茅坪滑坡是清江隔河岩库岸规模最大的滑坡，具有稳定性差、崩滑方量大等特点。在地表位移监测资料分析的基础上，选用逐步回归分析法建立了滑坡体位移模型。逐步回归法是将ｍ 个自变量先做为候选变量，计算出各自的贡献（ 偏回归平方和） 与方差比Ｆ，再与给定等级检验临界值Ｆα作比较，若Ｆ≥Ｆα，自变量被引入；若Ｆ＜ Ｆα，自变量将被剔除。如此计算、比较，使模型中的自变量减少到最合理的程度。此法与经典多元线性回归方法相比，不仅简化了计算过程，消除了计算误差，且能对所建模型进行修正。模型计算成果表明，茅坪滑坡体主要受库水位涨落与大气降水作用的影响，具有中后部推移和前部牵引下滑的变形特征，潜在危险性大 The Maoping landslide is the largest landslide on the Geheyan bank of the Qingjiang River. It has the characteristics of poor stability and large amount of landslides and landslides. Based on the analysis of surface displacement monitoring data, the landslide body displacement model was established by stepwise regression analysis. The stepwise regression method is to use m independent variables as candidate variables first, to calculate their respective contributions (partial regression square sum) and variance ratio F, and then compared with the given level test critical value Fα. If F≥Fα, independent variables Is introduced; if F