可解释人工智能在气象预报中的应用和展望.pdfVIP

2024年软件2024

第45卷第1期SOFTWAREVol.45,No.1

基金项目论文

可解释人工智能在气象预报中的应用和展望

仲夏

（内蒙古自治区气象台，内蒙古呼和浩特750306）

摘 要：本研究探讨了可解释的人工智能在现代气象预报服务业务中的应用和展望。目前，AI技术在强对流监测、临近

预报等方面提高了准确性，但仍存在训练数据集不完备、不平衡和模型解释性不足等问题。未来，可解释的人工智能将成为重

要发展方向，提高预测模型可靠性和可信度，其与数值预报融合将成为另一趋势，提供更准确、可靠的天气预报。研究应关注解

释性AI模型开发应用以及AI技术与传统数值预报融合方法，以推进可解释的人工智能在气象预报服务业务中的应用和发展。

关键词：可解释人工智能；气象预报；AI技术与数值预报融合

中图分类号：P409 文献标识码：A DOI：10.3969/j.issn.1003-6970.2024.01.014

本文著录格式：仲夏.可解释人工智能在气象预报中的应用和展望[J].软件,2024,45(01):050-052

ApplicationandProspectsofExplainableArtificialIntelligenceinWeatherForecasting

ZHONGXia

(MeteorologicalStationofInnerMongoliaAutonomousRegion,HohhotInnerMongolia750306)

【Abstract】：This study discusses the application and prospects of explainable artificial intelligence in

modern meteorological forecasting services. Currently, AI technology has improved accuracy in areas such as

severe convection monitoring and short-term forecasting, but there are still issues of incomplete and imbalanced

training datasets, and insufficient model interpretability. In the future, explainable artificial intelligence will

become an important direction of development, enhancing the reliability and credibility of forecastingmodels, the

integration with numerical forecasting will become another trend, providing more accurate and reliable weather

forecasts. Research should focus on the development and application of interpretable AI models, and the methods

of integrating AI technology with traditional numerical forecasting, in order to promote the application and

development of explainable artificial intellige