Temperature and precipitation variability in regional climate models and driving global climate models: Total variance and its temporal-scale components
Abstrakt
This paper presents an evaluation of the variability and its temporal-scale components of daily air temperature and precipitation simulated by two regional climate models (RCMs) and two driving global climate models (GCMs) from the EURO-CORDEX and CMIP5 projects. The analysis was performed for eight geographical areas of central Europe for the period 1971-2000.
After a brief evaluation of total variance, we focus on the ability of the simulations to represent short-term, seasonal and long-term variability components. This analysis was carried out for original time series as well as for series with the annual cycle removed.
A fast Fourier transformation and a Kolmogorov-Zurbenko filter were used to separate the three temporal components. The two methods of time series decomposition generally agree, which confirms the robustness of the results.
Although the observed total variability of daily air temperature and precipitation is not well represented in many cases and considerable differences among simulations exist, the separation of total variance into the three components is well represented. Regarding the air temperature, the results for the time series with the annual cycle removed are generally somewhat worse than for original series.
The simulations tend to underestimate the short-term component of temperature variability, while the results for the other two components are inconclusive. Moreover, it was found that temperature variability caused by the annual cycle is overestimated by the simulation of HadGEM2 GCM and CCLM4-8-17 RCM driven by HadGEM2, which explains the relatively poor results of these simulations regarding air temperature total variance and short-term as well as seasonal components in the original temperature series.
The RCMs provide better results than GCMs for temperature total variance but our results are rather inconclusive in the case of precipitation total variance and distribution of the variance of both studied variables into the three temporal components.