Analysis And Research On Spatial Characteristics Of Landslide Associated With Debris Flow Based On UAV

In Dongchuan District of Kunming City,Yunnan Province,mud-rock flows and landslide disasters frequently occur under the action of earthquakes,heavy rains and human factors.Therefore,rapid and accurate detection of landslides in this area and analysis of their characteristics are extremely important for disaster prevention and mitigation.As a research area,Nihe has the advantages of low cost,efficient and convenient data acquisition and safe operation due to its terrain restrictions and the particularity of disasters.It has the advantages of low cost,efficient and convenient data acquisition,and safe operation.At the same time,it is compared with aerospace remote sensing and high-altitude remote sensing.Low-altitude UAV remote sensing images have high resolution and rich textures,and can more effectively use the characteristics of their three-dimensional models,making them more suitable for landslide detection.This paper uses low-altitude UAV images to obtain the DEM with an elevation point accuracy of 0.315 m,a DOM with a plane point accuracy of 0.334 m,a threedimensional model,and related topographic factors in the test area,so as to detect the landslide of the Dabaini River in Dongchuan District.And three-dimensional feature analysis.The main results of this paper:(1)Based on the combination of Dabaini River DOM image and other topographic factor data,the optimal landslide segmentation data combination in the test area image is DOM,DEM,and slope data.The optimal landslide segmentation scale is 475 and 0.5 tight.Consistency and a shape index of 0.3.The landslide rule set is obtained from the segmented 3850 image objects,and the spectral characteristics and shape characteristic threshold values suitable for identifying the landslide are obtained.(2)The traditional method of landslide identification is mainly based on manual field investigation and visual interpretation of images.Although this method has high identification accuracy,the cost is too high and the cycle is too long.The article uses the nearest neighbor classification,decision tree,and fuzzy classification to identify the landslides in the Dabaini River imagery,and divides the landslides into soil flow landslides and mud flow landslides.After accuracy verification,the Kappa coefficients of the three classification methods are 0.74.,0.82,0.86,it takes about 2min,5min,1h,and the fuzzy classification method has the highest accuracy,but the decision tree classification is more suitable for landslide identification.(3)On the identified disaster locations of 6 mud flow landslides and 5 earth flow landslides,superimpose the 5 influencing factors of slope,aspect,roughness,undulation,and elevation to obtain the landslide prone index of the place,two types Landslides are prone to occur with a slope of 0?10 and a roughness of 1?1.5;earth-flow landslides are prone to slopes of 30?45°,with a slope of 67.5?157.5°;mudflow landslides are prone to slopes of 0?30°,slope The direction is 292.5?337.5°.(4)Perform statistical analysis on the back wall of the landslide and the landslide body for earth flow landslides,and perform statistical analysis on the formation area,circulation area and accumulation area of mud flow landslides.Mainly include the elevation change,arc length and radian value of the back wall of the landslide;the area,circumference,and aspect ratio of the landslide body;the area of the formation area;the change of the length,width,and elevation of the circulation area;the area and thickness of the accumulation area.Finally,through the formation and movement characteristics of the landslide,the unique rain erosion-landslide-debris flow disaster link in the study area is obtained.