Search

Zhongshan County of Guangxi Zhuang Autonomous Region was selected as the study area to investigate the intelligent assessment and early warning system of mountain flood geological disaster. Remote sensing images, spectral data and DEM data were processed on ENVI and ArcGIS platforms and the quantized data including slope, NDVI, soil looseness coefficient, valley and ridge classification and rainfall were obtained. And then a generalized regression neural network model for risk assessment of mountain flood geological disaster in Zhongshan County was established with the above quantized data as the input factors and the risk degree of the mountain flood geological disaster as the output factor. The trained model by using historical data has an excellent self-learning function and provide a good prediction on the risk degree of the mountain flood geological disaster in Zhongshan County.

On the basis of optimizing the frequency of ground-penetrating radar antenna, window length and sampling interval, a two-dimensional finite-difference-time-domain (FDTD) model for solving the Maxwell discrete equation for the electromagnetic wave is established. When there is one piping leakage point in the homogeneous soil, the piping size interpreted by the radar wave image is similar to the designed size. When there are two piping leakage points, the interference phenomenon of two reflected waves can be clearly simulated and the location and size of the two leakage points can be determined. A double horizontal layer soil is considered, and a piping point is located in the bottom soil layer, the piping position and size can still be accurately interpreted. The numerical test results show that the FDTD method for forward modeling is a very effective simulation technology and method for the identifying soil piping using ground-penetrating radar electromagnetic wave.