Rainfall Runoff Forecasting Based On Land-atmosphere Coupling

The leading time and reliability of runoff forecast are important basises for reservoir operation and the planning of basin flood control, water supply, and electricity generation, etc., and have important significance to the promotion of the management of water resources. Land-atmosphere coupling model, which works out the future runoff process by applying the data of rainfall forecast to hydrological models, is currently a research hotspot and edge for the extending of the leading time of runoff forecast, being a multi-disciplinary field. This paper, starting from the studies of numerical weather prediction model, distributed hydrological model, and land-atmosphere coupling model, carries out its research from the perspectives of theoretical approaches and application tests, and makes some meaningful progresses.The paper first analyzes the necessity to optimize the parameterization schemes for numerical weather prediction model, and establishes the indicator system for rainfall forecasting competence of the assessment model. Moreover, based on Euclid closeness, it sets up the quantitative and comprehensive evaluation model for the parameterization scheme above, which increases the objectivity of the selection of the parameterization scheme in multi-index evaluation. Then, the paper, taking full advantages of GLUE and PSO, improves the sampling rules of GLUE through PSO algorithm and by using the GLUE framework, as well as the optimization capability and the additive judgment rules of PSO, the paper improves the access to equivalent parameter groups. And based on the combination of GLUE and PSO, it proposes an uncertain calibration algorithm, which enhances the representativeness of equivalent parameter groups, and improves the calibration efficiency. Next, the paper, carrying out the application and research in Yalong River basin, establishes a numerical weather prediction model for the basin, and concludes the optimum parameterization scheme for the WRF model of the basin; based on HEC-HMS, it sets up a distributed hydrological model for Yalong River basin, thereby achieving the offline coupling with the basin’s numerical weather prediction model. On this basis, the paper, creates a database of the model parameters(parameter calibration) through using the calibration algorithms of GLUE and PSO, verifies it, and proposes a method for flood forecast, which is based on the database of the model parameters. Finally, the paper, using the Java language in high-performance computing environment, develops and establishes a prototype system for flood forecast in Yalong River basin, which is based on coupled land-atmosphere modeling system, achieving the automatic operation of the coupling model and the dynamic maintenance of the database of model parameters.The application shows: the land-atmosphere coupling model established in this paper can extend the leading time of runoff forecast considerably – it can extend the leading time(around 54 hours at most) in hydrological stations of Yalong River basin to about 6 days; the flood forecast method presented in this paper, which is based on the database of model parameters, can well shows the uncertainties in the forecasting process, and allows high use value of rainfall forecast even after the 96 hours. The research results have significance to the exploration of reservoir operation in combination with the forecast information, the construction and practical application of the ―digital basin‖ of Yalong River.