基于地类分层的土壤有机质光谱反演校正样本集的构建

发布时间：2018-05-17 05:39

本文选题：理化性质代表性 + 光谱代表性　；参考：《土壤学报》2016年02期

【摘要】：以江汉平原滨湖地区不同土地利用类型的土壤样本为例,比较了基于目标土壤理化性质的浓度梯度法、扩展的基于多种理化性质的综合法(P-KS)、基于光谱信息的KS法、最邻近样本去除法(reduce nearest neighbor samples,RNNS)法和基于浓度分层并结合光谱信息的C-KS、C-RNNS法,基于地类分层再结合上述方法,构建具有不同层次土壤信息代表性的校正集,采用偏最小二乘回归法,建立土壤有机质可见光/近红外光谱反演模型。结果表明,具有单一代表性的浓度梯度法、KS法、RNNS法难以建立适用模型;具有光谱与理化性质二元代表性的C-KS方法模型预测精度得到了明显的提升,相对分析误差(ratio of performance to standard deviation,RPD)为1.66;考虑土地利用类型后,浓度梯度法、RNNS法与C-KS法模型预测精度有明显的提升,RPD分别达到了1.84、1.51、1.75,模型具有良好的适用性。说明具有多层次土壤信息代表性的校正集构建方法对提高土壤有机质可见光/近红外光谱反演模型的适用性具有较好作用。
[Abstract]:Taking the soil samples of different land use types in Jianghan plain as an example, this paper compares the concentration gradient method based on the physical and chemical properties of the target soil, the extended comprehensive method based on various physical and chemical properties, and the KS method based on spectral information. The nearest neighbor sample removal method, reduce nearest neighbor samples-RNNSmethod and C-KSU C-RNNS method based on concentration stratification and spectral information are used to construct calibration sets with different levels of soil information representation based on ground stratification and combined with spectral information, and partial least square regression method is used. The visible / near infrared spectral inversion model of soil organic matter was established. The results show that the single representative concentration gradient method / KS method is difficult to establish a suitable model, and the prediction accuracy of the C-KS model with binary representation of spectral and physical and chemical properties has been improved obviously. The relative error ratio (RPD) of performance to standard deviation is 1. 66. After considering land use types, the prediction accuracy of RNNS and C-KS models is obviously improved, which is 1.84% 1.51% 1.75, respectively. The model has good applicability. It is concluded that the method of constructing correction set with multi-level soil information representation has a good effect on improving the applicability of the visible / near infrared spectral inversion model of soil organic matter.
【作者单位】：武汉大学资源与环境科学学院;地球空间信息技术协同创新中心武汉大学;武汉大学苏州研究院;
【基金】：公益性行业科研专项项目(201412023) 卫星测绘技术与应用国家测绘地理信息局重点实验室经费项目(KLAMTA-201401)资助~~
【分类号】：S153.621

【相似文献】