Regionalized extreme flows by means of stochastic storm generation coupled with a distributed hydrological model. The case of THE BASQUE COUNTRY.
David Ocio, Christian Stocker, Ángel Eraso, Paul Cowpertwait
Wednesday 1 july 2015
11:45 - 12:00h
at North America (level 0)
Themes: (T) Extreme events, natural variability and climate change, (ST) Hydrological extremes: floods and droughts
Parallel session: 9I. Extreme events - Flood Drought
An adequate flood risk management within a certain region or country should be based on a detailed knowledge of actual flood hazards, being extreme flows the most influential factor while defining potential flood-prone areas. While the use of long measured river flow series is clearly desirable to infer design discharges for different return periods, their availability and reliability are not always good enough to ensure that the related statistical analysis is representative. In those scenarios, the application of hydrological modelling based on design storms is a widely-used and accurate option providing an acceptable calibration/validation process is made. However, adopting this path implies that antecedent soil moisture conditions and spatial-temporal distribution of rainfall must be dealt with. As an alternative, a promising methodology for obtaining extreme flows in a regionalized way is presented here, combining a stochastic generation of hourly rainfall and temperature series with spatial consistency along a 500-year period together with a distributed and continuous hydrological simulation. The Basque country, located in northern Spain, is selected as the case study. TETIS model is used to simulate both the soil saturation-desiccation cycle and the generation of surface flow and interflow during flood events. Due to the great variability of possible hydrological situations and the small nature of the analysed basins, a significant number of historical events per gauge station was used to calibrate the model with an overall satisfactory fit (a mean R2 of 0.78 was obtained for the most important episodes). In addition, a spatial-temporal point process stochastic model for rainfall was developed per each month and regionally adjusted based on the daily mean and proportion of dry days and the coefficient of variation, skewness and lag autocorrelation at the 1, 6, and 24 h aggregation levels. As a result, hourly river flows series of 500 years were obtained at each location along the river network, allowing a systematic and consistent statistical analysis of extreme flows.
