麦田耕作层土壤有机质的高光谱监测  

Hyperspectrum Monitoring of the SOM in Plough Layer in Winter Wheat Filed

作　　者：王超[1] 冯美臣[1] 杨武德[1] 李广信[2] 赵佳佳[1] 朱智慧[1] 

机构地区：[1]山西农业大学旱作农业工程研究所,山西太谷030801 [2]山西省农业科学院作物科学研究所,山西太原030031

出　　处：《山西农业科学》2014年第8期869-873,共5页Journal of Shanxi Agricultural Sciences

基　　金：国家自然科学基金项目(31371572;31201168);山西省科技攻关项目(20110311038);山西省青年基金项目(2012021023-5);山西省研究生优秀创新项目(20133063)

摘　　要：为了利用高光谱技术实现麦田耕作层土壤有机质（SOM）的实时、快速、准确监测，对不同物理条件下的麦田耕作层（0-20cm）土壤进行了有机质和高光谱的测定，并利用多元逐步回归统计方法进行了土壤有机质信息波段的提取和监测模型的构建。结果表明，风干土和过筛处理土的光谱反射率要明显高于湿土处理，在350～1100nm处，光谱基本重合，说明二者含有相似的光谱信息，不同土壤粒径造成1100～2500nm的差异。利用多元逐步回归的统计方法构建了不同光谱处理方式的土壤有机质光谱监测模型，并取得了较好的验证效果，其中，基于Savitzky-Golay平滑法的光谱处理不适于试验土壤有机质监测模型的构建；而经一阶微分处理后所构建的光谱监测模型拟合精度高、验证误差小，其中，湿土的土壤有机质监测模型具有较好的实践应用性。研究结果将为麦田土壤有机质的高光谱监测提供一定的理论依据。To quickly, real-time and accurately monitor the SOM of plough layer （0-20 cm） in winter wheat filed using hyperspectrum, the SOM and soil spectral of different soil situation （wet soil, dry soil and sieved soil） were measured. The spectral reflectance was processed with the methods of original reflectance, Savitzky-Golay smoothing, and the first order differential, and then the sensible bands of SOM were extracted and the monitor models of SOM were constructed with the stepwise multi-regression method. The results showed that the spectral reflectance of dry soil and sieved soil was higher than wet soil, and the reflectance from 350 nm to 1 100 nm was coinci- dent which means the soil spectral contained similar soil information and the reflectance difference from 1 100 nm to 2 500 nm may be caused by the soil particle size for dry soil and sieved soil. The SOM monitor models under different soil treatments and different spectral processes were constructed with SMR method and the models were tested using validated sets. There are no band variables that entering the model equation based on smoothing process for three soil treatments, while the models, especially for wet soil treatment which had good practical application, based on the process method of first order differential had a higher accuracy and less error in predicting SOM. The study will provide some theoretical reference and practice application to monitor the SOM of plough layer （0-20 cm） in winter wheat filed using hyperspectrum.

关 键 词：土壤有机质 高光谱 模型 多元逐步回归 

分 类 号：S158[农业科学—土壤学]