The Identification Of Gully Features Of Loess Plateau Based On CTI

As an important geomorphological type of loess plateau,the valley system is of a high research value because of its active material energy exchange characteristics,variable morphological characteristics and severe potential risks to the downstream region.Compound topographic index(CTI)is a model used to predict the trajectories and initiation points of gullies.It is sensitive to multiple soil factors and easy to calculate.Besides,the predicted result almost coincides exactly with the actual location of the gully.Identifying the gully-heads and trajectories of the gully system based on CTI is of great significance for grasping the geomorphological development characteristics of the study area.In this study,the loess tested small watersheds and four realistic geomorphic types,loess Tab.land region(Chunhua),loess ridge region(Ganquan),wind drift sand region(Shenmu)and loess hillock region(Suide)are selected as the sample areas.For four realistic sample regions,the gully-heads and the trajectories are identified based on CTI model,and the CTI threshold and the most appropriate DEM resolution are discussed respectively.And for the loess tested small watersheds,the gully-heads are identified and the developmental stages are classified.In addition,the optimal resolution for each studies is discussed as well.The main research content of this paper and the conclusions are as following.1.Different CTI models are applicable to sample areas with different resolution levels..For the four geomorphic regions with a resolution of 5m,the CTI_a model is benefit to predict the gully-heads and trajectories of the gully,and contributes to discussing the appropriate resolution.For the loess tested small watersheds with a resolution of 10mm,the CTI_b model shows a more effective ability for the identification of gully-heads,the classification of the geomorphologic evolution and the discuss of the optimal resolution.2.The recognition rates of gully-heads and gully trajectories and thresholds of the CTI model are calculated in the realistic geographic sample area.In this study,the value of CTI_a is used to extract the gully trajectories and gully-head according to different classification methods.The results show that the threshold of CTIa in Chunhua is from 12.15 to 22.97,and the theoretical optimal recognition rate is about77%.For the Ganquan area,the threshold of CTI_a is from 10.45 to 20.32,and the theoretical optimal recognition rate is about 50%.For the Shenmu area,The threshold of CTI_a is from 11.09 to 19.17,and the theoretical optimal recognition rate reaches 59%.For the Suide area,the threshold of CTIa is from 9.84 to 18.66,and the theoretical optimal recognition rate is over 63%.In addition,the factors affecting the identification rate of the gully-heads and gully trajectories in the area can be concluded to four types:landform type,slope variability,the density of gully-head and gully trajectories density.And the landform type plays an important role for the effect the recognition rate.3.The optimal resolution of using CTI model to identify the gully trajectories in the actual geographic sample region.6 groups resolution data(5m-30m)are obtained by resampling the DEM of four realistic geomorphic regions,which are Chunhua,Ganquan,Shenmu and Suide.By analyzing the differences of the real gully trajectories length,the extracted gully trajectories length and the loss rate of the gully trajectories under different DEM resolution,the optimal resolution is obtained.For the Chunhua area,the optimal resolution is 20m when using the CTI model to extract the gully trajectories.For the Ganquan area,the optimal resolution is 10m when using the CTI model to extract the gully trajectories.For Shenmu,when using the CTI model to identify the gully trajectories,the best resolution is15m.For Suide,the optimal resolution using the CTI model to identify the gully trajectories is 30m.The factors that affect the gully trajectories length loss rate in the area are the rate of landform cutting depth and the length ratio of principle gully and branch gully.4.Use the CTI model to identify the gully-heads and classify the developmental stages.For the loess tested small watersheds,CTI_b is obtained by taking logarithm processing of the CTI values.According to the characteristics of CTI_b in each watershed,it is possible to identify the gully-heads and divide the watershed developmental stages.The results show that using the CTI model to identify the gully-heads in the early and mid-term of the loess tested small watersheds can basically achieve an efficiency over 56%.By re-sampling the loess tested small watersheds,6 sets of data with a resolution of 10mm to 35mm are obtained.When using the CTI model to identify the gully head,the effect achieved by 20mm resolution is the best.And when the CTI model is used to divide the watershed stages,15mm can obtain a better result.