Asia-Floods

Extreme sea-level events may be more disruptive for human societies and ecosystems than a slowly rising mean sea-level. If the frequency of sea-level extremes changes under the influence of anthropogenic climate change, it may have profound consequences in the estimation of climate change impacts and therefore on the planned adaptation measures. South East Asia, in addition to being one of the most populated regions in the world, is exposed to the impacts of typhoons and extra-tropical cyclones. It is however not yet clear how the frequency of storm floods and extreme rainfall may depend on the external radiative forcing and how large the amplitude of the internal variability and their frequency of occurrence may be.

Asia-Floods will analyze a series of global and regional climate simulations, with different spatial resolutions, conducted over the past millennium, present climate and future scenarios, with the objective of identifying the atmospheric weather patterns in the South East Asia region (defined here as including typhoons and extra-tropical storms) that are most effective in causing coastal flooding either due to storm floods or due to extreme continental rainfall or a combination of both. Since the spatial resolution of global and regional climate models makes it costly to perform a large number of simulations of storm floods and extreme localized precipitation, we will apply a statistical downscaling approach. Within this approach, large-scale predictors that represent the atmospheric dynamics are statistically linked to local climate using observational data sets. Once these statistical models are calibrated they can be applied to past and future global and regional climate simulations to estimate changes in the frequency of these types of extreme events.

In Asia-Floods we will use statistical models based on Classification and Regression Trees and Random Forest. These are suitable, though rather sophisticated, classification (weather typing) schemes that can be optimized towards a prescribed target, in this case storm surges and extreme coastal precipitation. To calibrate the statistical models we will use gridded observations from meteorological reanalysis products and climate simulations with data assimilation on one side, and local observations of daily sea-level and daily precipitation records on the other side.

These results will be linked to two other proposals within this SPP. In the case of the simulations over the past centuries (paleoclimate simulations), the results will be compared to other projects within this SPP that investigate the frequency of flooding from proxy data. In the case of the scenario simulations, the results will be used to estimate the increase in economic costs from coastal flooding.