Using Copulas for the evaluation of multi-parameter dependencies in reanalyses

Abstract:

Reanalyses are an unparalleled source of data for the evaluation of climate and its variability providing 4-dimensional reconstructions of multiple meteorological parameters describing the atmopsheric's system state. Yet, the vast majority of intercomparison studies focus on the evaluation of single parameters in time and space without looking at the statistical dependence between two or more parameters. But this is necessary especially if you are interested in specific events where two or more parameters are involved, i.e., so called compound events. Since the performance of a reanalysis varies considerably with respect to parameter, region, and time-scales, it is necessary to identify a reanalysis which optimally represents the conditions for compound events. Recent studies have therefore investigated the representation of natural hazards such as wildfires, heat stress, droughts by evaluating corresponding indices based on two or more parameters. We employ a more sophisticated approach by using copulas. With this method, we first aim at evaluating the multivariate statistical distribution between two or more parameters. In a second step, the identified dependencies are then related to different compound events based on the specific multivariate sample space. Copulas provide a sound statistical framework to assess such multivariate statistical distribution by modeling the dependence structure of variables separately from their marginal distributions. We will present results for a joint copula-based evaluation of temperature and humidity related to the natural hazards mentioned above.