Spatially-explicit characterization of the extreme distribution of tropical cyclone-induce flood risk

---

Manuel del Jesus, Javier Díez Sierra, Alexandra Toimil, Fernando Méndez Incera, Raúl Medina Santamaría

Thursday 2 july 2015

15:35 - 15:50h at Europe 2 (level 0)

Themes: (T) Flood risk management and adaptation, (ST) Flooding along in rivers and coasts

Parallel session: 12K. Floodrisk - Assessment

---

Coastal areas are among the most populated regions on Earth. They constitute the interface between continental land and the ocean. They are subject to complex flooding dynamics that arise from the interaction of coastal and continental dynamics. In this work, we develop a methodology to characterize the extreme distribution of flooding induced by tropical cyclones in coastal environments under different climates, considering marine dynamics (storm surge and wave run-up) and continental dynamics (precipitation and runoff). The approach followed in this work begins by selecting the tropical cyclones that affected the study area in the past; augmenting the set with synthetically-generated cyclones. The maximum dissimilarity algorithm is then used on the dataset to select for dynamical downscaling the K tropical cyclones that best represent the variability on the data. Numerical simulations are carried out for the K cyclones to derive the spatial fields of wind and rainfall induced by the cyclone. SWAN model is used to derive the wave fields, H2D to derive the storm surge fields and a CUENCAS-like model (IH-Mole) to derive runoff fields. All the flood-inducing dynamics are the input to the RFSM-EDA model that computes flood depths for the study area. Having completed the dynamical downscaling database, a Monte Carlo simulation is used to generate synthetic time series of tropical cyclone occurrence. Tropical cyclone climate is related to the spatial patterns of sea surface temperature (SST) fields, which are used in turn as the main driver of a Monte Carlo simulation. Flood time series are derived from cyclone time series using the dynamical downscaling database and interpolation, for those cyclones that have not been simulated. Our hybrid approach (mixing statistical and dynamical downscaling) allows us to compute any statistic of the complete flooding distribution at every location of the study site. Moreover, making use of SST data from simulations of future climate, obtained from general circulation models, we can study the effects of climate change in these distributions of extremes.