中文摘要：

通过传统土壤类型图所得的土壤属性图已不能满足精准农业和生态环境模型所需土壤属性的精度，而目前应用较多的统计方法和地统计方法均存在一定的局限性。鉴于此，本文探索了一种采用模糊聚类获取模糊隶属度进行土壤属性制图的方法。首先，采用模糊C均值聚类（Fuzzy c—means clustering，FCM）方法对环境因子进行聚类，通过野外采样（称为建模点）建立土壤-环境关系知识；然后，计算区域内各像元点对土壤类型的模糊隶属度；最后，对模糊隶属度采用加权平均的方法获取土壤属性值。将该方法应用于黑龙江鹤山农场老莱河流域的研究小区，以土体厚度和表层有机质为例进行土壤属性制图。为了评价该方法的有效性，将其与采用环境因子所建立的多元线性回归模型进行比较，通过野外验证点集评价两种模型所得的土壤属性，评价指标为观测值和预测值的相关系数、平均绝对误差（MAE）、均方根误差（RMSE）和准确度（AC）。结果表明，尽管通过建模点建立的多元线性回归方程砰较大，但该方程并不适用于研究区内的其他样本点，这表明多元线性回归方法在该区具有一定的局限性。与之相比，模糊隶属度加权平均的方法则可以通过较少的建模点得到更好的预测效果。

英文摘要：

Use of fuzzy membership values obtained with the fuzzy c-means clustering （FCM） method was explored to predict soil properties over space. First, environmental factors were fuzzy-clustered. Then, soil samples were collected at modeling points in the fields to determine soil-environment relationship and fuzzy membership of each pixel point to soil type in the study region was calculated. Finally, the weighted average model was applied to fuzzy membership, thus acquiring soil properties. To evaluate effectiveness of this method, it was compared with the multiple linear regression model based on environmental factors in soil property and terrain attributes. Four indices were set up for evaluation of the performance of these two models, i.e. correlation coefficient between predicted and observed values, mean absolute error （MAE）, root mean square （RMSE） and agreement coefficient （AC）. To validate the method, it was applied to a study area, in the Laolai River Valley, Heshan Farm of Nenjiang County in Heilongjiang Province of China. Two soil properties were chosen, i. e. A-horizon organic matter and soil thickness. Results show that the fuzzy membership weighted method produced reasonably better performance than the regression model by using less modeling points, while the linear regression model demonstrated its limitations in the study area. Although R^2 of the regression equations based on modeling points was high, the equation does not fit other sampling points of the area. It is, therefore, concluded that the weighted average method using fuzzy membership was an efficient way to predict soil properties, and it is more extendable than the regression approach.