A climate emulator for coastal flooding events

Abstract:

The evaluation of coastal flooding requires the definition of the multivariate marine climate conditions (wave height, wave period, wave direction, wind, surge levels). Historical reanalysis databases are a valuable information source. However, the limited time period covered implies uncertainty in the statistical characterization of extremes. Besides, downscaling is needed to extend data to climate change scenarios or long-term historical periods, or even to understand the interannual variability. A statistical downscaling approach is adopted due to its low computational cost. The relationship between large-scale atmospheric variables (predictor) and local marine climate variables (predictand) is established by means of a physical division of the predictand based on weather types. The multivariate dependence structure of the predictand (extreme events) is introduced linking the independent marginal distributions of the variables by a probabilistic copula regression. Therefore, the climate emulator is a stochastic hybrid model with the following steps: 1) Collecting historical data for the predictor (atmospheric variables) and predictand (sea state parameters, storm-surge); 2) Predictor definition, i.e. using ESTELA method in the case of wave generation characteristics (Pérez et al., 2014a); 3) Defining the most appropriate statistical model (distribution modeling based on weather-type); 4) Stochastic simulation of the present climate; 5) Marine climate downscaling under climate change scenarios (selecting the best GCMs from CMIP5, Pérez et al., 2014b).

References:

Perez, J., Menéndez, M., Méndez, F.J., Losada, I.J. (2014a). Evaluating the performance of CMIP3 and CMIP5 global climate models over the north-east Atlantic region, Climate Dynamics, DOI 10.1007/s00382-014-2078-8.

Perez, J., Menéndez, M., Méndez, F.J., Losada, I.J. (2014b) ESTELA: A method for evaluating the source and travel-time of the wave energy reaching a local area. Ocean Dynamics, DOI 10.1007/s10236-014-0740-7.