中文摘要：

研究分析了典型崩岗泥沙源地的31种土壤理化性质,利用Kruska-Wall无参检验结合多元回归分析筛选指纹因子体系,通过多元混合模型得出侵蚀泥沙来源。结果表明：该典型崩岗筛选出的8种指纹因子累积正确判别率达到96.87%,可以组成指纹因子体系。多元混合模型结果表明崩积堆泥沙样品超过60%来源于红土层,但在冲积扇中的沉积泥沙仅有10%左右来源于红土层,其余90%的泥沙都来源于砂土层和碎屑层。这表明红土层有较强抗侵蚀能力,崩塌后会残留在崩积堆上。但由于红土层细颗粒含量较多,在二次侵蚀作用下红土层泥沙大都被径流带出崩岗。组合指纹法在崩岗中的成功应用,为高效、定量研究崩岗侵蚀泥沙来源提供可能。

英文摘要：

In this study,soil samples were collected from sediment sources in typical collapsing hill.After analyzed 31 soil characteristics,the composite fingerprint factors were selected by using Kruska-Wall test and multiple regression analysis.Then multivariate mixing model was applied to estimate the source of sediments.Results showed that 8 factors could be used as composite fingerprinting factors,and the rate of cumulate accuracy was 96.87%.Multivariate mixing model results showed that there were over 60% of sediments in slumping deposit（SD） coming from red soil layer（RSL）.However,there were only 10% of sediments in alluvial fan（AF） coming from RSL,other 90% of sediments in AF were coming from another two layers in collapsing hill.This meant RSL had high resistant erosion ability,it would hold back in SD after collapsed.However,most of the RSL’s sediments easily moved out by runoff,because most of RSL’s particle sizes were small.Composite fingerprinting technique offers a quantitative and efficient way to evaluate collapsing hill’s sediment sources.