Today with increasing in population and water demand in agriculture, industrial and urban use, water resources managers have been faced to the problem of supply water with suitable quality for different users. Because of interactions between surface water and groundwater in hydrosystems, these two resources should be seen simultaneously, then optimizing the conjunctive use of surface water and groundwater is one of the ways that is used in recent decades. Combined simulation-optimization models are powerful techniques to achieve the optimal conjunctive use policies. In this research, conjunctive use of surface water and groundwater in Koohpayeh-Sagzi subbasin located in Zayandeh roud basin is studied. In this region, the agricultural demand is the most important part that uses surface water transmitted by Abshar and Rudashtayn networks and pumped groundwater from surface and subsurface aquifers. A simulation-optimization model is developed using Support Vector Machines (SVMs) and Genetic algorithm (GA). SVMs is used as an approximated simulator of groundwater simulation model (GV4) to determine the change of groundwater level in each time period and GA searches the feasible space of problem until achieving the optimal solution. Some constraints are used to limit changes in groundwater level and available surface water. The embedding method is applied for linking between simulation and optimization models. The proposed model is applied for four different scenarios with different conditions. Results show that if sufficient surface water would be available, demand could be supply perfectly. In conditions that surface water is limited because of water shortage in Zayandeh roud basin, demand can’t be supplied by available water resources. Model shows that groundwater level is affected by agricultural returned water excessively, but with decreasing water demand by changing crop pattern or decreasing crop area a durable use of water resources could be achieved.
