PREDICTION OF WATER CONSUMPTION IN ISTANBUL BY MEANS OF STATISTICAL FORECASTING MODELS & GEOGRAPHIC INFORMATION SYSTEMS (GIS)

Abstract

Environmental pollution and future scenarios associated with increasing population foresees the water shortage even water scarcity in parallel to climatic conditions. In large cities, municipalities have to plan and expand their infrastructure investments based on the projects related with water consumption values which are estimated according to population growth using prediction methods. Thus water consumption is vital in large metropolitan city planning. In this study, water consumption of the 39 districts of Istanbul was investigated during the period of 2010-2014. However, in 18 districts of Istanbul, there is no statistically significant established pattern. Main reasons of failures of modelling are considered as migration, restructuring of the administrative borders, and re-planning of the districts as either industrial or commercial districts. In the study, water consumption values of Istanbul are determined by means of statistical forecasting models. The ratio model results are found close to logistic model outcomes. In addition, different properties of the changes in water consumption are evaluated based on periods of years. However, due to the possible differentiations in the scale and development level, similar districts are separated by cluster analysis. Four different water consumption patterns are obtained after the process. Thus, determination of the different patterns is turned out to be more meaningful and their relationships with other parameters are defined for interpretation. When the results were analyzed, extreme changes in 8 districts were figured out. In addition, by using two-phase least mean square method and only by accepting water consumption statistics as associated sequentially; it is possible to obtain and determine units which are not suitable to be used the model. This application presented that, 3 districts whose administrative boundaries were changed, are different from the others. Statistical results were visualized using Geographic Information Systems (GIS). Planning can be done using mathematical methods especially in developing cities. As such, effective investment can be considered and supported.