Evaluating Daily Temperature Characteristics in the km-scale CORDEX-FPS Convection Ensemble in the Greater Alpine Region

The evaluation of km-scale climate models is often centered on variables and time scales for which added value is expected, e.g. short-term wind and precipitation extremes. However, potential shortcomings of basic temperature characteristics might be overlooked, even though temperature is a key variable. Especially in regions with complex topography, parameterizations and model physics that were originally designed for larger-scale applications might be inadequate; although a better resolved orography might support an improved representation of temperature, it is not guaranteed.

In our research, we investigate the representation of temperature characteristics in complex terrain in km-scale regional climate models. For this purpose, we exploit the CORDEX-FPS ensemble on convective phenomena over the Alps at two resolutions: km-scale (2.2 - 4 km) and coarse-scale (12 - 15 km). We evaluate these ensembles against SPARTACUS, a high-resolution gridded observation-based dataset in Austria, during the period 2000-2009.

We find season-dependent biases in both the coarse-scale and km-scale model ensembles accompanied by a high ensemble spread. Hot and cold extremes are generally overestimated, with the km-scale models showing a higher overestimation than their driving coarse-scale models. Consequently, most km-scale models overestimate the frequency of hot days (daily maximum temperature >30°C) and frost days (daily minimum temperature <0°C). Additionally, both ensembles exhibit a strong and mostly negative elevation-dependent bias, which is most pronounced for daily minimum temperatures. The biases become more negative with increasing elevation for both ensembles and become as large as -5.0°C for the multi-model mean. The near-surface temperature lapse rate is therefore systematically overestimated for both daily minimum and maximum temperature and season-dependent. Our results highlight that adequately representing temperature characteristics remains challenging even at the km-scale.