Atmosphere forecasting method utilizing similar set algorithm based on time weight

Abstract

The invention belongs to the field of atmospheric science, and provides an atmospheric forecasting method using a similar set algorithm based on time weight. Similarity judgment is performed accordingto the distance between a current numerical mode forecasting result and historical forecasting, the importance of different forecasting times in the judgment process is considered, and different weights are given to different forecasting times. and finally, the most similar real-time observation result in history is selected as a similar ensemble forecasting result. After the weights of differentforecasting times are increased, the importance of the current forecasting time can be highlighted, and the similarity judgment between the current forecasting time and the historical forecasting time is more accurate, so that a more accurate similarity set is obtained, and the atmospheric variable forecasting precision is improved.

Description

technical field [0001] The invention relates to the field of atmospheric science, in particular to an atmospheric forecast method using a time weight-based similarity set algorithm. Background technique [0002] Similar Ensemble Forecast (AnEn) was first proposed by Professor Luca of the US National Environmental Atmospheric Administration in 2013 (Luca et al., 2013). It is based on the assumption that the weather system is relatively stable, and it is believed that similar historical numerical model forecast errors , which can be used to correct the current numerical model forecast results. Using the historical forecast results of the numerical model as the training data to find similar ones can greatly increase the sample size of the training data, and on the other hand, it can avoid the systematic error of the model that needs to be considered when looking for the observation samples directly. In addition, considering the time continuity of predictors and the similarity ...

AI Technical Summary

Problems solved by technology

[0010] 1. There are too many combinations of variable weights, and the amount of calculation is huge. For example, assuming that the minimum weight factor is 0.1 and the predictor variable is 5, there are as many as 252 weight combinations, and the amount of calculation is huge;

Images