is over north-eastern Europe (−2.8 C); this location is indicative of land–atmosphere rather than cloud-radiation interactions. Precipitation is systematically overestimated in summer by all model configurations, especially the higher quantiles, which are associated with summertime deep cumulus convection. The Kain–Fritsch convection scheme produces the larger summertime precipitation biases over the Mediterranean. Winter precipitation is reproduced with lower biases by all model configurations (15–30 %). The results of this study indicate the importance of evaluating not only the basic climatic parameters of interest for climate change applications (temperature-precipitation), but also other components of the energy and water cycle, in order to identify the sources of systematic biases, possible compensatory or masking mechanisms and suggest methodologies for model improvement.
QR Code: Link to publication