Precipitation changes in Greece over the past century; what type of stochastic description should we use?

In the presence of long-range dependence, several difficulties emerge in stochastic methods, especially in intermittent and highly-skewed processes, such as precipitation, which cannot be fully supported by the established models in the literature. Here, we analyze a large set of rainfall data in Greece comprising ground records as well as non-conventional data from reanalyses and satellite, and we identify cluster periods of droughts and wet-years in both extreme tails, raising the challenge for their stochastic description. In this light, and after statistical analysis of the whole dataset, we apply the latest version of a genuine stochastic method (i.e., direct use of the process of interest without any transformation, and with a focus on the long-range dependence under various stochastic behaviours; Koutsoyiannis and Dimitriadis, 2021), and we discuss on the implications of the results for future hydrological design scenarios.

Koutsoyiannis, D., and P. Dimitriadis, Towards generic simulation for demanding stochastic processes, Sci, 3, 34, doi:10.3390/sci3030034, 2021.