GIS IN COMPUTING KYIV SLOPES STABILITY: THREE-DIMENSIONAL MODELING

The paper reports on landslide processes in Kyiv, and focuses on GIS methods in three-dimensional assessment of slope stability. Sliding hillsides of St. Andrew's Church in Kyiv make an area of study. The paper presents computation of the slope stability factor based on K. Tertsahi classical two-dimensional method and Hovland’s three-dimensional method. There have been recorded about 100 active landslides and 6 sliding reliefs in Kyiv. Most landslides have been temporarily stabilized by engineering constructions. However, landsliding increases because of unregulated property development on the slopes. The development of deep ravines is accompanied by complex movements of soil masses on the sides of ravines and on their bottoms resulting in landsliding or slumping. In general, to estimate a threat of landsliding there should be considered a scale of local and general slope destabilization. Three-dimensional modelling (3D) is widely used to calculate slope stability. At present, it is becoming essential in engineering calculations, with numerical simulation of the soil mass strain being applied to solve a wide range of environmental and hydrogeological problems. Stability factor for the investigated landslide site has been computed on (2D) sustainability model by K. Tertsahi and (3D) Hovland’s method. The GIS 3D method to assess stability of the landslide-prone slopes provides their further monitoring, for landslide preventive measures to be taken when necessary. The site under study makes a base slope of the Dnieper valley and is located at the foot of St. Andrew's hill. The geological structure of the investigated area is represented by the Quaternary and Neogene sediments to a depth of 23 m. Computation reveals two landslide-prone slopes, to the north and to the east of St. Andrew's Church. Three-dimensional spatial assessment enhances the accuracy of predicting landslide risks. Key words: landslides, stability factor, three-dimensional stability model, Hovland’s method.