中国区域作物气象产量统计预报研究进展

ccurate crop yield prediction is crucial for governments to understand production levels, plan agricultural activities, and ensure national food security. Meteorological factors critically influence crop yields, and yield prediction methods and technology systems based on these factors serve as important references. Meteorological yield prediction predominantly employs statistical methods because of their simplicity, ease of implementation, and high accuracy, making them the most widely used techniques in China. This study reviews the application ofthe most commonly used statistical methods in meteorological yield prediction in Chinaincluding the key meteorological factor, climate suitability, and historical meteorological impact index methods. Through extensive data collection and investigation, a detailed overview is provided regarding the crop types and regions where each statistical method has been applied, the quantities and time scales of selected meteorological factors, various calculation approaches for meteorological indicators, and the modeling techniques adopted. The paper elaborates on the effectiveness of each statistical method across different regions and crops, evaluates the performance of integrated statistical models, and compares the forecast accuracy of different approaches. In doing so, several issues in the statistical prediction of meteorological yields are identified. For example, the key meteorological factor method offers advantages such as easy model parameter acquisition and operational applicability; however, it primarily considers the effects of light, temperature, and water, potentially overlooking other meteorological factors and disasters. The climate suitability method comprehensively accounts for the light, temperature, and water resources required for crop growth but mainly focuses on average states with lower temporal resolution, making it difficult to capture the impact of short-term disastrous weather. The historical meteorological impact index method objectively and quantitatively predicts the influence of meteorological conditions on crop yields; however, it is challenging to identify truly similar years. These issues contribute to unstable forecast results. To overcome these limitations, future efforts can focus on integrating data from multiple sources (such as satellite remote sensing, wireless sensor networks, Internet of Things, etc.), introducing advanced data analysis technologies and new statistical methods (such as machine learning and deep learning algorithms), and combining these with crop growth modelsto establish an integrated technology system based on agriculture, meteorology, remote sensing, and artificial intelligence. This will facilitate the development of mixed forecasting models suitable for various spatiotemporal scales, which are efficient and highly accurate. By conducting applicability analyses for different regions and crops, the precision, accuracy, and comprehensiveness of agricultural meteorological services will be enhanced.