چكيده به لاتين
Water resources conservation, as it plays an indispensable role in achieving sustainable development, should be at the core of managerial actions in both local and national scales to ensure accessing to clean and sufficient water supply in the long-term. The unprecedented water quality deterioration caused by the unscientific application of chemical fertilizers and extensive industrial activities have heightened the need for establishing effectual approaches for water quality monitoring and management. Javeh reservoir, on the one hand, receives sewage effluent from nearby rural areas and, on the other hand, immediately adjoins the Sanandaj city landfill, which both are nutrient-rich and contribute to nourishing algae blooms, poisoning ecosystems. This study set out to develop a web-based decision support tool capable of predicting the concentration of chlorophyll-a and dissolved oxygen, and computing water quality and quantity indices, which provide deeper insight into the current situation of the reservoir. At first, the reservoir was modeled in a 2D hydrodynamic and water quality simulation, CE-QUAL-W2, and nine additional outflow scenarios were developed to further extend the available data set to be fed to machine learning models. Two machine learning models, i.e., Random Forest and XGBoost were selected to eliminate the time-consuming computation of CE-QUAL-W2 model. Due to exhibiting superior performance, XGBoost was chosen as the primary prediction model. Using open-source software, a web-based decision support tool was developed, which was capable of receiving daily data, predicting the concentration of dissolved oxygen and chlorophyll-a using the developed machine learning model, calculating water quality and quantity indices based on the entered data. The Carlson trophic state index and Vollenweider trophic state classification were used to evaluate the reservoir trophic state for the simulation period. Also, the water resource vulnerability index and social water stress index were selected and added to the framework in order to provide an insightful description of the current water quantity state. To evaluate the performance of the tool, the reservoir water quality and quantity data from September 2016 to September 2017 were entered into the tool, and the results were obtained. According to both quality indices, the results indicate the risk of eutrophication in the warm seasons, while the reservoir is facing mild and weak risk of eutrophication in the cold seasons. Also, the vulnerability assessment suggests the vulnerable situation of the studied area, although, social water stress index indicates that no stress is facing the water resources in the area. Overall, in the studied period, the reservoir is threatened by relatively mild eutrophication risk, and the studied area is not experiencing water stress, yet is vulnerable.
