基于空间回归模型的土壤有机质区域分布特征研究

发布时间：2018-05-09 01:33

本文选题：土壤有机质 + 环境因子　；参考：《华中农业大学》2016年硕士论文

【摘要】：平原丘陵过渡带属于生态交错带,其土壤有机质在陆地生态系统中的积累变化对土壤生产力及全球气候变化的影响日益受到关注。研究土壤有机质的空间分布特征,对于获取精确的土壤养分信息,理解陆地生态系统碳收支平衡具有重要的理论和现实意义。土壤有机质具有较强的空间依赖性和异质性,其空间格局同时受到气候、地形、成土母质、土壤类型、土地利用等诸多环境因素的影响,运用空间回归模型进行土壤有机质空间分布特征的研究是促进资源环境信息定量化、自动化管理的重要途径。为提高土壤有机质空间分布预测的精度,本文以土壤-景观模型的相关方法和理论为基础,以空间统计及地统计模型为工具,研究了平原丘陵过渡带土壤有机质的空间特征,预测了土壤有机质的空间分布,明确了土壤类型和土地利用程度的作用范围,主要结果如下:1.分析了平缓地区和丘陵山地及过渡条带等局部土壤有机质的空间聚集性,土壤有机质在平原丘陵间的核心过渡带上具有显著的高-低‖、低-高‖聚集现象,能表征环境梯度的转换特征。比较了空间自回归模型和地理加权回归模型预测土壤有机质空间分布的特点和能力,两者的残差均存在较弱的空间自相关,但地理加权回归模型的残差更小,地理加权回归模型残差在800m处已基本不存在空间自相关,而空间自回归模型残差至1km处仍有微弱自相关,所有模型的残差均在5km处无自相关。2.以地形因子为辅助变量,构建了回归克里格和地理加权回归克里格模型。(1)全域范围内土壤有机质变化范围为3.80~69.40g/kg,变异系数为39.59%,属中等程度变异。(2)回归和地理加权回归克里格模型的插值精度分别较普通克里格模型提高25.84%和27.61%,两者估计的土壤有机质的变化范围均小于实测范围,平原丘陵间的核心过渡带有明显的有机质积累。(3)在残差存在空间自相关的情况下,地理加权回归克里格模型生成的土壤有机质空间分布图斑更加破碎,细节更加丰富。3.定量化了土壤类型间差异信息和土地利用程度信息,土壤有机质与两者均存在显著相关性,土壤有机质在不同土壤类型中具有不同的变异程度,土地利用程度在平缓地区具有显著的高-高‖聚集现象。通过构建地理加权回归模型,明确了土壤类型和土地利用程度作为土壤有机质空间变异主导因子的作用范围,土壤类型在全局范围内具有主导性,土地利用程度在平缓地区具有较高主导性。
[Abstract]:The transition zone of plain hills belongs to ecological ecotone. The influence of soil organic matter accumulation in terrestrial ecosystem on soil productivity and global climate change has been paid more and more attention. It is of great theoretical and practical significance to study the spatial distribution characteristics of soil organic matter in order to obtain accurate soil nutrient information and to understand the balance of carbon budget in terrestrial ecosystem. Soil organic matter has strong spatial dependence and heterogeneity. Its spatial pattern is affected by many environmental factors such as climate, topography, soil-forming parent material, soil type, land use and so on. The study of spatial distribution of soil organic matter by using spatial regression model is an important way to promote the quantification and automatic management of resources and environment information. In order to improve the accuracy of soil organic matter spatial distribution prediction, based on the relevant methods and theories of soil-landscape model, the spatial characteristics of soil organic matter in the transitional zone of plain hilly region were studied by using spatial statistics and geostatistical models. The spatial distribution of soil organic matter was predicted, and the range of soil type and land use degree was determined. The main results were as follows: 1. The spatial accumulation of soil organic matter in flat region, hilly area and transition zone is analyzed. The accumulation of soil organic matter in the core transition zone between plain hills and hills is significant. It can characterize the transformation characteristics of environmental gradient. The characteristics and abilities of spatial autoregressive model and geo-weighted regression model in predicting spatial distribution of soil organic matter were compared. The residuals of the two models were both weakly spatial autocorrelation, but the residuals of geo-weighted regression models were smaller. There is no spatial autocorrelation at 800m, but there is still a weak autocorrelation between spatial autoregressive model residuals and 1km. The residual errors of all models are not autocorrelation at 5km. 2. Taking terrain factors as auxiliary variables, The regression Kriging model and geo-weighted regression Kriging model were constructed. The range of soil organic matter variation was 3.80 ~ 69.40 g / kg and the coefficient of variation was 39.59 g / kg. The interpolation accuracy of regression and geo-weighted regression Kriging models were obtained. Compared with the ordinary Kriging model, the variation range of soil organic matter estimated by both methods was 25.84% and 27.61% higher than that of the conventional Kriging model, respectively. In the case of spatial autocorrelation of residual, the spatial distribution of soil organic matter generated by geographical weighted regression Kriging model is more fragmented, and the details are more abundant. 3. The difference information and land use degree information among soil types were quantified. Soil organic matter had significant correlation with both, and soil organic matter had different degree of variation in different soil types. The land use degree has the remarkable phenomenon of high-high-high-high concentration in the flat area. Through the construction of geographical weighted regression model, the range of soil type and land use degree as the dominant factor of spatial variation of soil organic matter was defined, and the soil type was dominant in the global scope. The land use degree has the higher leading position in the flat area.
【学位授予单位】：华中农业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：S153.621

【参考文献】