Extreme high temperature inter-year prediction method and device and storage medium

By using a Transformer-based regional prediction model, leveraging multi-mode, multi-initial-field data and deep learning methods, the shortcomings of dynamic models in interdecadal prediction of extreme high temperatures are addressed, achieving more accurate predictions of extreme high-temperature intensity and supporting scientific decision-making and disaster prevention and mitigation measures.

CN121009927AActive Publication Date: 2025-11-25NANJING UNIV OF INFORMATION SCI & TECH
6 Cites 1 Cited by

Patent Information

Application Number
CN202511539409.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-10-27
Publication Date
2025-11-25
Estimated Expiration
2045-10-27

AI Technical Summary

Technical Problem

Existing dynamic models suffer from initialization shocks, initial value errors, boundary value errors, and systematic model errors in predicting interdecadal extreme high temperatures, resulting in limited predictive capabilities and an inability to meet practical needs.

Method used

A Transformer-based regional prediction model is adopted. By acquiring multi-mode, multi-initial-field return test data, the extreme high-temperature intensity index is extracted and predicted using deep learning methods. Combined with linear projection to reconstruct the spatial distribution, the prediction accuracy is improved.

Benefits of technology

It significantly improves the ability of dynamic models to predict the intensity of extreme summer temperatures in the interdecadal range, providing a reliable scientific basis for formulating adaptive policies and optimizing resource allocation, and enhancing the effectiveness of disaster prevention and mitigation strategies.

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

Abstract

The invention discloses an extreme high temperature inter-year prediction method and device and a storage medium, and belongs to the technical field of inter-year prediction, and the method comprises the steps: obtaining a second extreme high temperature strength index extracted based on multi-mode multi-initial-field return test data in a target prediction time period, inputting the second extreme high temperature strength index into a pre-trained Transform-based region prediction model, and obtaining a prediction result of the second extreme high temperature strength index; acquiring an extreme high-temperature strength inter-year prediction sequence of each region in the target geographic region; and reconstructing the extreme high-temperature strength inter-year prediction sequence of each region to the spatial grid points through a linear projection method to obtain a spatial distribution prediction result of the extreme high-temperature strength in the target geographic region. According to the method, the capability of a power mode for summer extreme high temperature strength interannual prediction is remarkably improved, and the defects of an existing mode in extreme high temperature strength interannual prediction are effectively overcome.
Need to check novelty before this filing date? Find Prior Art