Liu Ling, Wang Hai-Yan, Dai Wei, Yang Xiao-Iuan, Li Xu
Ying Yong Sheng Tai Xue Bao. 2014 Sep;25(9):2460-8.
Soil samples were collected in Jincang Forest Farm, Wangqing Forestry Bureau to study spatial distribution of soil organic carbon (SOC) and nutrients. Geostatistics was used to predict their spatial distribution in the study area, and the prediction results were interpolated using regression-kriging and ordinary kriging. Multiple linear regression was used to study the relationship between SOC and spatial factors. The results showed the SOC density (SOCD) at 0-60 cm was (16.14 ± 4.58) kg · m(-2). Soil organic carbon decreased significantly with the soil depth. With the increasing soil depth, total N, total P, total K, available P and readily available K concentrations decreased. Stepwise regression analysis showed that SOC had good correlation with elevation and cosine of aspect, with the determination coefficient of 0.34 and 0.39, respectively (P < 0.01). Soil organic carbon at 0-20 cm and 0-60 cm soil layers conformed to Gaussian model and exponential model. Compared with ordinary kriging, the prediction accuracy was improved by 18%-58% using regression-kriging. Regression-kriging interpolation was also applied to study spatial heterogeneity of soil total N.
在汪清林业局金仓林场采集土壤样本，以研究土壤有机碳（SOC）和养分的空间分布。利用地统计学方法预测研究区域内它们的空间分布，并采用回归克里格法和普通克里格法对预测结果进行插值。运用多元线性回归研究SOC与空间因素之间的关系。结果表明，0 - 60 cm深度处的土壤有机碳密度（SOCD）为（16.14 ± 4.58）kg·m⁻²。土壤有机碳含量随土壤深度显著降低。随着土壤深度增加，全氮、全磷、全钾、有效磷和速效钾含量均降低。逐步回归分析表明，SOC与海拔和坡向余弦具有良好的相关性，决定系数分别为0.34和0.39（P < 0.01）。0 - 20 cm和0 - 60 cm土层的土壤有机碳符合高斯模型和指数模型。与普通克里格法相比，回归克里格法的预测精度提高了18% - 58%。回归克里格插值法还被用于研究土壤全氮的空间异质性。
