基于光谱信息的龟裂碱土盐渍化程度预测研究

Soil salinization and its induced soil degradation is a major problem in arid and semi-arid area. Real-time and accurate acquisition of soil information is important for the evaluation of soil salinization to prevent its degradation and realize agriculture sustainable development. In our prophase study, close relationship was found between field measured spectra and the salt-alkali parameters of Takyr solonetzs. We established models to predict the degree of salinization, and qualified the effect of Takyr solonetzs on the characteriscs of canopy spectra reflectance for aboveground vegetation. In this project, northern Ningxia Yinchuan would be select as the study area. Base on the soil science, geostatistics,and remote sensing methods, take soil salinization parameters, the measured spectra data of surface soil and remote sensing image as the source of information, analyze the characteristics of spectra reflectance of different configuration of surface and effect factors, combine the measured spectra data related to different soil salinization degree with soil salinization parameters of laboratory in order to establish a soil salinization monitoring model base on the best spectral transformation and response wavelengths, explore the scale effect of different remote sensing data of Takyr solonetzs. Combine the investigation of field, the field measured spectra and remote sensing images, differences between field measured spectra and remote sensing images could be detected to reveal the characteristics of soil and vegetation, analyze match field measured spectra with remote sensing images using the spectral matching methods. Field measured spectra could be used to validate and improve the accuracy of remote sensing images in predicting salinization degree of Takyr solonetzs. The study would provide a convenient, efficient choice for real-time and precise prediction on Takyr solonetzs.

盐碱地是我国重要的后备耕地资源，及时、精准获取土壤盐渍化信息，对治理盐渍土、防止其进一步退化和促进农业可持续发展至关重要。本项目选择宁夏北部地区龟裂碱土为研究对象，运用土壤学、地统计学及遥感科学方法，以土壤实测光谱数据、遥感影像数据和室内盐渍化指标测定数据为基本信息源，研究不同表观龟裂碱土光谱特征及其影响因素，从龟裂碱土实测光谱数据中选取与土壤盐渍化指标相关性最好的光谱变换方式和响应波段，构建龟裂碱土盐渍化信息预测模型；探索不同来源土壤遥感信息的尺度效应，分析土壤及其上覆植被实测光谱和遥感影像信息表达的耦合程度，揭示实测光谱与遥感影像的匹配关系，通过实测光谱来拟合遥感影像光谱，选用修正的遥感影像建立龟裂碱土盐渍化程度预测模型，为实现龟裂碱土及其同类地区盐碱地信息大面积、实时、准确预测提供科学依据。

土壤盐渍化已经成为干旱和半干旱地区最危急的环境问题之一。宁夏银北地区盐渍化土壤面积是非盐碱耕地面积的2.15倍，有较大的开发潜力。掌握盐渍化土壤的分布、性质和盐渍化程度等信息，对及时采取应对措施，防治其进一步退化、增加后备耕地资源、保障粮食安全具有重要意义。.本项目选择宁夏银北地区为研究区，以龟裂碱土野外实测光谱和遥感影像为基本信息源，运用土壤学、空间信息技术和地统计学等分析方法，系统研究不同表观龟裂碱土光谱特征；明确土壤及其上覆植被在实测光谱和遥感影像上信息表达的异同点，确定二者的匹配关系；利用经过尺度转换的实测光谱特征来修正遥感影像光谱信息，最后利用修正过的Landsat 8 OLI影像来实现龟裂碱土盐渍化信息的大面积、快速、准确预测。主要结论如下：.（1）通过野外光谱反射率可以将研究区主要盐结皮类型进行分类。除了Ca2+和K+外，利用光谱反射率或光谱指数可以准确预测0～5 cm结皮层pH值、EC和其他盐分离子。光谱反射率或光谱指数对0～5 cm结皮层各指标的预测稳定性和能力较0～20 cm强。重采样处理后的高光谱波段反射率在不同程度土壤含盐量和含水率条件下和OLI影像波段反射率具有显著相关性。基于OLI影像敏感波段的土壤含盐量和pH反演效果普遍优于基于盐分指数，但反演精度明显低于实测光谱。经实测光谱校正后的OLI影像土壤含盐量反演模型精度明显提高。.（2）根据冠层光谱特征能够区分野外典型植被类型。土壤pH值与植被冠层红边参数普遍呈极显著负相关关系；而土壤全盐含量与植被冠层各红边参数均无相关性。基于重采样光谱建立的土壤含盐量和pH反演模型精度均高于OLI影像模型。基于OLI影像敏感波段的含盐量与pH反演模型略优于植被指数。基于实测高光谱植被指数和敏感波段分别对OLI影像反演含盐量模型进行校正，精度均有了显著提高。