Study on susceptibility of karst collapse based on normal cloud model in Yonghe town, Liuyang City

Ground collapse is a common geological disaster in karst area, which would bring great harm to people's life and property. The zoning evaluation on karst collapse susceptibility is beneficial to the classification and treatment of the disaster in order to ensure safety and economy. Karst ground collapse is characterized by suddenness, concealment, multi-factor, randomness and fuzziness; therefore, it is difficult to be fully quantified. The normal cloud model could effectively reflect the fuzziness and randomness of objective things, and integrate them to form the mapping between qualitative and quantitative analyses. In this paper, based on the normal cloud model, the study on susceptibility of karst collapse in Yonghe town, Liuyang City has been conducted in order to provide a basis for the classification and treatment of ground collapse in this area, and also provide a reference for the susceptibility evaluation of karst collapse in other areas.

The study area is located to the northeast of Liuyang City, 24 km away from it, and its administrative division is located in Yonghe town, Liuyang City, Hunan Province. Ground collapses in this area are all developed in the distribution area of soluble rock layer. The most developed layer is the second member of the Lower Permian Qixia Formation (P₁q²) with thick-layered carbonate rocks. Secondarily, ground collapses are distributed in the middle and upper Carboniferous Hutian Group (C₂₊₃ht). The study area covers an area of 14 km², and 38 karst collapses have occurred in this area so far. Among these collapses, 22 have been surveyed and filled. Sixteen collapses, four of which have been filled, have been investigated in detail. The collapse sites are mainly distributed in the region of soluble rock where faults are developed and surface water and groundwater are closely connected. Geographically, the collapse sites are mainly distributed in Yueshan Formation-Oujia Formation-Dahe Formation, Yongfu village, Yonghe town, Juxiang community-the old street of Yonghe-Huayuan village, Yanxi town, Nanshan Formation-Lizhen Primary School (old)-Xinwan Formation-Xinping Formation.

Six evaluation indexes of karst collapse susceptibility were selected in this study, including karst development degree (dissolution rate of borehole), distance from the fault in the area, thickness of overlying soil layer, characteristics of karst water, distance from the pumping funnel center and current situation of ground collapse (ground collapse density). Firstly, according to the detailed exploration results of existing subsidence pits, the weights of six evaluation indicators of 16 existing subsidence pits were calculated and assigned based on entropy weight method. The calculation results indicate that the weight of development density of existing ground collapses is the largest, and that of the fluctuation amplitude of groundwater is the smallest. Secondly, the scoring standard for the risk level of normal cloud model was determined, and the susceptibility of karst collapse in each unit area in the study area was evaluated combined with the weight of each evaluation index. The evaluation results show that all the karst collapses that occurred are distributed in the area highly subject to karst collapse, indicating that the zoning in the evaluation of this study is reasonable.

When the normal cloud model is used to evaluate the susceptibility of karst collapse, the size of grid cells could be controlled by the cloud similarity of comprehensive risk evaluation of karst collapse. For areas with high similarity and small cloud droplet dispersion, the unit area of each evaluation could be appropriately increased in order to reduce the evaluation workload. In addition, the high risk area calculated by analytic hierarchy process is 10.8% larger than that by normal cloud method, but three existing collapse pits distributed in the area at medium-level risk of karst collapse, indicating that normal cloud model takes more advantages than analytic hierarchy process in dealing with fuzzy and random problems such as the risk evaluation of karst collapse.