Summary and Conclusions--USGS OFR 95-470

SUMMARY AND CONCLUSIONS

The effects of ground-water withdrawals on streamflow have become an issue of major concern in the Puget Sound Lowland of western Washington as continuing population growth increases the demand for water. Surface-water resources are completely allocated in important segments of the region, and future growth there will most likely depend on the availability of ground water. Though the basic nature of interactions between the ground-water and surface-water systems is well known, the details of these interactions in small basins of the Puget Sound Lowland are not well understood. Ground-water development will, in most cases, affect the baseflow to streams. The lack of understanding about the details of the hydrologic system is due to a number of factors, but the most important may be the complexity of the Quaternary geology of the region. Repeated series of glacial advances and retreats, punctuated by interglacial periods of deposition and erosion, created an extremely complex system of aquifers and confining layers through which ground water passes as it moves from recharge areas to discharge points at streams and springs. The lack of understanding is compounded by the difficulty and expense of collecting the data necessary to characterize these systems. Finally, many of the traditional tools for assessing the interactions between ground water and surface water are either too simplistic to be useful (analytical models) or too complex and expensive (numerical models) to be practical for solving site-specific problems that face regulators. Nonetheless, regulators must maintain minimum streamflow and protect the interests of surface-water rights, so they need a more detailed conceptual understanding of the effects of ground-water withdrawals on streamflow to help guide decisions.

The purpose of this study was to provide a better understanding of relations and interactions between the ground-water and surface-water systems in small basins of the Puget Sound Lowland and, particularly, to identify some of the important factors controlling the response of the systems to ground-water withdrawals. The primary tool in this investigation was a numerical ground-water-flow model. The model was developed for a hypothetical basin in the Puget Sound Lowland and was based on a conceptual model synthesized from the work of many previous investigators in the region. Topography, geology, drainage, and climate were defined for the 262 square mile hypothetical basin. Hydrologic conditions simulated by the numerical model, such as ground-water levels and discharge to streams and springs, were compared with conditions typical of the region, and model parameters were adjusted until simulated and typical conditions agreed closely. The calibrated, or baseline, model was then used to simulate the effects of ground-water withdrawals on discharge to streams and springs under a variety of scenarios. Seven series of scenarios were simulated in which the effects of 1) distance from the well to a stream, 2) the presence of a confining layer, 3) pumping rate, 4) depth of the pumped aquifer, 5) distance from the well to a bluff, 6) well density, and 7) recharge rate were evaluated.

The results of each simulation were compared with the baseline model results to compute the percentage of the well discharge that was derived, or captured, by diverting flow that otherwise would have discharged to streams and springs. All simulations were of equilibrium, or steady-state conditions. That is, they simulated conditions after water levels had adjusted to the pumping stress and no changes in ground-water storage were occurring.

The central part of the hypothetical basin is a drift plain composed of layered Pleistocene glacial drift and interglacial sediments bounded on the east and south by low-permeability Tertiary bedrock and on the west and north by steep bluffs. At the base of the bluffs, 200 to 600 feet below the drift plain, lies a broad valley drained by a major stream. The valley contains up to 500 feet of alluvium consisting of a heterogeneous mixture of gravel, sand, silt and clay. The drift plain has relatively low relief and the streams that drain the plain have low gradients until they descend the bluff to the major stream valley; where the stream crosses the bluff it has incised a deep canyon, exposing the drift deposits. Recharge to the ground-water system depends on annual precipitation and the permeability of the geologic layer at the surface. The mean annual precipitation in the basin is 44 inches per year and the recharge rate in areas where the more permeable outwash deposits are exposed is 27 inches per year compared to recharge of only 18 inches per year in areas where the less permeable till is exposed; till covers most of the basin and the average recharge is 20 inches per year (389 cubic feet per second).

A three-dimensional numerical model of the ground-water-flow system of the hypothetical basin was constructed using the U.S.Geological Survey's MODFLOW model. The ground-water system was subdivided horizontally into a regular grid of cells, each having dimensions of 1,500 feet per side; 50 columns and 70 rows were included in the grid. The vertical dimension was subdivided using 13 layers of cells. Three glacial sequences, each consisting of recessional outwash, till, advance outwash, and interglacial deposits, were part of the conceptual model of the hypothetical basin. Each hydrogeologic layer was simulated using a separate model layer and, therefore, the three glacial sequences made up the upper 12 layers of the model. Beneath the drift plain, the 13th (bottom) layer represented undifferentiated glacial and interglacial deposits. The Quaternary alluvium underlying the major stream valley was represented in layers 9 through 13. The lower boundary of the model represented the contact between the Quaternary unconsolidated sediments and the consolidated Tertiary siltstones and mudstones that form a low-permeability (no-flow) boundary to the model.

Thickness and hydraulic characteristics of the hydrogeologic layers were initially assigned on the basis of values published from previous investigations in the Puget Sound Lowland. Values of hydraulic characteristics were modified during model calibration to provide a better fit to expected hydrologic conditions in the hypothetical basin. The horizontal hydraulic conductivity of the glacial sequences ranged from 0.25 foot per day and 1.0 foot per day for the till and interglacial confining layers to 100 feet per day for the outwash aquifers. The alluvial deposits of the major stream valleys were assigned a value of 50 feet per day and the undifferentiated deposits a value of 25 feet per day. Ratios of horizontal to vertical hydraulic conductivity ranged from 10 for outwash and alluvial aquifers to 100 and 200 for till and interglacial confining layers. Each layer was assumed to be homogeneous.

Ground water generally flows downward beneath the principal recharge area on the drift plain and then flows laterally from the south and east toward the primary discharge areas, where it flows upward. The primary discharge areas are the major stream valley and the springs that discharge on the bluffs to the north and west; however, shallower, local flow systems also discharge to streams and springs on the drift plain. In the baseline model, 73 percent (285 cubic feet per second) of the ground water discharged to the major stream valley and springs on the bluffs; the remaining 27 percent (104 cubic feet per second) discharged to streams and springs on the drift plain. The proportions of discharge to the major stream valley and the drift plain were reasonable on the basis of expected baseflow to streams on the drift plain of 100 to 175 cubic feet per second. The simulated range in specific discharge to streams on the drift plain of 0.3 to 3 cubic feet per second per mile also compared well with the expected range of 0.8 to 3.9 cubic feet per second per mile based on gain-loss data for a small watershed in southwest King County.

The following principal conclusions were drawn from the simulation of various pumping scenarios:

A well pumped from an unconfined outwash aquifer that is in contact with a streambed will capture nearly all of its discharge by diverting flow from the nearest reaches of the stream. Increasing the distance between the well and the stream allows the well to capture some discharge from other streams on the drift plain, but does not affect discharge to springs on the bluffs or to the stream in the lower valley.
When a confining layer separates the nearest stream from the pumped aquifer, the effects of pumping spread over a much larger area. The low-permeability confining layer forces the cone of depression of the well to extend to greater distances to divert the natural discharge required to offset pumping.
The presence of a confining layer between the well and the stream is more important than the distance between the well and the stream in determining the distribution of capture of natural discharge throughout the basin.
At equilibrium, the magnitude of drawdown and capture at any point are a function of the pumping rate.
As the depth of a well and the number of confining layers between it and discharge areas increases, capture of discharge to streams and springs is distributed over increasingly larger areas.
The bluffs are important hydrogeologic boundaries. Discharge to springs on the bluffs is very sensitive to the distance of wells from the bluffs.
The density of wells does not have a significant effect on the equilibrium distribution of capture of natural discharge; however, higher well densities result in greater local drawdown effects.
Impervious areas associated with development can reduce ground-water recharge. Natural discharge to streams and springs will be reduced by an equivalent amount, in addition to the reduction due to capture by wells. Artificially enhancing recharge to exceed natural recharge rates will increase natural discharge to streams and springs and can offset capture of natural discharge by wells.

These conclusions are based on the simulated equilibrium response of the ground-water flow system in the hypothetical basin to the various pumping scenarios. The results of the steady-state (equilibrium) model are a very simplified representation of a system that, in reality, changes temporally in very complex ways. The equilibrium model allows evaluation of scenarios based on their long-term (equilibrium) effects, but simulation of the transient response of the system to seasonal variations in pumping or to long-term climatic changes (drought) would allow greater insight as to the time required to reach equilibrium and to the short-term as opposed to long-term response in different parts of the system. For example, when pumping from a confined aquifer near a stream on the drift plain, drawdown in the confined aquifer may be transmitted very quickly to the bluffs where it captures discharge to springs, whereas capture from the nearby stream may take much longer because of the time required for the drawdowns to be transmitted across the confining layer. Development of a transient version of the hypothetical basin model would involve (1) estimating values of storage coefficient for each hydrogeologic layer, (2) estimating the seasonal distribution of recharge in the basin, and (3) calibrating the model to expected conditions.