Multi-model analysis of RCM simulated 1-day to 30-day seasonal precipitation extremes in the Czech Republic

Abstract

► Statistical analysis of 1-day to 30-day seasonal precipitation extremes. ► Ensemble of 14 transient regional climate model simulations. ► Negative ensemble mean bias in the quantiles in summer, decreasing with duration. ► Extreme quantiles better reproduced than moderate quantiles in the other seasons. ► Increases in the quantiles for all durations, return periods and seasons. This paper presents an analysis of seasonal precipitation extremes in an ensemble of fourteen transient regional climate model (RCM) simulations for durations varying from 1 to 30 days. It is assumed that these precipitation maxima follow a generalized extreme value (GEV) distribution with time-varying parameters. In addition, spatial pooling of the maxima at various grid boxes is applied to reduce the uncertainty of the parameters of the GEV model. Relatively large positive biases in moderate quantiles of long-duration precipitation extremes have been found in autumn, winter and spring. The extreme quantiles are much better reproduced in these seasons. In summer, the ensemble mean bias in the quantiles is negative and decreases (in absolute value) with increasing duration. The ensemble mean changes in the quantiles are positive for all durations, return periods and seasons considered. Large quantiles of the distribution of summer precipitation extremes increase more than moderate quantiles and this increase is relatively large at short durations. The dependence of the changes on duration and return period is much weaker in the other seasons, in particular in winter and spring. Although the spread in the RCM ensemble is considerable, most of the RCM simulations agree on the sign of the changes.