摘要
本文以宁夏银川贺兰县为研究区,针对农作物倒伏监测中传统方法效率低、主观性强的问题,综合利用Landsat-8光学影像与Sentinel-1合成孔径雷达(SAR)数据,结合监督分类、植被指数分析、纹理特征提取与随机森林算法,构建了多模态遥感协同的农作物倒伏监测与面积反演模型。研究结果表明,倒伏作物的NDVI、EVI和LSWI等光谱指数显著低于正常作物,纹理特征可有效辅助倒伏区域识别;随机森林模型的整体分类精度达90.939%,Kappa系数为0.8312,表现出“极好一致性”,能够准确提取倒伏区域。研究验证了多源遥感数据在农作物倒伏监测中的可行性与有效性,为区域农业灾害评估提供了可靠的技术支持。
Taking Helan County,Yinchuan,Ningxia as the study area,this research aimed to address the problems of low efficiency and strong subjectivity of traditional methods in crop lodging monitoring.It comprehensively utilized Landsat-8 optical images and Sentinel-1 Synthetic Aperture Radar(SAR)data,combined with supervised classification,vegetation index analysis,texture feature extraction,and random forest algorithm,to construct a crop lodging monitoring and area inversion model based on multi-modal remote sensing collaboration.The results showed that the spectral indices(e.g.,NDVI,EVI,LSWI)of lodged crops were significantly lower than those of normal crops,and texture features could effectively assist in the identification of lodging areas.The overall classification accuracy of the random forest model reached 90.939%,with a Kappa coefficient of 0.8312,showing"excellent consistency"and enabling accurate extraction of lodging areas.This study verified the feasibility and effectiveness of multi-source remote sensing data in crop lodging monitoring,providing reliable technical support for regional agricultural disaster assessment.
作者
马竣
MA Jun(School of Civil and Hydraulic Engineering,Ningxia University,Yinchuan 750021,China)
基金
宁夏大学大学生创新创业训练计划项目资助:“多模态遥感协同的主粮作物倒伏及轻量化评估系统”(S202510749074)。
