Innovative Flood Forecasting for Small and Medium Sized Catchments (PAP014359)

Código

PAP014359

Autores

Prof. Dr. Günter Meon, Dr. Gerhard Riedel, Markus Anhalt, Prof. Dr. Joseph Hoelscher

Tema

Flood forecasting and early warning systems

Resumo

In 2010 the hydrological modeling system PANTA RHEI was implemented in the Flood Forecast Center of Lower Saxony, Germany. PANTA RHEI is a conceptual water balance model with special program extensions for the use as operational system for ensemble based river flood forecasting. Numerical weather predictions from the National Weather Service (DWD) are incorporated in order to obtain an extended lead time for the river flood forecasts. The Flood Forecast Center now operates several catchment models from lowland areas to lowland mountain range and covers the half of Lower Saxony (20.000 km²). Here larger cities and industrial plants like Volkswagen are areas at risk. All of these catchments comprise intense water resource management measures like reservoirs and detention ponds (dams), bifurcations, culverts and weirs. These structures can operate automatically or can be operated manually with respect to specific orders by water authorities (NLWKN). The implementation and the successful operation of these partially networked hydraulic structures make up one innovative aspect of the forecast model. For the simulation of small and medium sized catchments the average hyrologic response unit (HRU) element size is about 2.5 km². All the HRUs are further subdivided into hydrotops with emphasis on land use. All the basic catchment models were fitted with a focus on both river flood events and long term water budget. The flood forecast models are operated semi-continuously. All important state variables are saved daily and at the end of any simulation. Once stopped, the operation can be continued immediately e.g. for flood forecast simulations or they can be continued at any previously saved date e.g. for recalibration purposes. When continued in forecast mode, the model continues the water balance simulation but switches automatically from observed to forecasted meteorological input data. The model operator may select an ensemble of meteorological forecasts. At the same time an ensemble of operation rules for the hydraulic structures can be determined. The operation will always show deviations between observed and simulated runoff mainly due to uncertainties in the processing of the sparse meteorological input data. These irregular and temporary deviations need to be corrected by the model operator interactively. For this time consuming operation an efficient and innovative concept was developed, that uses a single comprehensive model parameter (M-parameter) to perform temporally interventions into the water balance simulation. The flood forecast models have proven their forecast quality during three winter periods with intense snow influence as well as during extreme thunderstorms in summer. During the flood event in May 2013 in Germany all of the tributaries from smaller and medium size catchments were reproduced highly satisfying and the published river flood forecasts quite reliable. On the other hand the numerical weather predictions still showed quite high variations in the forecasted precipitations heights during the continuation of the flood event.