The Pearl Harbor aquifer in southern O‘ahu is one of the most important sources of freshwater in Hawai‘i. A thick freshwater lens overlays brackish and saltwater in this coastal aquifer. Salinity profiles collected from uncased deep monitor wells (DMWs) commonly are used to monitor freshwater-lens thickness. However, vertical flow in DMWs can cause the measured salinity to differ from salinity in the adjacent aquifer or in an aquifer without a DWM. Substantial borehole flow and displacement of salinity in DMWs over several hundred feet have been observed in the Pearl Harbor aquifer. The objective of this study was to evaluate the effects of borehole flow on measured salinity profiles from DMWs. A numerical modeling approach incorporated aquifer hydraulic characteristics and recharge and withdrawal rates representative of the Pearl Harbor aquifer. Borehole flow caused by vertical hydraulic gradients associated with both the natural regional flow system and groundwater withdrawals was simulated.

Model results indicate that, with all other factors being equal, greater withdrawal rates, closer withdrawal locations, or higher hydraulic conductivities of the well cause greater borehole flow and displacement of salinity in the well. Borehole flow caused by the natural groundwater-flow system is five orders of magnitude greater than vertical flow in a homogeneous aquifer, and borehole-flow directions are consistent with the regional flow system: downward flow in inland recharge areas and upward flow in coastal discharge areas. Displacement of salinity inside the DMWs associated with the regional groundwater-flow system ranges from less than 1 to 220 ft, depending on the location and assumed hydraulic conductivity of the well. For example, upward displacements of the 2 percent and 50 percent salinity depths in a well in the coastal discharge part of the flow system are 17 and 4.4 ft, respectively, and the average salinity difference between aquifer and borehole is 0.65 percent seawater salinity. Groundwater withdrawals and drawdowns generally occur at shallow depths in the freshwater system with respect to the depth of the DMW and cause upward flow in the DMW. Simulated groundwater withdrawal of 4.3 million gallons per day that is 100 ft from a DMW causes thirty times more borehole flow than borehole flow that is induced by the regional flow field alone. The displacement of the 2 percent borehole salinity depth increases from 17 to 33 ft, and the average salinity difference between aquifer and borehole is 0.85 percent seawater salinity. Peak borehole flow caused by local groundwater withdrawal near DMWs is directly proportional to the pumping rate in the nearby production well. Increasing groundwater withdrawal to 16.7 million gallons per day increases upward displacement of the 50 percent salinity depth (midpoint of the transition zone) from 4.6 to 77 ft, and the average salinity difference between aquifer and borehole is 1.4 percent seawater salinity. Simulated groundwater withdrawal that is 3,000 ft away from DMWs causes less borehole flow and salinity displacements than nearby withdrawal. Simulated effects of groundwater withdrawal from a horizontal shaft and withdrawal from a vertical well in a homogeneous aquifer were similar. Generally, the 50 percent salinity depths are less affected by borehole flow than the 2 percent salinity depths. Hence, measured salinity profiles are useful for calibration of regional numerical models despite borehole-flow effects. Commonly, a 1 percent error in salinity is acceptable in numerical modeling studies. Incorporation of heterogeneity in the model is necessary to simulate long vertical steps observed in salinity profiles in southern O‘ahu. A thick zone of low aquifer hydraulic conductivity limits exchange of water between aquifer and well and creates a long vertical step in the salinity profile. A heterogeneous basalt-aquifer scenario simulates observed vertical salinity steps and borehole flow that is consistent with measured borehole flow from DMWs in southern O‘ahu. However, inclusion of local-scale heterogeneities in regional models generally is not warranted.