沙漠与绿洲气象2025，Vol.19Issue(3)：132-138,7.DOI:10.12057/j.issn.1002-0799.2312.02001
基于数值—循环神经网络相结合的回南天客观预报方法
An Objective Prediction Method for High-Humidity Weather Based on the Combination of Numerical Forecasting and Recurrent Neural Networks
摘要
Abstract
Using data from the Heyuan Meteorological Station and observational records of the"High-Humidity Weather"event in Dongyuan from 2015 to 2021,a Gate Recurrent Unit(GRU)model was developed to predict indoor surface temperatures for the next 24 and 48 hours.The model was evaluated using observation data in 2022.A key innovation of this research lies in combining artificial intelligence with numerical forecasting to improve the prediction accuracy of the"High-Humidity Weather"event.The validation results indicate that the GRU model is significantly better than the multiple linear regression model.Sensitivity experiments further reveal that accurate numerical prediction enhances the robustness of the GRU model.This method of combining artificial intelligence and numerical prediction highly depends on the accuracy of dew point and air temperature prediction.关键词
回南天/室内地表温度/循环神经网络/数值预报Key words
High-Humidity Weather/indoor surface temperature/recurrent neural network/numerical forecasting分类
天文与地球科学引用本文复制引用
赖玉林,胡家晖,甘海,钟海彬,林伟丽..基于数值—循环神经网络相结合的回南天客观预报方法[J].沙漠与绿洲气象,2025,19(3):132-138,7.基金项目
新疆维吾尔自治区气候中心科研基金课题(QHCX-2023-06) （QHCX-2023-06）
