As the development of mountain areas has recently increased in Korea, existing roads are being renovated, and new highways are being constructed, which increases driving speeds in mountainous areas. However, the mountainous region in northeastern Korea is more likely to form black ice due to higher humidity, frequent fog, and hillshade, depending on the terrain, which can cause serious traffic pileups. In this study, therefore, we present a method to build a more effective black ice prediction and warning system by linking spatial information to the existing road management system that estimates the road surface temperature based on real-time weather information. The spatial information enabled a prediction to be made of the risk level of black ice formation for each time zone by simulating changes in the shadow area based on precise 3D terrain information. Moreover, this information also presented slope and curvature information of the road to estimate the risk zone. The spatial information was integrated with weather data to predict road surface temperature. The proposed system was tested in two mountainous regions with weather data accumulated from 2017 to 2018. As a result, the proposed system anticipated 71% of traffic accidents caused by black ice during the testing period. The results show that the system can contribute significantly to preventing black-ice-related traffic accidents by providing reasonable predictions.

A method to find corresponding feature class pairs, including hierarchical M:N pairs between two geospatial datasets is proposed. Applying an overlapping analysis to the object sets within the feature classes, the similarities of the feature classes are estimated and projected onto a lower-dimensional vector space after applying the graph embedding method. In this space, conventional mathematical tools—agglomerative hierarchical clustering in this study—could be used to analyze semantic correspondences between the datasets and identify their hierarchical M:N corresponding pairs. The proposed method was applied to two cadastral parcel datasets; one for latest land-use records in an urban information system, and the other, for original land-use categories in the Korea land information system. To quantitatively assess identified feature pairs, F-measures for each pair are presented. The results showed that it was possible to find various semantic correspondences of the feature classes and infer regional land development characteristics.