Abstract:  On the basis of the Mahalanobis distance discriminant theory and in combination with the project practice, a distance discriminant analysis model of hazard degree assessment of a typical landslide was established according to the typical landslide strength indexes and the possibility indexes of the landslide, in which nine parameters were selected as the discrimination factor. The parameters are: landslide size, gradient, difference of height, slope structure, stratigraphic relations, slope shape, rubble content, stratum and lithology. Eighteen typical landslide examples in Fengjie County were used for training and verification. The back substitution method was introduced to verify the stability of the distance discriminant analysis model and the ratio of mistake-discrimination was equal to zero after the distance discriminant analysis model was trained. By means of the model, other 6 groups of measured data were tested as forecast samples, and the predicted results are consistent with the measured data. The study result shows that the distance discriminant analysis model has a higher accuracy and a low mis-discrimination ratio. It is a new approach to predict the hazard degree of a typical landslide, which can be used in practical engineering.

Key words: typical landslide , hazard degree assessment,  , distance discriminant analysis, prediction