EIRP model driven by machine learning for predicting the occurrence risk of southern corn rust (Puccinia polysora Underw.) in northern China,Agricultural and Forest Meteorology

当前位置： X-MOL 学术 › Agric. For. Meteorol. › 论文详情

Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)

EIRP model driven by machine learning for predicting the occurrence risk of southern corn rust (Puccinia polysora Underw.) in northern China
Agricultural and Forest Meteorology ( IF 5.6 ) Pub Date : 2024-07-14 , DOI: 10.1016/j.agrformet.2024.110149
Lujia Yang , Lili Li , Zhaoke Dong , Junsheng Zhu , Wenxiu Guo , Yingying Song , Hongying Cui , Suhong Lv , Lara Sindhu , Xingyuan Men

Global maize production is persistently confronted with the threat of pests and diseases that are migrating to novel regions. Southern corn rust (SCR) is becoming one of the most serious maize diseases within China, with its pathogens overwintering in southern regions and then spreading northward in response to warm and moist airflow. To fill the gap in SCR prediction frameworks in northern China, the ecoclimatic index risk prediction (EIRP) model was developed by integrating the infection risk of SCR spores migrating from southern regions and meteorological parameters. This model is trained on a comprehensive ground meteorological dataset spanning 15 years (2007–2022) from Shandong Province within a high-performance computing environment. Its temperature and leaf moisture index and leaf area index effectively simulate pathogen spore dynamics, identifying the primary infection of maize by the pathogen in the 28th week of the year. A threshold of 29.5 for the SCR ecoclimatic index (SCREI), established by the model, signifies a high risk of SCR infection. Furthermore, by incorporating a spore dispersal index negatively determined by the wind speed at 10 m, the accuracy of the model-predicted SCR disease index (SCRDI) is markedly enhanced. The validation conducted using observational data from 66 maize-growing regions in 2023 demonstrated that the classification algorithm for SCREI achieved a precision rate of 93.51 %. Furthermore, the linear regression model comparing SCRDI to DI yielded a training R² of 0.93 and a testing R² of 0.84, confirming the robustness and reliability of the model. The model undoubtedly offers a valuable approach for enhancing the accuracy of predicting serious airborne diseases such as SCR.

更新日期：2024-07-14

点击分享查看原文

点击收藏

阅读更多本刊新发论文本刊介绍/投稿指南