Data mining on the ossibility study of urban fabric's physical change modeling

Urban expansion has been a very important topic not only in the management of sustainable development, but also in the fields of remote sensing and geographic information science. Urban planners also encounter a huge challenge to require the understanding of the complex urban growth process, which involves various factors with different patterns of behavior. Modeling of an urban development pattern is the prerequisite to understanding the process and might be the first step in making a decision on urban planning. The main issues of great importance in land use modeling include spatial dynamics, temporal dynamics, incorporation of human drivers of land use changes, and scale dynamics. Dynamic simulation models and empirical estimation models have been used to model land use changes. Rule-based simulation models are mostly suitable for incorporating spatial interaction effects and handling temporal dynamics. However, Cellular Automata (CA) models do not focus on interpretation or understanding of Spatio-temporal processes of urban growth. Most dynamic simulation models cannot incorporate enough socioeconomic variables. Empirical estimation models use statistical techniques to model the relationships between land use changes and the drivers based on historic data. As an empirical estimation method, a regression model has been used in deforestation analysis, agriculture, and urban growth modeling.This paper tends to apply a regression to model urban changes in the old part of the Kermanshah city (Faizabad neighborhood) from 1956 to 2011. In this regard, multi-temporal airborne images were used as a data source. According to the common assumptions, urban physical forms are characterized as the results of a complex deliberation process that involves consideration of many factors. Monitoring of urban fibers transformation through airborne images and translating the obtained data provide a systematic database which can be used in empirical analyses. Applying a grid network for the first time yields the images to quantify obtained results from every cell of the network. In the second step, each cell value recorded for the available temporal period and the Minitab 16 software were used to gain regression equations from these values. The nearest relation between cells value in an observed period was provided by the software as a quadratic equation. Adding an appropriate value to equations gives an estimated amount for its related cell in the selected period of time. The approach was calibrated for 2016 by cross comparing of actual and simulated cells value. In order to examine the modeling precision, the same process was done for 2016 images and cells data were extracted. After the cross comparing, the simulation results were consonant more than 70% with actual data of 2016, which was satisfactory to approve the calibration process. Urban development programs and non-professional interventions in this case study area caused to more disparities and dismantle the logic of the model. Simplicity and the easiness of the proposed model are main privileges in comparison with the previous ones. In summary, this model can be used as a quick responsible way to predict urban changes in specific areas which give acceptable schematic responses.