Medium-and-long-term rainfall forecasting method based on weather science method and mathematical statistics method

Abstract

The invention provides a medium-and-long-term rainfall forecasting method based on a weather learning method and a mathematical statistics method. The method comprises the steps: carrying out the annual classification according to the existing rainfall data; screening influence factors by adopting a grade correlation coefficient method, performing qualitative analysis on the factors passing the significance test, and establishing a rolling model; checking each rolling forecasting model, calculating absolute error distribution of the rolling forecasting models after the rolling forecasting models are qualified, and performing interval forecasting; evaluating a model forecasting effect, selecting a forecasting model with a better effect, and comprehensively forecasting and judging a precipitation level by taking an interval forecasting result and a qualitative analysis conclusion of factors of the forecasting model as a criterion; and selecting similar years from the historical rainfall data according to the predicted precipitation level, and representing monthly and annual precipitation of the watershed by the monthly and annual cumulative precipitation of the year respectively. The method has the advantages that the qualitative model and the quantitative model are combined, and the prediction result precision is high.

Description

technical field [0001] The invention relates to the technical field of hydrological models, in particular to a medium- and long-term precipitation forecast method based on synoptic methods and mathematical statistics methods. Background technique [0002] The medium- and long-term hydrological forecast is based on the previous hydrological elements, using the methods of cause analysis and mathematical statistics to scientifically predict the hydrological elements for a long time in the future. Due to the long forecast period, medium and long-term forecasts cannot be calculated using measured data, and various influencing factors of hydrological processes or their own changing laws must be considered for forecasting. At present, the more common methods of hydrological forecasting include genetic analysis, hydrological statistics, fuzzy analysis and artificial neural network. [0003] Precipitation forecast generally has a clear correlation with meteorological factors and cos...

AI Technical Summary

Problems solved by technology

Synoptic methods and cosmic-geophysical analysis methods are only suitable for large-scale trend forecasting, and require a large amount of non-hydrological data, and the accuracy of quantitative forecasting is average; regression analysis methods are difficult to find and find the most important factors due to the limited observation data. Influencing factors, the current accuracy of the time series method is general

Images