中文摘要：

利用已知类型的土壤样点及其所处位置的高程、坡度、平面曲率、剖面曲率、复合地形指数等数据,基于Fisher判别分析方法对安徽宣城样区的土壤类型进行预测和制图表达。结果表明,土纲级别的预测效果较好,正确率达到84.2%。但随着从土纲到亚类级别的降低,由于受土壤样点数量限制,土壤类型预测的准确率也逐步降低。通过与样区1986年基于发生分类的土壤图进行对比,采用的方法无论是在制图精度,还是图面信息的负载量方面都要优于传统方法,能够更加客观真实地反映土壤在自然界的空间分布。

英文摘要：

Based on Fisher discriminant analysis（FDA） method,environmental data containing elevation,slope gradient,profile curvature,plan curvature and compound terrain index was utilized to predict soil types in Xuancheng area of Anhui province.The results show that prediction accuracy of soil orders is 84%,but that accuracy decreases with the soil type-down to subgroup level-because of quantity limitation of the soil sample.Compared to conventional soil mapping,digital mapping of soil by the FDA method is superior in terms of accuracy,information capacity etc.,and can represent soil spatial distribution in the real world more objectively.