不同尺度冬小麦氮素遥感监测方法及其应用研究

Approach to estimation of winter wheat nitrogen by using remote sensing technology on diverse scale and its application

该文以航空影像、地面冠层光谱数据及同步观测的植被生化数据为基础,探讨了冬小麦冠层氮素监测的遥感方法。该方法应用于Lukina变量施肥模型,研究了基于遥感影像变量施肥量的计算方法。为实现以上目标,首先采用矩匹配和反射率转换方法,对获取的机载实用模块化成像光谱仪(OMIS)影像进行辐射校正;然后结合航拍相片及地面高精度差分GPS定位点坐标对高光谱影像进行几何校正。以预处理后的反射率影像和冠层光谱数据为数据源,采用倒高斯模型拟合冬小麦红边光谱曲线,并构建红谷位置、红边位置和红边宽度等光谱特征参量。通过对红边光谱特征量和实测氮素进行统计分析,寻找相关性显著、拟合误差小的最佳光谱特征量,并用于预测冬小麦冠层的氮素含量。统计相关分析结果表明:拟合曲线和图像反射率曲线面积差和实测的氮素含量有最高的相关性,且相关性达到极显著。最后,把该氮素预测方法集成到Lukina变量施肥模型中,结合反射率影像数据生成变量施肥处方图。文中探讨的最佳氮素预测方法改善了氮素预测的精度;基于影像的面状信息获取技术克服了点状信息的不足,使变量施肥技术更利于实用和推广。

This paper focused on estimating winter wheat nitrogen content from hyper-spectral airborne images and ground measured spectral data, and research methods for variable-rate fertilization based on remote sensing images by combining the nitrogen estimation method and Lukina＇s diverse fertilization model. In order to reach the target, firstly, moment-matching and reflectance conversion were employed to correct the image＇s radiation difference, then the OMIS image was geo-corrected using a geo-referenced aerial photograph image. The Inversed-Gaussian model was used to simulate winter wheat red edge of the processed OMIS image spectra and ground measured spectra, and the red edge parameters, such as red edge position, red edge wide were derived. The correlation between red edge parameters and measured nitrogen content was calculated, and the optimal red edge parameter whose correlation was the most significant and RMSE was the smallest was selected to estimate wheat nitrogen content. The results show that the correlation coefficient between winter wheat nitrogen content and the area difference between simulated spectral curve and actual reflectance curve is the most significant. Lastly, the plant nitrogen estimating method was integrated to Lukina＇s variable-rate fertilization model, and the variable-rate fertilization prescription map was created based on the OMIS hyper-spectral image. The optimal nitrogen estimating method in this paper improved the nitrogen estimating accuracy in Lukina model, and the technology of obtaining large area information overcame the shortcoming of point sampling technology, which made variable-rate fertilization technology more practical and easier to be generalized.