Intelligent identification of landslides in loess areas based on the improved YOLO algorithm:a case study of loess landslides in Baoji City

Loess landslides are one of the geological hazards prevalent in mountainous areas of Loess Plateau,seriously threatening people's lives and property safety.Accurate identification of landslides is a prerequisite for reducing the risk of landslide hazards.Traditional landslide interpretation methods often have the disadvantage of being laborious and difficult to use on a large scale compared with the recently developed deep learning-based landslide detection methods.In this study,we propose an improved deep learning model,landslide detectionyou only look once(LD-YOLO),based on the existing you only look once(YOLO)model for the intelligent identification of old and new landslides in loess areas.Specifically,remote sensing images of landslides in Baoji City,Shaanxi Province,China are acquired from the Google Earth Engine platform.The landslide images of Baoji City(excluding Qianyang County)are used to establish a loess landslide dataset for training the model.The landslide data of Qianyang County is used to verify the detection performance of the model.The focal and efficient IoU(Focal-EIoU)loss function and efficient channel attention(ECA)mechanism are incorporated into the 7th version of YOLO(YOLOv7)model to construct the LD-YOLO model,which makes it more suitable for the landslide detection task.The experiments yielded an improved LD-YOLO model with average precision of 92.05%,precision of 92.31%,recall of 90.28%,and F1-score of 91.28%for loess landslide detection.The landslides in Qianyang County were divided into two test sets,new landslides and old landslides,which were used to test the detection performance of LD-YOLO for both types of landslides.The results show that LD-YOLO detects old landslides with a detection precision of 82.75%and a recall of 80%.When detecting new landslides,the detection precision is 94.29%and the recall is 91.67%.It indicates that our proposed LD-YOLO model has strong detection performance for both new and old landslides in loess areas.Through a proposed solution that can realize the accurate detection of landslides in loess areas,this paper provides a valuable reference for the application of deep learning methods in landslide identification.

Deformation Characteristics of Loess Landslide along the Contact between Loess and Neocene Red Mudstone

The loess landslide along the contact between loess and Neogene red mudstone (NRM) is one of those that have occurred extensively and frequently in loess areas of China. To better understand the mechanism of this type of landslides, a distressed loess slope being subjected to deformation along the loess-NRM contact was comprehensively investigated through approaches of field monitoring and laboratory physical modeling. Field observation and physical modeling shows that the slope deformation will experience two distinct processes: 1) laterally retrogressive and vertically progressive propagation, which was initiated by falling of the slope head; and 2) retrogressively separate mass sliding along the weak basal zone of the loess-NRM contact with minor sliding along the paleosols within the loess. Shear failure of the loess-NRM contact may initiate in the middle section, followed by a progressive propagation towards the slope toe and inner slope. Analysis reveals that the deformation characteristics of the distressed slope are largely constrained by slope topography, the unique structure, physical and mechanical properties of loess and paleosols, and occurrence and nature of the loess-NRM contact. Rainfall has significantly influence on the deformation characteristics of the slope through its interaction with the loess and soil of the loess-NRM contact. Additionally, improper style and intensity of cutting on the slope greatly enhance and accelerate the deformation course of the slope.

The Volume Calculation Method of Rock Collapses and Loess Landslides Triggered by the 1556 AD Huaxian M81/2 Earthquake in Shaanxi Province, China

Accurate volume calculation of each individual landslide triggered by strong historical earthquakes can help understand the characteristics of the typical earthquake-induced landslides,thus providing significant information for the modification of the focal parameters of historical earthquakes.In this study,we select one rock fall and three loess landslides triggered by the 1556 AD Huaxian M8⅟􀄟earthquake,compute their volumes using the low-altitude high-precision Unmanned Aerial Vehicle(UAV)photogrammetry and landslide profile restoration methods.The results show that:①the whole influencing area of the Huangjiagou Rock Fall is approximately 3.03×105 m2 and the area of the collapsed rock accumulated at the slope foot is 3.33×104 m2,accounting for approximately 10%of the entire influencing range.However,the estimated volume of the collapsed rock is only 0.699×106 m3,indicating a rock fall with large influencing range but limited collapsed rock;②the geological form of thethree loess landslides are preserved intactly,with volumes of 0.283×108 m3,0.074×108 m3,and 0.377×108 m3.These important geological hazard relics reflect the strong vibrations and severe casualties in the meizoseismal area;③loess landslides are the key reason of the serious death toll in the hilly-gully loess area.Our new method can be used to estimate the influencing area and the actual volume of each individual landslide,and rationally evaluate the role of earthquake landslides in the disaster.In addition,quantitative research on secondary disasters triggered by strong historical earthquakes is beneficial for understanding the surface process and focal parameters of the earthquakes.

A combined survey to evaluate the thermal behavior of loess for a landslide-prone slope on the Heifangtai terrace in Northwest China

A combined survey including infrared thermography(IRT)and field-laboratory tests were conducted to analyze the thermal responses and thermal properties of loess on a landslide-prone loess slope in the Heifangtai terrace in Northwest China aiming at preliminarily demonstrating the potential of IRT as a complementary technique to the investigation of irrigation-induced loess landslides.Multitemporal thermographic surveys corresponding to various solar radiation intensities during the afternoon were carried out on the landslide-prone loess slope.Accordingly,the spatiotemporal distribution of the thermal responses within the observed slope surface was analyzed qualitatively and quantitatively.Meanwhile,field and laboratory investigations were also performed on the thermal properties of different landslide materials.The results indicate that loess,a landslide-prone deposit that usually has a relatively high water content,exhibits different thermal properties and anomalies,including a lower surface temperature and greater thermal inertia,compared to surrounding zones without landslides.The groundwater table and corresponding seepage line could also be obtained by determining the potential boundary between the thermal response distribution of landslide scarps and that of saturated deposits in the presence of landslides.The results of these investigations are expected to provide insight for future endeavors combining infrared thermography with other efficient survey methodologies(e.g.,InSAR,which can monitor the active displacement of a loess slope)to evaluate the activity of this kind of excessive irrigation-induced loess landslide.

Vibration-induced acceleration of infiltration in loess

Coupled effects of railway vibration, rainfall, and drying-wetting cycles contribute to new types of loess disasters such as the collapse and vibration-induced settlement of the railway subgrade, loess mass disintegration, and slope sliding,during the large-scale construction and operation of high-speed or heavy-haul railways in China. This has significant consequences for railways that require millimetre-level deformation control, seriously threatening railway operation and people's lives. Therefore, for the first time, "five vibration-induced effects" on loess subjected to vibration, water immersion, and drywetting cycles are discussed including the vibration-induced acceleration of cracking, infiltration, disintegration, sliding, and subsidence. In this paper, the vibration-induced acceleration of infiltration is discussed mainly. The response of loess to locomotive vibration along railways was determined using field sampling and data processing. The dominant frequency of locomotive vibration are 18-22 Hz and the maximum amplitude is 0.4 mm. Tests regarding the vibration-induced acceleration of the permeability were conducted on intact loess using a newly developed vibration permeability apparatus. The results reveal that the vibration accelerates the water permeability and increases the water saturation of loess. The permeability coefficient of saturated loess under vibration is ~1.5-20 times of that without vibration, reaching a maximum at a vibration frequency of 20 Hz.Vibration-induced infiltration tests were carried out on a loess column and the vibration-induced acceleration of water infiltration, wetting front evolution, and air outflow were analysed. The vibration-induced acceleration of infiltration strongly depends on the dry density, vibration frequency, and vibration amplitude. This pioneering work explores the mechnism of loess disasters triggered by locomotive vibration, rainfall, and drying-wetting cycles.