[1]覃月凤,董良淼,刘国忠,等.广西区域性暴雨深度神经网络预测模型设计与应用[J].气象研究与应用,2024,45(03):29-36.[doi:10.19849/j.cnki.CN45-1356/P.2024.3.04]
QIN Yuefeng,DONG Liangmiao,LIU Guozhong,et al.Design and application of deep neural network prediction model for regional rainstorm in Guangxi[J].Journal of Meteorological Research and Application,2024,45(03):29-36.[doi:10.19849/j.cnki.CN45-1356/P.2024.3.04]
点击复制
广西区域性暴雨深度神经网络预测模型设计与应用()
气象研究与应用[ISSN:1673-8411/CN:45-1356/P]
- 卷:
-
第45卷
- 期数:
-
2024年03期
- 页码:
-
29-36
- 栏目:
-
研究论文
- 出版日期:
-
2024-09-15
文章信息/Info
- Title:
-
Design and application of deep neural network prediction model for regional rainstorm in Guangxi
- 作者:
-
覃月凤; 董良淼; 刘国忠; 梁存桂; 杨明鑫
-
广西壮族自治区气象台, 南宁 530022
- Author(s):
-
QIN Yuefeng; DONG Liangmiao; LIU Guozhong; LIANG Cungui; YANG Mingxin
-
Guangxi Meteorological Observatory, Nanning 530022, China
-
- 关键词:
-
气象对象表示; 深度神经网络(DNN); 模型性能基准; 数据增强; TimeDistributed包装器
- Keywords:
-
meteorological object representation; deep neural network(DNN); model performance baseline; data enhancement; TimeDistributed wrapper
- 分类号:
-
P457
- DOI:
-
10.19849/j.cnki.CN45-1356/P.2024.3.04
- 文献标志码:
-
10.19849/j.cnki.CN45-1356/P.2024.3.04
- 摘要:
-
遵循AI技术典型应用流程,从确定研究对象的数据表示开始,探讨从测试选用算法工具构建“区域性暴雨”深度学习预测模型,到调整优化模型超参数、对模型性能进行泛化增强,最终实现上线部署,成功对2023年广西“龙舟水”过程作出合理预测。结果表明,结合实际预报服务需求,在深入理解算法和模型架构特点基础上,选用能够模拟天气分析过程、合理解释预测机理的AI算法,可显著增强模型的实用性能。采用TimeDistributed层对样本时间层进行封装,先提取学习气象要素场空间特征和关联特征后再进一步学习时间变化知识,是一种符合天气分析思路且预测性能较好的气象AI模型构建方法。而针对“小概率”气象关注事件,在设定事件标签时适当降低气象评判指标、基于相似天气形势来增强样本数据,引导模型针对关注目标进行正向增强训练,是提升气象AI模型性能的有效技术手段,实现一个“有/无区域性暴雨发生”定性预报的基础分类模型。
- Abstract:
-
Following a typical application process of AI technology, and starting from determining the data representation of the research object, this article discusses the construction process of a deep learning prediction model for“regional rainstorm”from using testing and selecting algorithmic tools, to adjusting and optimizing the model hyperparameters, enhancing the generalization of the model’s performance, and ultimately achieving its deployment online, thus successfully making sensible predictions of the“Dragon-boat Rain”event in Guangxi in 2023. The results show that the practical performance of the model can be significantly enhanced by employing AI algorithms that can simulate the weather analysis process and provide reasonable explanations for prediction mechanisms based on an in-depth understanding of the characteristics of the algorithm and model architecture in combination with the actual demand for forecasting services. The application of TimeDistributed layers to encapsulate the temporal dimension of samples, which first extracts the spatial and related features of meteorological elements before further learning about temporal changes, is a meteorological AI model construction method that aligns with weather analysis approaches and shows better predictive performance. For“low probability”meteorological events of interest, it is an effective technical means to improve the performance of weather AI models by appropriately lowering meteorological evaluation indicators when setting event labels, enhancing sample data based on similar weather situations, and guiding the model to carry out the positive enhancement training for the focus of attention, so as to achieve a basic classification model for qualitative prediction of“with/without regional rainstorm occurrence”.
参考文献/References:
[1] 杨绚,代刊,朱跃建. 深度学习技术在智能网格天气预报中的应用进展与挑战[J]. 气象学报,2022,80(5): 649-667.
[2] CHOLLET F.Deep learning with Python[M].Manning Publications,2022.
[3] WU X D,KUMAR V.The top ten algorithms in data mining[M].Chapman and Hall/CRC,2009.
[4] VASWANI A,SHAZEER N,PARMAR N,et al.Attention Is All You Need[J].Advances in Neural Informalion Processing Systems, Curran Associates, ine .2017,30:5998-6008.
[5] 黄嘉佑.气象统计分析与预报方法[M].北京:气象出版社,2004.
[6] 邓华,张颖超,顾荣,等. 基于PCA-RBF的风电场短期风速订正方法研究[J].气象科技,2018,46(1):10-15.
[7] 陈永义,俞小鼎,高学浩,等. 处理非线性分类和回归问题的一种新方法(I)——支持向量机方法简介[J]. 应用气象学报,2004,15(3):345-354.
[8] 冯汉中,陈永义. 处理非线性分类和回归问题的一种新方法(II)——支持向量机方法在天气预报中的应用[J] .应用气象学报,2004,15(3):355-365.
[9] 熊秋芬,曾晓青. SVM方法在降水预报中的应用及改进[J].气象,2008,34(12):90-95.
[10] 李晶,唐全莉. 基于1DCNN和LSTM的单站逐时气温预报方法[J]. 热带气象学报,2022,38(6):800-811.
[11] 周高进,杨智鹏,彭静. 基于LSTM的地基微波辐射计湿度廓线反演[J].气象科技,2022,50(1):21-29.
[12] 周满国,黄艳国,杨训根. 基于GRU神经网络与灰色模型集成的气温预报[J]. 热带气象学报,2020,36(6): 855-864.
[13] 雷蕾,徐邦琪,高庆九,等.基于卷积神经网络的长江流域夏季日最高温度延伸期预报方法研究[J].大气科学学报,2022,45(6):835-849.
[14] 徐月飞,赵放,毛程燕,等. 基于深度卷积神经网络的阵风锋识别算法[J].暴雨灾害,2020,39(1):81-88.
[15] 陈锦鹏,冯业荣,蒙伟光,等. 基于卷积神经网络的逐时降水预报订正方法研究[J].气象,2021,47(1):60-70.
[16] 张烨方,冯真祯,刘冰. 基于卷积神经网络的雷电临近预警模型[J].气象,2021,27(3):373-380.
[17] 张延彪,宋林烨,陈明轩. 基于卷积神经网络的京津冀地区高分辨率格点预报偏差订正试验[J] ,大气科学学报,2022,45(6):850-862.
[18] 周必高,鲁小琴,郑峰,等. 基于改进的VGG16模型的副热带高压相似识别及应用评估[J]. 气象,2022,48(12):1608-1616.
[19] 陈训来,刘军,郑群峰,等. 基于卷积门控循环单元神经网络的临近预报方法研究[J]. 高原气象,2021,40(2): 411-423.
[20] 谢文鸿,徐广珺,董昌明. 基于ConvLSTM机器学习的风暴潮漫滩预报研究[J]. 大气科学学报,2022,45(5): 674-687.
[21] BI K F,XIE L X,ZHANG H H,et al.Pangu-Weather: A 3D high-resolution model for fast and accurate global weather forecast[J]. Atmospheric and Oceanic Physics, 2022.
备注/Memo
- 备注/Memo:
-
收稿日期:2023-12-18。
基金项目:中国气象局“揭榜挂帅”项目(CMAJBGS202217)、中国气象局气象能力提升联合研究专项(22NLTSY011)、广西气象科研计划项目(桂气科2024M03、桂气科2019M07)
作者简介:覃月凤(1988-),工程师,主要从事天气预报技术研究。E-mail:qinyfjune@163.com
通讯作者:董良淼(1973-),正高级工程师,主要从事天气预报技术研究。E-mail:nn172172@163.com
更新日期/Last Update:
2024-09-15