MAPPING URBAN AREAS OF THE STEPPE ZONE USING THE ARTIFICIAL NEURAL NETWORK METHOD
S.A. DUBROVSKAYA, R.V. RYAKHOV
Institute of Steppe, UB RAS, 460000, Orenburg, ul. Pionerskaya, 11, Russia skaverina@bk.ru
Keywords: ландшафтное картографирование, цифровая модель рельефа, геоморфометрические показатели, урбогеосистема, самоорганизующиеся карты, территориальное планирование, landscape mapping, digital elevation model, geomorphometric indicators, urbogeosystem, Self-Organizing Maps, spatial planning
Abstract
Based on the automated classification technique for artificial neural networks, an integrated cartographic model of the genetic types of topography and structural and functional zoning of the city of Orenburg was constructed using geomorphometric data to identify spatially homogeneous sections of landscape structures of the urbogeosystem. The spectral classification method without a teacher allows spatial differentiation of the urban technical systems and reliable information that is necessary to improve the environmental components and comfort of urban space. Data on the topography and properties of the underlying surface are inhomogeneous values that are brought into a single discrete form. To do this, at the processing stage, modern effective GIS tools are involved, creating high-precision analytical and cartographic material that is processed and presented in the form of grid structures. Based on the Self-Organizing Map method, a typological generalization of operational-territorial units, a thematic interpretation of classes, and a cartographic model of the urban landscape differentiation of the territory were compiled. The selected sections of the floodplain (high and low) and the terraces are concentrated in the same ordination plane of the Kohonen neural network but are different in terms of the characteristics of the relief (height, slope, exposure, and other morphometric indicators). The identification of the 1 st and 2 nd floodplain river terraces became possible with the use of field research data and medium-scale geomorphological maps. The verified model of the landscape base is superimposed on the modern scheme of the ecological and functional zoning of the urbogeosystem. The final map reveals the spatial structure of the development of the urban technological system. The suggested method of artificial neural networks makes it possible to update data obtained accord ing to the purpose of typological mapping. A comprehensive landscape classification of urbogeosystems is a reflection of the features of the geographical environment and shows the patterns of anthropogenic impacts, and the development of processes of changing the states of natural and anthropogenic geosystems.
|