AN ASSESSMENT OF THE ABILITY OF GLOBAL CLIMATE MODELS TO REPRODUCE OBSERVED REGIONAL TRENDS IN THE FREQUENCY OF FLOODING

Abstract

Climate change impacts hydrological systems differently in different parts of the world. In this study I assess the ability of various global climate models (GCMs) to accurately reproduce observed historical trends in the frequency of flooding at the global scale. I use a peak-over-threshold (POT) method and focus on the period 1985-2014. First, I compare trends in regions belonging to different climate zone types and find that tropical and arid areas adhere to the “wet-get-wetter, dry-get-drier” pattern, with significant increasing trends in tropical areas and significant decreasing trends in arid areas; the results are less definite in temperate, boreal, and polar regions. Second, I assess whether the outputs from 12 different GCMs are consistent with the results from the reference data, considering both annual and seasonalized time scales. I find that, in general, the GCMs are not able to accurately model the observed regional trends. Northern Australia (NAU), however, is an exception to this, and I use a subset of ten GCMs that perform well in this region to determine the projected changes in the frequency of flood events by the end of the 21st century. My results indicate that flooding is projected to decrease in frequency under all four emissions scenarios considered.