Abstract:

Landslide triggered by Wenchuan earthquake was interpreted based on multisource remote sensing data. A logistic regression model was built between the landslide distribution and the control factors of the landslide. The number of landslide area interpreted was 5145, which covered a total area of 1139 km2. By using spatial analysis technology, it is found that (1) there are two types of distribution patterns existing in the landslide, one is a linear distribution along seismogenic fault and another is clusterlike around epicenter of Wenchuan earthquake; that (2) in the Ⅹ and Ⅺ degree seismic intensity zones, there are 73.2% landslide area and in the other region with Ⅷ degree or Less seismic intensity, there are few landslides; that (3) the distribution of landslide was not only controlled by seismogenic fault rupture, but also affected by the topographic amplification and faults upper and lower plate; there is 90% landslide area located in the upper plate of seismogenic fault. Furthermore, the logistic regression model indicated that (1) the seismogenic fault was the dominant factors controlling landslide distribution and the models partial regression coefficient of the factor of distance to BeichuanYingxiu fault was larger than that of any other factors in the logistic regression model; that (2) the factors of rock hardness are not significant, which indicated that rock hardness did not affect the distribution of landslide triggered by earthquake in our study area; that (3) in our model, the factors of elevation, terrain slope, slope change rate, the accumulated rainfall, human activity and the vegetation coverage all affect the landslide spatial distribution.

Key words:  Wenchuan earthquake, remote sensing interpretation, landslide, spatial distribution, logistic regression