The top elevation of the uppermost layer represents the land surface and was approximated using an imported GSK-J4 and resampled digital elevation model. The bottom elevation of the fifth layer represents the bedrock surface, thereby constraining flow within the valley fill thickness ranging between 3 and 120 m thick. The bedrock surface was interpolated from available well logs, both in published literature (Randall, 1972) and public records from NYSDEC. The first upper two layers of the model represent the unconfined aquifer system. The third layer is a clay unit, which serves to confine the lowest two layers. The thickness and elevation

of the third layer was also interpolated from well logs (Randall, 1972). Both aquifer systems – upper and lower – were split into two layers apiece, with their interlayer elevation set at half of the aquifer thickness in each cell. There are four hydraulic conductivity units in this model (Fig. 4). The uplands are considered one homogeneous, low-conductivity unit, primarily serving as a transmitting media between the external boundary conditions and the valley walls. Separate hydraulic conductivity units were assigned to

the upper and lower aquifer systems. Cells representing the clay confining unit were assigned to the fourth conductivity DNA Damage inhibitor field. Any cell in the third layer with a thickness greater than 3 m is considered part of the confining unit. The remainder of the third layer, where the confining unit is thin or absent, is part of Dipeptidyl peptidase the upper aquifer hydraulic conductivity unit. Manual calibration indicated that this model was not significantly sensitive to conductivity of the confining unit in layer three at

the regional scale. Although there is extensive heterogeneity within the valley drift sequences, it is difficult to capture such a variability at this scale. Therefore, these hydraulic conductivity values better represent regional, effective conductivity. Uniform recharge of 62 mm/year was applied to the top of the model, representing the component of groundwater recharge derived from the infiltration of precipitation falling directly in the valleys. This value was approximated by adding the total volume removed from the system (through municipal pumping) to the net regional recharge estimated from the analytic element model (Best, 2013). Constant head boundaries on the outside of the active model area provide the lateral aquifer recharge derived from overland runoff, tributary infiltration, and interflow. In the baseline model, the constant head contribution to groundwater inflow from the boundary of the model was approximately 42%. Constant head contributions in the withdrawal scenarios were evaluated to ensure that this fraction of groundwater input did not unrealistically increase, results of which will be discussed in the sensitivity analysis. The Streamflow-Routing Package (Prudic et al.