基于高光谱的滩涂土壤演化序列发育程度反演方法与机理研究

The degree of soil evolution is important to the study of soil pedogenesis. It has important significance to study the soil evolution process, alteration of properties and the effect of human activity to soil properties. Soils chronosequences are valuable tools for investigating rates and directions of soil evolution as well as for testing theories of pedogesis.Some properties such as iron oxides and clay minerals will change with the age of soil evolution. So the type and content of iron oxides and clay minerals can indicate the degree and environment of evolution..There have been no significant technological breakthroughs in the characterization of soil profiles which is crucial for soil genesis, although significant advances have been achieved in other soil branches.Traditional soil analyses of determining soil age are time-consuming with high cost. Soil spectroscopy can track chemical and physical properties that can serve as tracers to monitor soil genesis and formation rapidly, which provides new method to the study of pedogenesis. Tidal soil derived from marine sediments formed different years ago which has relatively homogeneous parent material. So we can establish soil chronosequence to study soil evolution..The main objective of this study is to evaluate the ability of soil reflectance spectroscopy to predict the degree of soil evolution or soil age. In this study, we collect the tidal soils developed in different historical periods and construct the soil chronosequences after determining the parent material uniformity and soil age. Then we analyze the evolution characteristics of different iron oxides and clay minerals and select the spectral characteristics or index indicating the degree of soil evolution. The multiple regression method will be used to relate the spectral characteristics or index to predict the degree of soil evolution. Then the content and type of iron oxides and clay minerals will be predicted using the method of partial least squares regression and the relationship of spectroscopy and the degree of soil evoltion will be discussed. Finally, we compare the precision of predictions of iron oxides and clay minerals and discuss the mechanism of predicting the degree of soil evolution using spectroscoy. This research is a cross study between soil science and remote sensing. It has theoretical significance of revealing the spectral characteristics of iron oxides and clay minerals and clarify the mechanism of predicting the degree of soil evolution using spectroscoy. It also has practical significance of broadening the utility of spectroscopy technology to pedogesis.

土壤高光谱技术可以快速反演土壤理化属性，为土壤发生学研究提供了新的思路。本项目采集不同成陆时间的滩涂土壤，构建沿海滩涂土壤时间序列，揭示土壤演化序列中氧化铁及粘土矿物的演变规律。利用连续统去除、微分等方法进行光谱数据预处理，结合美国地质调查局（USGS）光谱库中氧化铁及粘土矿物光谱曲线，筛选能够反映土壤发育程度的光谱特征参数，利用多元回归法反演土壤发育程度；利用偏最小二乘回归法反演氧化铁及粘土矿物含量，进而构建土壤发育程度反演模型。对比分析多元回归、偏最小二乘法的反演精度，探讨高光谱技术反演土壤发育程度的机理。本研究是土壤科学与遥感科学的交叉性综合研究，对于揭示滩涂土壤中氧化铁及粘土矿物演变特征、阐明高光谱反演土壤发育程度的机理具有一定理论意义。

土壤时间序列是一种可以用于验证土壤发生学理论的研究方法。土壤剖面中的化学、物理和矿物属性随时间发生有规律的变化，从而表现出系统的、与时间相关的变化趋势，这些属性可以用于分析成土过程的变化方向及变化速率。土壤反射光谱技术是一项可以实现快速、低成本预测土壤属性的新技术。所以，将土壤反射光谱技术应用于时间序列研究将为土壤研究提供新的方法与思路。.本研究以苏北滨海滩涂土壤为研究对象，建立千年土壤时间序列，分析土壤发生属性(有机质、粘粒含量、铁元素、碳酸钙等)的变化规律，讨论这些土壤发生属性变化在土壤反射光谱曲线中的反映及估算结果，探讨反射光谱技术预测土壤剖面发育程度的机理。研究结果表明，在此千年尺度土壤时间序列中，可以看出有机质积累过程、碳酸钙淋溶过程、粘粒迁移过程。土壤剖面反射光谱可以较好估算这些发生属性。土壤有机质的时间变化最显著，可以利用整段光谱结合偏最小二乘法估算，也可以建立“弓曲差”指数结合多元回归方法估算。当全氮含量较低时，土壤全氮与有机碳相关系数也较小，实现反射光谱估算全氮是基于氮的吸收特征；当全氮含量较高时，全氮与有机碳相关性较强且有机碳模型精度高于全氮，说明有机碳对光谱曲线的影响随其含量增加而增大，并掩盖了氮的吸收特征，实现反射光谱估算全氮是基于其与有机碳的相关性。土壤反射光谱对铁及碳酸钙的预测结果也较好。土壤光谱的反射强度及吸收特征可以识别出有机质积累过程、碳酸钙淋溶过程、粘粒迁移过程这些成土过程。.通过准多深度光谱曲线可以估算土壤剖面的发育程度，剖面内0-80cm深土样光谱数据可以得到最优的建模及交叉验证结果。所以，由于土壤风化或成土过程而发生规律变化的土壤发生属性架起了反射光谱与发育程度/土壤年龄之间的桥梁，从而可以实现光谱技术或遥感技术应用于土壤时间序列研究并进行土壤发育程度数字制图。