Abstract: The existing analysis of the degree of karst development is mainly based on the general understanding of regional scale, but under the influence of topography, landform, groundwater runoff and other factors. The degree of karst development in the same geological unit varies and the regional general understanding of karst development often cannot meet the requirements of site-scale karst development degree analysis. A drill hole data is an important source for studying karst development characteristics at the engineering site. But various factors such as the core recovery rate and the large long steep dip-vertical karst fractures can influence the estimates for the karst development degree using the traditional karst statistical methods. By collecting drill holes data within the scope of the project and using the karst fractal theory, this paper combined the borehole image recognition system and the graphic fractal calculation program to determine the karst development degree discriminant index (IKAF), and proposed a site-scale, high-precision karst development degree identification method. The drill hole data of a reservoir site in the left bank mountainous area of the middle and upper reaches of the Shiting River basin in Mianzhu were analyzed using this method. The results show that this method can quickly and accurately identify the morphology of karst development in the drilling hole, reducing errors in determining the karst rate caused by the development of long and steep to vertical karst fractures. The degree of karst development obtained using this method is basically consistent with the degree of karst development reflected by the karst phenomena such as sinkholes and karst caves obtained from field investigations. This method is a beneficial addition to the traditional method in analyzing the degree of karst development. 

Key words: fractal theory；Drill Hole Image Recognition, fractal evaluation index, degree of karst development