Meteorological dry and wet spells: Observations, GCM and RCM Simulations in the MedCORDEX domain

Forecasting and projection of dry and wet spell characteristics is of major interest. This paper discusses the representativity of wet/dry spell characteristics estimated from a few decades of data only. Two dry/wet spell indices (EDI, SPI) are applied on daily station observations series and gridded observations. In a next step we show and discuss the added value of regional climate model systems (RCMs)(with grid-spacings of about 50 and 10 km) on global climate model system (GCM) simulations in the MedCORDEX domain. In summary, the predictive capability of the models is limited for change in spell characteristics (it is a multi-decadal variability process and thus initialization sensitive) and the RCMs add some value on the spatial patterns of wet spell characteristics, but no value to patterns of dry characteristics. Given these limitations the models imply more extreme dry spells in the period 2061-2100 compared to 1961-2000, but because of a change in the mean annual cycle of precipitation and not because of a climate with more intense relative extremes.