摘要土壤属性高精度建模可被定义为是一种通过数值模型从土壤各方面观测值以及 附近的环境信息来预测土壤属性的空间变异情况，从而获取土壤属性空间信息的模 型。这种模型有利于我们研究土壤属性，并且我们可以将这种研究应用于土壤的可持 续利用，同时为农业发展，环境保护和社会进步提供有效保障。因此，我们需要不断 提高土壤属性建模的精度。本文针对复杂地貌类型区土壤属性，建立新的高精度自适 应模型。并且对土壤属性进行空间分异分析。70797

本文利用青海湖流域 111 个样本点数据，在 GIS 技术支持下，先用反距离权重法( IDW)、克立格插值法( Kriging)、径向基函数插值 （RBF)、局部多项式插值法[1（]LPI）

和基于地学信息的插值方法[3]对样本点进行空间插值。对预测结果进行误差分析，对 比平均值误差、平均绝对误差和均方根误差后精度分析，查找原因。

对于传统土壤属性精度建模来说，本文选择应用方差分析模型选取与土壤属性空 间分异密切相关的地学环境要素，构建一系列地学环境信息支持的土壤属性插值模 型，作为集成学习的基插值模型。针对土壤空间的不连续性和空间异质性，探索以构 建误差曲面的方式，实现插值曲面的自适应分区，以筛选合适的插值模型；对筛选的 插值模型进行优化组合，构建一个可对不同插值曲面进行自适应分区与集成的土壤属 性插值曲面模型(不同区域选取不同插值模型组合)，实现复杂地貌类型区土壤属性的 高精度模拟。以期解决通常全局模型所不能解释的空间变量之间存在的空间非稳定性 问题。

对于插值出来的空间分布差异，主要体现为土壤对于不同属性的表现不同，如钾 含量因土地利用类型不同，分布状况也不同。

该论文有图 8 幅，表 3 个，参考文献 20 篇。 

毕业论文关键词：空间插值 自适应 土壤属性高精度建模 空间分异

Complex Landform Types Adaptive Modeling and Spatial Differentiation Analysis of Soil Properties

Abstract Soil properties and high precision modeling can be defined as one model which is based on the analysis of soil observations by the numerical model as well as the nearby environment information to predict the spatial variation of soil properties so that to get the model of soil attribute space information.This model is helpful to our study of soil properties, and this kind of research can be applied to the sustainable utilization of the soil, at the same time for agricultural development, environmental protection and social progress to provide effective protection.Therefore, we need to constantly improve the soil properties of modeling accuracy.This article in view of the soil properties of areas that is full of complex landform types to establish a new high precision adaptive model.And the analysis of spatial variation was carried out on the soil properties.

In this paper, using the data from the qinghai lake basin in 111 sample points under the GIS support, we start to spatial interpolation of sample points by four methods such as the inverse distance weighting method (IDW), Radial Basis Functions method (RBF), local polynomial interpolation method (LPI) and the interpolation method based on the geological information. For the predicted results we should analyse the error.At the same time comparing the average error, mean absolute error and root mean square error after precision analysis to find the reasons.

For traditional soil properties precision modeling, this paper chose using analysis of variance model and soil properties is closely related to the spatial differentiation of geological environment factors, build a series of soil attribute interpolation model of geological environment information support, as the integrated study of the interpolation model;In view of the discontinuity and the spatial heterogeneity of soil space, exploring the way to build error of surface, the adaptive partition interpolation curved surface, by screening suitable interpolation model.Optimize the screening of the interpolation model of combination, to build an adaptive partition and integration of different interpolation surface