中文摘要：

研究了基于环境因子和混合插值的林地土壤有机质预测模型。首先应用数字地形与遥感影像分析技术获取地形因子与遥感指数,然后分析土壤有机质与环境因子的相关性,最后用环境因子对土壤有机质进行空间预测。针对回归克里格法（RK）需要计算半变异函数的缺陷,提出了一种空间插值方法,即回归-光滑薄板样条插值法（R-STPS）。将这2种插值方法用于顺昌县土壤有机质的空间预测。结果表明,RK与R-STPS的预测精度、计算效率、预测的研究区土壤有机质空间分布的总体趋势相近。R-STPS无需计算半变异函数,使用方便,因此更有优势。

英文摘要：

The spatial prediction model of forest soil organic matter was studied based on environmental factors and mixed interpolation methods. Firstly, digital terrain and remote sensing image analysis technologies were applied to get topographic factors and index of remote sensing. Then, the correlation of soil organic matter and environmental factors was analyzed. In the end, soil organic matter was predicted spatially according to the environmental factors. Aiming at the flaw of regression Kriging （RK） which needs to compute semi-variogram, a spatial interpolation method named regression-smoothing thin plate spline （R-STPS） was presented. This two interpolation methods were applied to predict soil organic matter of Shunchang county spatially. The results showed that the prediction accuracy and computation efficiency of RK and R-STPS were almost consistent. The overall trend of spatial prediction distribution of soil organic matter of study area was similar. However, R-STPS was not needed for calculation of semivariogram and easy to use. Therefore, R-STPS has more advantages.