基于SWCI-NDVI特征空间的县域耕地地力遥感反演

Remote sensing inversion of cultivated land fertility at county scale based on SWCI-NDVI feature space

利用遥感反演县域耕地地力状况,快速、准确、高效地实现耕地定级,是区域耕地资源利用与管理的客观需求。本研究以东平县为研究区,利用Landsat-TM卫星影像和耕地地力评价资料,构建以地表含水量指数(SWCI)、归一化植被指数(NDVI)为特征参量的水分植被地力指数(MVFI),进而优选得到最佳反演模型,并在县域空间上进行推广应用和精度验证。结果表明:MVFI与耕地地力综合指数(IFI)的相关系数为-0.753,能综合反映小麦长势、土壤墒情和地力水平,具有明确的生物物理意义;最佳反演模型为二次模型,反演精度较高;模型适用于县域耕地地力的反演,反演结果与地力评价结果的空间分布规律和均匀程度相似,高、中、低等级耕地的面积差均低于2.9%。本研究提出的基于特征空间理论的耕地地力遥感反演方法,可有效提高县域耕地地力的评价效率和预测精度。

It is objective needs during utilization and management of regional cultivated land resource to use remote sensing to accurately and efficiently retrieve the status of cultivated land fertility at county level and realize the gradation of cultivated land rapidly.In this study,with Dongping County as a case,using Landsat TM satellite imagery and cultivated land fertility evaluation data,the moisture vegetation fertility index(MVFI)was constructed based on surface water capacity index(SWCI)and normalized difference vegetation index(NDVI),and then the optimal inversion model was optimized to obtain the best inversion model,which was further applied and verified at the county scale.The results showed that the correlation coefficient between MVFI and integrated fertility index(IFI)was-0.753,which could comprehensively reflect the growth of winter wheat,soil moisture and land fertility,and had clear biophysical significance.The best inversion model was the quadratic model,with high inversion accuracy.This model was suitable for the inversion of cultivated land fertility in the county.The spatial distribution and uniformity of the inversion results were similar to the results of soil fertility evaluation.The area differences between the high,medium and low grades were all less than 2.9%.This study provided a remote sensing inversion method of cultivated land fertility based on the feature space theory,which could effectively improve the evaluation efficiency and prediction accuracy of cultivated land fertility at the county scale.