Aquifer thermal energy storage (ATES) systems use aquifers for thermal storage to improve the efficiency in heating and cooling. Since water pumping and injection operations are repeated between groups of wells in an ATES system, the balance of injected heat and cold heat is important to prevent the excessive change of aquifer temperatures. Hence, the ground temperature during the operation of an ATES system needs to be accurately predicted through scientific approaches to achieve high energy efficiency in the ATES system. In this research, long-term heating and cooling tests were carried out in an ATES system installed in Akita City, Northern Japan. During the tests, field data, i.e., the temperatures and flow rates of the pumped and injected water, performance of heat pump, etc., were recorded and these data were analyzed to demonstrate the high energy efficiency of the system. Then, based on the geological and groundwater information at the test site, a 3D numerical model was developed using a groundwater and heat transport simulator. The model was validated through the history-matching between the measured and simulated pumping water temperatures. Finally, sensitivity studies were carried out to optimize the operation scheme of the ATES system. The sensitivity studies showed that the ratio of the number of pumping and injection wells should be close to one to effectively use the stored heat and that the period of the cooling and heating should be close enough to prevent the change of the aquifer temperatures.