Planetary scale assimilation and machine learning external forcing based climate mode prediction method

Abstract

The invention relates to a planetary scale assimilation and machine learning external forcing based climate mode prediction method. The mode comprises the following steps: (1) separating 1-3 wave information in a background field; (2) forming initial conditions of a mode by using a flow-dependent assimilation technology; (3) forming a sea temperature external forced field of a climate mode by adopting a machine learning method; (4) performing modeling by adopting a machine learning method to form a land external forced field of a climate mode; (5) carrying out modeling on the frozen circle slow change signal through machine learning by means of observation and reanalysis data, obtaining a cross-seasonal extrapolation prediction value of the frozen circle signal, wherein the prediction value serves as a mode exogenous forcing item; (6) forming an atmospheric boundary field; (7) performing seasonal climate prediction; (8) checking and correcting to obtain a revised value; (9) superposing the nonlinear information and the linear change information as a predicted value; and (10) gathering the revised value and the predicted value according to a historical fitting rate to obtain a final prediction result. The climate prediction result can be effectively improved.

Description

technical field [0001] The invention relates to the technical field of climate prediction, in particular to a climate model prediction method assimilating planetary scale and machine learning external forcing. Background technique [0002] Cross-seasonal climate forecasting refers to climate forecasting with a forecast period ranging from 2 weeks to 1 season. Western developed countries, led by the European Community and the United States, have made remarkable achievements in the field of short-to-medium-term numerical weather prediction. At present, the 1-5 day numerical weather prediction technology in Europe has achieved satisfactory results, and has achieved satisfactory results in the field of Widely used in business. However, the climate forecast results for more than 10 days are not satisfactory, such as figure 1 As shown, especially the 15-30-day and monthly-scale climate prediction is in the exploratory stage, which is far from meeting people's needs, and it is no...

AI Technical Summary

Problems solved by technology

[0004] my country's model-based technology is in its infancy. At present, it mainly uses a single climate (or Earth system model) method to assimilate existing observation data, and mainly uses statistical methods. This method is linear, and the "external forcing" signal used "is the sea temperature of the Pacific Ocean (the signal of air-sea interaction - ENSO) and the snow cover of the Qinghai-Tibet Plateau considered as a whole. When the changes of the two are within the normal range, the prediction technology is not predictable

There are also some results of climate models, but they are not used as the main basis for prediction

Images