项目名称: 基于光谱分类的区域土壤有机质遥感预测模型研究
项目编号: No.41501357
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 天文学、地球科学
项目作者: 张新乐
作者单位: 东北农业大学
项目金额: 20万元
中文摘要: 不同类型土壤反射光谱特征的主导影响因素不同,土壤有机质(SOM)对反射光谱特征的作用程度及其他土壤属性的影响也不同,单一模型难以准确预测区域不同土壤SOM含量。本研究在土种级别上采集松嫩平原北部主要土壤类型样本,实验室测定其高光谱反射率;基于正交试验与因子分析方法,分析各属性对土壤反射光谱特征的影响及作用大小;计算光谱指数,根据因素影响程度构建光谱特征量(表征土壤反射光谱特征主要影响因素的光谱指数集合);区别于已有研究试图将土壤光谱分类结果与土壤传统分类保持一致,以SOM对土壤反射光谱特征的作用大小及其它土壤属性对SOM光谱响应波段的影响程度为标准、以光谱特征量为输入量,进行土壤光谱分类;利用数理统计、偏最小二乘法与神经网络方法建立基于光谱分类的SOM预测模型,实现区域SOM光谱速测。研究成果将为土壤理化参数速测仪器研制与遥感反演、土壤表层碳库估算等相关研究提供理论依据与技术支持。
中文关键词: 土壤有机质;高光谱遥感;可见光-近红外遥感;土壤水分
英文摘要: The accuracy of soil organic matter (SOM) predicting model at regional scale is decreased by significant spatial heterogeneity of soil type and properties, different determining factors on soil reflectant characterisitics and the integrated effects of soil properties on soil reflectant characteristics. To accelerate the usefulness of soil property hyperspectral predicting models, samples at soil species level in the northern Songnen Plain will be collected and measured with hyperspectral spectroscopy, to build regional SOM predicting model with spectral data. The main factors determining soil reflectant characteristics and the integrated effects of soil properties on soil reflectant characteristics will be identified with orthogonal experiment, spectral and factor analysis methods. Spectral indices will be built for main factors, and Spectral Feature Set (SFS), a collection of spectral indices indicating key factors influencing soil reflectance, is built on the importance of factors. Spectral classification is conducted basing on the effects of SOM on soil reflectant characteristics and the affecting degree of other soil properties on SOM responding bands, with SFS as input. Then regional SOM predicting model will be built basing on Discriminant analysis of SFS and spectral classification with statistic regression, partial least squares method, Neural network and One-in-out strategy, to realize regional SOM quick measuring. The research will provide scientific and technique support for soil property monitering with remote sensing, quick measuring machine inventing and soil carbon pool estimation.
英文关键词: Soil organic matter;Hyperspectral remote sensing;Visible-NIR remote sensing;Soil moisture