Early Recognition Of Loess Landslide Based On UAV Photogrammetry

Loess is a kind of structural soil with large porosity,vertical joint development,weak cementation and water sensitivity.Loess can keep certain strength under the dry condition.But under the conditon of external disturbance,especially with the erosion of water,the mechanical properties will yield and soften,and then the strength will reduce,called strong collapsibility.The Loess landslide as one of the most common geological disasters,is widely distributed with high frequency and the formation of complex mechanisms.It is a serious threat to people's lives and property.Heifang Terrace is a typical representative of the loess landslide,known as the "natural loess landslide museum."Heifang Terrace belongs to the ? terrace of Yellow River.The average different height from top to bottom is 70 m.This Terrace was developed into farmland as Resettlement site for the construction of Yanguo Gorge Hydropower Station by means of flooding irrigation from Huanghe River.The landslide occurred soon afterwards.Now there are 75 sites of landslide and above 100 slides,including loess-bedrock slide,loess collapses,loess flow and loess static liquefaction.Among them,loess static liquefaction type has the characteristics of big size and long distance,which seriously threatens the local residents' safeties of life and property.In recent years,the more frequent areas of landslide for the Hei Terrace are Dangchuan section,Jiaojia section and Chenjia section,with an average value of 3-5 times a year.Although we can recognize the inevitability of the landslide from the historical landslides,it is difficulty to monitor the landslide on the precise position because of the wide area.If we can detect the area of deformation,it is possible to monitor the landslide and warn early,which can protect the local residents' productions and safeties.In order to realize the early deformation identification of the landslide on Heifang Terrace,this paper uses the UAV photogrammetry technology to carry out multi-phase measurements,and studies its precision and deformation recognition method.Firstly,the UAV flights are executed to obtain the images of the survey area,and then the images are processed to obtain the point clouds data,the 3D models data and the orthophoto maps.The accuracy of the UAV photogrammetry is increased from the decimeter level to the centimeter level by increasing the numbers and optimizing the distribution of GCPs.This paper also introduces the use of a new multi-period data registration method,which will register two-phase three-dimensional grid models,and then subtract the elevation data to obtain the differential model to display the deformation areas.Finally,the reliability of deformation identification by differential model is verified by field investigation,orthophoto map and ground monitoring data.We can conclude that It is very effective to use the UAV photogrammetry technique to identify the early deformation of landslide from the above study.Several cases show that the error of deformation detection is less than 20 cm.There still are some examples that the deformation areas were detected at the earlier time and then some collapses were happened in these areas.So we can identify the deformation areas as potential landslides.