A real-time flood forecasting system with dual updating of the NWP rainfall and the river flow

Abstract

Numerical weather prediction (NWP) models are gaining more and more attention in providing high-resolution rainfall forecasts for real-time flood forecasting. In this study, the weather research and forecasting (WRF) model is integrated with the probability distribution model (PDM) to make real-time flow forecasts in a small catchment located in Southwest England. In order to improve the accuracy of the NWP rainfall and flow forecasts, dual real-time updating is carried out in the forecasting system through data assimilation. The three-dimensional variational data assimilation technique is coupled with the WRF model to assimilate radar reflectivity and traditional meteorological data; meanwhile, the autoregressive moving average model works with the rainfall–runoff model PDM to assimilate real-time flow observations. Four 24-h storm events with different characteristics of rainfall–runoff responses are selected from the study catchment to test the performance of the constructed forecasting system. The flood forecasting accuracy is found to be largely improved by incorporating the NWP forecasted rainfall when the lead time is beyond the catchment concentration time. The assimilation of radar and meteorological data also shows great advantage in improving the NWP rainfall forecasts.