论文部分内容阅读
用传统统计学方法探讨土壤水分空间分布的前提条件是用于研究的数据在统计上必须独立且均匀分布,但土壤水分在空间上一般存在空间自相关,这些空间自相关包含一些有用信息。本文利用Moran I系数研究了祁连山北坡甘肃臭草(Melica przewalskyi)单优种群斑块浅层剖面(0~30 cm)土壤水分空间自相关关系、空间相关尺度,建立了各层土壤水分影响因子的线性回归模型和空间自回归模型,并比较了2种模型的分析结果。结果表明:各层土壤水分均具有空间正相关性与空间集聚特征,10~20 cm层土壤水分的空间集聚特征较0~10 cm和20~30 cm土层更为明显;土壤水分的Moran I系数随着分析间隔距离的增大而减小,在<4 m的范围内各层土壤水分均存在正的空间自相关关系。影响甘肃臭草浅层剖面土壤水分空间分布的因素在不同土层不尽相同;空间自回归模型的LIK值和R2的值比线性回归模型的值要大,从而显示出空间自回归模型的解释能力要优于经典线性回归模型。
The traditional statistical methods to explore the spatial distribution of soil moisture prerequisite is that the data used for the study must be statistically independent and uniform distribution, but there is generally spatial spatial autocorrelation of soil moisture, the spatial autocorrelation contains some useful information. In this paper, the spatial autocorrelation and spatial correlation of soil moisture in the shallow section (0-30 cm) of single dominant population of Melica przewalskyi in the northern slope of Qilian Mountains were studied using the Moran I coefficient. The soil water content Linear regression model and spatial autoregressive model, and compared the results of two models. The results showed that the spatial distribution of soil moisture in 10 ~ 20 cm layer was more obvious than that in 0 ~ 10 cm and 20 ~ 30 cm soil layers. Soil Moran I The coefficient decreases with the increase of the analytical interval distance. There is a positive spatial autocorrelation of soil moisture in all layers within the range of <4 m. The factors influencing the spatial distribution of soil moisture in stinkbloom shallow section of Gansu were different in different soil layers. The value of LIK and R2 in the spatial autoregressive model was larger than that of the linear regression model, which showed that the spatial autoregressive model was explained Ability is better than the classical linear regression model.