Revealing the future complexity of urban water scarcity and drought via support vector machine: Case from semi-arid bursa urban area

Abstract

The Mediterranean Basin is a significant area will be affected by drought and water scarcity in future. In this context, Bursa urban area, the fourth largest city in terms of population in T & uuml;rkiye was used for quantification. A high-resolution global climate model of MPI-ESM-MR based RCP4.5 and 8.5, and population projections based on arithmetic and exponential growth models until 2100 was utilised. Support Vector Machine (SVM) regression was established between observed precipitation, evapotranspiration, runoff and reservoir volume for the reference period. Climate model outputs like precipitation and derived outputs such as evapotranspiration based on Penman-Monteith, runoff from SCS-Curve Number were used for SVM future dam volume prediction. Reference (observed data) and near and distant future (projected) dam volumes were converted to the Standardized Reservoir Index (SRI), and water scarcity as water per capita was also calculated. As a result, increased droughts and extreme conditions are identified in the near and distant future compared to the reference period. In addition, decrease in water per capita was determined with respect to the reference period. Therefore, results demonstrate that water scarcity is worsened by both semi-arid climate and population in urban area. Hence, water management in urban areas should address climatic variability and economic processes together.