A Hydrological Forecasting Method

Abstract

The invention discloses an ideal boundary and multiple linear regression-based hydrological forecasting method of an upper and lower limit interval. According to the method, methods of magnifying and reducing actually measured flow at the same multiple proportion are respectively adopted to construct ideal upper and lower limit boundaries, in a calibration phase, the ideal upper and lower limit boundaries are used as targets and a least-squares method is used as a principle to determine a structure and parameters of an upper and lower model of multiple linear regression, and hydrological forecasting of the upper and lower limit interval is realized through an upper and lower limit forecasting result of the calibration phase and a testing phase. An inclusion rate, a relative width, symmetry and a root-mean-square error (adopting a median of the interval as a forecasting value) of the forecasting interval are used as precision assessment indexes, and compared with interval forecasting results of existing neural network methods and regression models with different relative widths, the forecasting result of the method provided by the invention shows better forecasting precision and predicting effect. According to the method adopted in the invention, calculation is simple and quick, a large number of parameter optimization and searching processes and possibility of falling into a local optimum by an optimization algorithm are avoided, and hydrological forecasting time is greatly shortened.

Description

technical field [0001] The invention belongs to the field of short-term runoff hydrological forecasting in hydrology, and more specifically relates to a hydrological forecasting method based on ideal boundaries and multiple linear regression between upper and lower limits. Background technique [0002] The current hydrological prediction methods mainly include empirical correlation method and model method. The empirical correlation method includes hydrological prediction method based on multiple regression method. This method establishes a statistical method of linear or nonlinear mathematical model quantitative relationship between multiple variables. It reflects the law between the quantity of one phenomenon or thing and the quantity change of multiple phenomena or things. The probabilistic hydrological forecast method includes the probabilistic hydrological forecast model based on Bayesian theory. This method first assumes the type of prior distribution and posterior dist...

AI Technical Summary

Problems solved by technology

[0003] The existing hydrological forecast method based on the multiple regression method can realize the forecast of related hydrological variables such as flow, but this method does not consider the uncertainty of the hydrological process, and cannot quantitatively predict the uncertainty, so this method cannot fully reflect the Hydrological process

[0004] Existing probability forecast and ensemble forecast calculations based on assumed error distribution can produce probability forecasts with a certain degree of confidence, but the calculation process is complicated

The method of using the neural network model for interval forecasting is also due to the complex internal structure of the black box model, and its optimization algorithm is prone to fall into local optimum, resulting in unsatisfactory model forecasting accuracy

[0005] Based on the problems existing in the current hydrological forecasting methods mentioned above, hydrological forecasting methods still have shortcomings in terms of forecasting accuracy, algorithm optimization, and computational efficiency, and new hydrological forecasting methods need to be developed

Images