In this Report, the possibility of utilizing joint probability methods in coastal flood hazard component calculations is investigated, since flood risk is rarely a function of just one source variable but usually more of two or three variables such as river discharge, storm surge, wave etc. Joint probability values provide the likelihood of source variables taking high values simultaneously and resulting to a situation where flooding may occur. This report focuses on data preparation, parameter selection and methodology application. The source variable-pairs presented here, which include enough information for calculations, are: (i) surge & wave, (ii) surge & discharge and (iii) wave & discharge. The analysis is focused over 32 river ending (RIEN) points that have been selected to cover a variety of coastal environments along European riverine and estuary areas. In the absence of coincident long-term measurements, the methodology of simulating data observations by modelling was adapted resulting to a set of hindcasts for the three source variables (surge, wave height and discharge).Storm surge hindcasts were performed by utilising the hydrodynamic model Delft3D-Flow that was forced by wind and pressure terms from ECMWF ERA-Interim reanalysis. In a similar way, wave hindcasts were generated by utilizing the latest version of ECMWF ECWAM wave (stand-alone) model, forced by neutral wind terms from ERA-Interim. For the construction of river discharge hindcasts the LISFLOOD model – developed by the floods group of the Natural Hazards Project of the Joint Research Centre (JRC) – was employed. Validation of hindcasts was made over the RIEN point of river Rhine (NL) where coincident observations were available. Considering the physical driver complexity behind interactions among surge, wave height and discharge variables, hindcasts were found to perform quite well, not only simulating observation values over the common interval of interest, but also in resolving the right type and strength of both correlations and statistical dependencies.Results are presented by means of analytical tables and detailed maps referring to both correlation and dependence (chi) values being estimated over RIEN points. In particular, dependencies coming from such analytical tables can be used in an easy way to calculate the joint return period for any combined event by inserting chi in a simple formula containing the individual return periods of source variables. It is then straightforward to estimate the joint probability value as the inverse of the joint return period.Overall, the highest values of (strong / very strong) correlations and dependencies were found between surges and waves mainly over North Sea and English Channel with (such combined) events to take place on the same day (zero-lag mode). Moderate to well category dependencies were found for most sea areas, also on a zero-lag mode. In the case of surge and river discharge, moderate to well category values were found in most cases but not in a zero-lag mode as in surge & wave case. It became clear that in order to achieve such (relatively high) values, a considerable lag time interval of a few days was required with surge clearly leading discharge values. For the case of wave and river discharge, well to strong category values were found but once more mostly in non-zero lag mode indicating the necessity of a considerable lag time interval for dependence to reach such (well / strong) values with wave distinctly leading discharge values.
展开▼
