Relevance and purpose of the work. Currently, the area of development of Tomsk is increasing. New neighborhoods are growing on previously undeveloped land (for example, on the left bank of the river Tom). The central part of the city is being redeveloped and reconstructed. It is impossible to develop a high-quality territory without taking into account the dynamics, mechanisms, factors and patterns of development of dangerous natural and technological processes, the forecast of their development. The purpose of the work is to establish the patterns of gully erosion, assess the intensity of its development, and predict the probability of its occurrence within the new city boundaries. Methods of research. We performed an assessment and forecast of the development of gully erosion in Tomsk using GIS technologies, which are an important tool in the city management process due to their ability to process and analyze multidimensional data about the geological environment. We compared the traditional model of data-driven frequency ratio (FR) and expert-based multi-criteria assessment, i.e. analytical hierarchical process by weighting of gulley conditioning factors. Results of the work. We constructed a map of the distribution of gullies on the territory of the city, including 23 polygons. These polygons were then randomly divided into training (16 polygons or 70%) and validation data (7 polygons or 30%). We used seven gulley-conditioning factors for the two models to produce gulley susceptibility maps: slope angle, slope aspect, curvature, elevation, geological structure of the territory; types of filtration sections; distance to the river, to analyze the spatial patterns that determine the development of gully erosion. The spatial correlation between gulley locations and the conditioning factors were identified using GIS-based statistical models. We constructed gulley susceptibility maps based on the ranking of each factor by two methods using a training data set. Receiver operating characteristics (ROC) were used to validate the resulting susceptibility maps. The area under the curve (AUC) was 0.905 for the AHP model and 0,800 for the FR model, respectively, which indicates excellent and high quality of forecast maps. We proved that both methods are beneficial for assessment the susceptibility of the territory to gully erosion. We recommend using the constructed maps for regional planning and hazard mitigation, as well as in education by teaching the discipline “Engineering geodynamics”.