【正文】 em (GIS).Spatial data for the subsidence area, topography and geology and various ground engineering data were collected and used to make a factor raster database for a ground subsidence hazard map. To determine the importance of extracted subsidencerelated factors, frequency ratio model and sensitivity analysis were employed. Sensitivity analysis is a method for paring the bined effects of all factors except one. Sensitivity analysis and its verification showed that using all factors provided % accuracy. The best accuracy was achieved by not considering the groundwater depth (%) and the worst by not considering the lineament (%). The results show that the distance from the lineament and the distance from the drift highly affected the occurrence of ground subsidence, and the groundwater depth, land use and rock mass rating had the least effects. Thus, we determined causes of ground subsidence in the study area and this information could help in the prediction of ground subsidence in other areas. Keywords GIS Abandoned underground coal mine Ground subsidence Frequency ratio Sensitivity analysis Korea Introduction Coal industry has played a major role in the development of the Korean economy since the 1960s. However, the coal industry began to decline from the end of the 1980s because the oil price fell globally. There were 173 coal mines in 1988 in Jeongseon, Kangwondo, the largest coal mining region in Korea. However, valueless coal mines were abandoned after the implementation of coal industry rationalization action by Korean government. Today, no coal mines are in operation. In the abandoned coal mines in Jeongseon, environmental pollution, having never been considered a future problem at the time the mines ceased operating, is on the increase. Heavy metals in wastewater from coal mines and muck reaching rivers and soil is a serious problem. In particular, most abandoned coal mines are in mountain ranges and their damage to the environment is not widely known. Ground subsidence due to an underground coal mine is a hazard to human life as well as to property such as railways, roads, houses and buildings. In addition, it is difficult to pletely recover an area of subsidence and rehabilitation is expensive. Moreover, most countermeasures for ground subsidence are simply reinforcements after ground subsidence has already occurred. Therefore, it is necessary to have a systematic prediction and management plan for an area of ground subsidence. There have been many studies on the prediction of ground subsidence locations and how to decrease related damage. Methodology In general, to predict ground subsidence it is necessary to assume that subsidence occurrence is determined by subsidencerelated factors, and that future subsidence will occur under conditions that are the same as those for past subsidence according to the relative frequency concept of probability theory. On this basis, the relationship between areas where subsidence has occurred and subsidencerelated factors can be distinguished from the relationship between areas without past subsidence and subsidencerelated factors. In this work, the spatial relationships between subsidence locations and each subsidencerelated factor were analyzed using the frequency ratio which is one of the probability models. The frequency ratio is the ratio of the area where subsidence has occurred to the total study area, and is also the ratio of the probability of subsidence occurrence to no occurrence for a given attribute. In the case of subsidence occurrence, if the subsidence event is denotedB and a given factor’s attribute is denoted D, then the frequency ratio of D is the ratio of the conditional probabilities of B . Therefore, the greater this ratio, the stronger the relationship between subsidence occurrence and the given factor’s attribute. The frequency ratio was calculated for each