土壤湿度空间分布特征分析与模拟研究

发布时间：2018-03-16 22:38

本文选题：土壤湿度　切入点：HASM模型　出处：《山东农业大学》2016年硕士论文　论文类型：学位论文

【摘要】：本文以东营、德州和滨州为研究区,基于地面实测获取的土壤湿度数据,利用GS+7.0软件进行空间变异特征分析,选取最适宜半变异方差函数参数。采用统计分析、GIS插值、缓冲区分析等方法,分析研究区域土壤湿度状况和空间变异规律。常用的插值模型和HASM插值模型相结合,对比得出模型的优缺点和适用范围。对空间变异程度大的区域增加采样点数量,提高了HASM模型插值结果的精度。采用野外高光谱数据与采样点数据建立土壤湿度的高光谱估测模型,实现快速获取关键点的土壤湿度。将插值所得到的土壤湿度值作为“空间扩张”后的点数据,建立与遥感卫星波段反射率的关系模型,实现大面积土壤湿度值的获取。土壤湿度属于中等空间变异性。建立了野外高光谱土壤湿度预测模型,模型决定系数R2为0.85。建立了野外土壤湿度值与对应室内测量值关系模型,模型决定系数R2为0.97。在LandSat8影像中,建立了基于波段多种组合形式光谱参量的土壤含水量反演模型,模型决定系数R2=为0.65。本文研究的主要内容包括:(1)根据土壤湿度实测数据,分别采用统计分析、GIS插值、缓冲区分析等方法,对山东省德州、滨州、东营三地区土壤湿度进行空间分析,通过半变异函数参数选取实验,进行空间探索性分析,选出最优空间变异参数,找到最适合的插值模型,经过交叉验证,比较插值方法优缺点,提高插值结果精度。分析研究区土壤湿度的空间分布特征。(2)为解决采样点数量不足的问题,采用HASM高精度模型插值方法,对研究区的六组数据分别进行插值实验,得到精度更高的土壤湿度空间分布模拟。对研究区空间变异性大的区域增加采样点数量,重新对土壤湿度进行空间分布模拟。(3)利用相关分析和多元逐步回归分析,地面实测土壤湿度数据结合野外高光谱数据建立土壤湿度估测模型,获取关键点土壤湿度值。(4)建立高光谱数据与土壤湿度关系模型。选取土壤湿度敏感波段,采用多元线性回归方法,运用插值后的作为“空间扩张”后的点数据,建立高光谱数据与土壤湿度关系模型,获得大面积土壤湿度。
[Abstract]:Taking Dongying, Texas and Binzhou as the study areas, based on the soil moisture data obtained from the ground measurements, the spatial variation characteristics were analyzed with GS 7.0 software, and the most suitable parameters of semi-variance variance function were selected. The statistical analysis and GIS interpolation were used. Buffer analysis and other methods to analyze and study the regional soil moisture status and spatial variability. Commonly used interpolation model and HASM interpolation model combined, Compare the advantages and disadvantages of the model and the scope of application. For areas with large spatial variation, increase the number of sampling points, The accuracy of interpolation results of HASM model was improved. The hyperspectral estimation model of soil moisture was established by using field hyperspectral data and sampling point data. The soil moisture value obtained by interpolation is regarded as the point data after "spatial expansion", and the model of reflectivity of remote sensing satellite band is established. The prediction model of field hyperspectral soil moisture is established, and the determination coefficient R2 is 0.85. The relationship model between field soil moisture value and corresponding indoor measurement value is established. The model determination coefficient R2 is 0.97. In the LandSat8 image, a soil moisture inversion model based on spectral parameters in various spectral forms is established. The model determination coefficient R _ 2 = 0.65. The main contents of this study include: (1) based on the soil moisture measured data, the model determination coefficient R _ 2 = 0.65. The spatial analysis of soil moisture in three areas of Texas, Binzhou and Dongying in Shandong Province was carried out by means of statistical analysis, GIS interpolation and buffer analysis, and the spatial exploratory analysis was carried out through the experiment of parameter selection of semi-variable function. The optimal spatial variation parameters are selected and the most suitable interpolation model is found. After cross-validation, the advantages and disadvantages of the interpolation method are compared. In order to solve the problem of insufficient number of sampling points, HASM high precision model interpolation method was used to carry out interpolation experiments on six groups of data in the study area. The spatial distribution simulation of soil moisture with higher precision was obtained. The number of sampling points was increased in the study area with high spatial variability, and the spatial distribution simulation of soil moisture was carried out again by means of correlation analysis and multivariate stepwise regression analysis. Soil moisture estimation model was established by combining soil moisture data measured on the ground with field hyperspectral data. The relationship between hyperspectral data and soil moisture was established by obtaining the key point soil moisture value. The sensitive bands of soil moisture were selected. By using the multivariate linear regression method and the interpolated point data after "spatial expansion", a model of the relationship between hyperspectral data and soil moisture was established, and a large area of soil moisture was obtained.
【学位授予单位】：山东农业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：S152.7

【参考文献】