Trends in annual and seasonal maxima of daily rainfall and peak river flow

With the public perception that climate has been changing in the last decades, more and more doubts are raised regarding the validity of the assumption of stationarity for hydrological processes. In this work we explore, in a descriptive and preliminary way, trends for annual and seasonal maxima in daily rainfall and peak river flow, applying a linear regression model allowing the mean of the distribution to vary with time. Over a longer period (50 years), trends do not seem to be significant, while there is indication of some slight, although not spatially coherent, trends in shorter series (20 years). Particularly in summer, there seem to be little concordance in the direction of trends for rainfall and flow, suggesting that other factors, such as evaporation, land use change, water management practices, etc, affect the river flows. Another concern is whether or not the standard linear regression can cope with extreme records. We discuss these issues and look into possible alternative approaches.