Deep learning-based extended-period air temperature deviation correction method and system

Through the 3D U-Net model based on deep learning and grid search optimization parameters, the problem of poor temperature deviation correction effect in the existing technology is solved, and more accurate temperature forecasts and more effective climate disaster warnings are achieved.

CN119644481APending Publication Date: 2025-03-18GUANGDONG CLIMATE CENT +1
0 Cites 1 Cited by

Patent Information

Application Number
CN202411831188.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-12
Publication Date
2025-03-18

AI Technical Summary

Technical Problem

When the prior art applies statistical learning methods to systematic deviations of different climate physical quantities, there is a problem of poor correction effect, especially in the temperature forecast of extended periods, which makes it difficult to effectively utilize time and space information.

Method used

Using a deep learning-based method, a temperature deviation correction model based on 3D U-Net is constructed by acquiring and preprocessing the extended period replication data set and reanalyzing the data set, and the optimal parameter combination is determined through grid search to achieve the correction of the temperature deviation.

Benefits of technology

During the deviation correction process, the time information characteristics and spatial information characteristics can be considered at the same time, to reduce errors, improve the accuracy of forecasts, and is suitable for climate prediction of important meteorological elements such as temperature and precipitation, and to provide better early warning information.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to the technical field of weather forecast, and provides an extension period air temperature deviation correction method and system based on deep learning, and the method comprises the following steps: obtaining an extension period back calculation data set and a reanalysis data set, carrying out the preprocessing of the data sets, and obtaining a four-dimensional training data set; the dimensions comprise a set member number, forecast time efficiency, latitude and longitude; inputting the training data set into an extension period air temperature deviation correction model based on 3D U-Net for training; wherein an optimal parameter combination is determined through gridding search to update the extension period air temperature deviation correction model; and pre-processing extension period air temperature forecast data to be corrected, and inputting the pre-processed extension period air temperature forecast data into the trained extension period air temperature deviation correction model to obtain a correction result.
Need to check novelty before this filing date? Find Prior Art