Spatial And Temporal Distribution Characteristics Of Giant Landslides In The Upper Yellow River Based On Surface Roughness Analysis

The area from Laganxia to Sigouxia in the upper reaches of the Yellow River,located in the northeastern part of the Tibetan Plateau,has a wide distribution,a long time span,a large number of landslides and a complex mechanism of genesis,which are representative of China and even the world.Many scholars have studied giant landslides in terms of distribution characteristics,developmental characteristics and clustering mechanisms,but the dating of landslides usually differs depending on the dating methods and sampling locations.The existing methods of landslide dating have certain limitations,and most of the landslide dating studies are limited to a single giant landslide,but fail to analyze multiple landslides from a larger area,which directly limits the scientific understanding of their cluster mechanism and affects the study of the main control factors of giant landslides in the upper Yellow River area.In recent years,landslide roughness has been successfully applied to chronological studies as high-precision topographic data can be obtained more efficiently.The commonly used ¹⁴C dating,Optically Stimulated Luminescence（OSL）dating,Terrestrial Cosmogenic Nuclides（TCN）dating and other methods are limited by the difficulty of sampling and the high cost of measurement,and the use of landslide roughness for chronological studies provides new ideas and methods for Many foreign scholars have discussed the applicability of this method and successfully predicted the age results of several landslides in the study area.However,there are still many problems restricting the development of chronological studies using landslide roughness,mainly including the applicable differences of different roughness algorithms,and the number of landslide samples.Based on previous studies on the chronology of giant landslides in the upper reaches of the Yellow River,this paper selects the middle section of the Qunke/Jianzha basin from Sigou Gorge to Lagan Gorge in the upper reaches of the Yellow River as the study area and obtains a digital elevation model（DEM）of the whole area with 5m resolution based on the Resource Three（ZY-3）satellite.The standard deviation of slope（SDS）,root mean squared height（RMSH）,2D Discrete Fourier Transform（2D DFT）,and 2D Continuous Wavelet Transform（CWT）were firstly applied.Continuous Wavelet Transform（2D CWT）algorithms were used to calculate the roughness of nine landslides in the study area,including Tangse Village,Kang Yang,Quhetan,and Xazang Tan,and analyzed in terms of frequency distribution curves,statistical eigenvalues,and response to landslide sequential relationships.After that,a total of 18 chronological samples were collected for these 9 landslides,and the dating of landslides was completed by Optically Stimulated Luminescence（OSL）dating technique.The results of landslide dating were compared with the results of roughness analysis,and finally the elevation mean square difference and two-dimensional Fourier transform algorithms were selected to construct the relationship between landslide surface roughness and chronology.The research results obtained in this paper are as follows:（1）From the comparison of the frequency distribution curves of the four roughness algorithms,it is found that most of the landslide curves of the slope standard deviation algorithm and the elevation mean square algorithm have a bimodal distribution,while most of the landslide curves of the two-dimensional Fourier transform and the two-dimensional wavelet transform algorithm have a single-peaked distribution.（2）From the comparison of the statistical eigenvalues of the four roughness algorithms,it is found that the difference in standard deviation and the difference in mean values among the landslides of the slope standard deviation algorithm are smaller than those of the other three algorithms,the difference in mean values among the landslides of the 2D wavelet transform algorithm is large,and the difference in mean values among the landslides of the elevation mean square and 2D Fourier transform algorithms is moderate.In terms of the mean values,the two-dimensional Fourier transform and two-dimensional wavelet transform algorithms have larger mean values than the other two algorithms.（3）The OSL dating results showed that the right bank landslide of Mananggoukou was dated around 40,600 a B.P.,the Kang Yang landslide was dated around 30,000-40,000 a B.P.,the Mananggoukou landslide was dated around 34,300 a B.P.,the Tangse Village landslide was dated around 30,000 a B.P.,and the Xiazangtan landslide was dated around 20,000-30,000 a B.P.,the age of the landslide in Quhetan left bank is about 23,900 a B.P.,the age of the landslide in Guanxiangkou is about 13,600 a B.P.,the age of the landslide in XaQiongsi is about 13,100 a B.P.,and the age of the landslide in Senguotan is about 7,000-8,000 a B.P.（4）The relationship between elevation mean square difference and twodimensional Fourier transform algorithm is more accurate.The landslide surface roughness and age relationship curve constructed by applying elevation mean square difference is t=272744×e^（-0.41807R）,R²=0.91105;the landslide surface roughness and age relationship curve constructed by applying two-dimensional Fourier transform algorithm is t=57841×e^（-0.09838R）,R²=0.90458.