The demand for water from the Warwick aquifer, which underlies the Fort Totten Indian Reservation in northeastern North Dakota, has been increasing during recent years. Therefore, the Spirit Lake Sioux Nation is interested in resolving questions about the quantity and quality of water in the aquifer and in developing a water-management plan for future water use. A study was conducted to evaluate the surface-water and ground-water resources of the Fort Totten Indian Reservation and, in particular, the ground-water resources in the area of the Warwick aquifer. A major component of the study, addressed by this report, was to define the ground-water flow system of the aquifer.

The Warwick aquifer consists of outwash deposits of the Warwick outwash plain that are as much as 30 feet thick and buried-valley deposits beneath the outwash plain that are as much as 200 feet thick. The aquifer is bounded on the north and west by end-moraine deposits and Devils Lake, on the south by the Sheyenne River Valley, and on the east by outwash deposits and ravines. The aquifer is underlain by Pierre Shale or by glacial till, clay, or silt. Ground-water gradients generally are small and rarely are more than 3 or 4 feet per mile. From 1982 to 1993, withdrawals from the Devils Lake well field averaged 1.5 cubic feet per second, and withdrawals from irrigation wells averaged 1.29 cubic feet per second. The combined discharge from springs may be about 3 cubic feet per second. During the early 1990s, the Warwick aquifer probably was in a steady-state condition with regard to storage change in the aquifer.

A finite-difference, three-dimensional, ground-water flow model provided a reasonable simulation of ground-water flow in the Warwick aquifer. The aquifer was divided vertically into two layers and horizontally into a grid of 83 by 109 cells, each measuring 656 feet (200 meters) per side. The steady-state simulation was conducted using 1992 pumpage rates and October 1992 water levels. The mean absolute difference between simulated and derived water-level altitudes during final calibration of the model was 1.52 feet. The two transient simulations were conducted for 20 time intervals of 1 year each using both the small and large storage estimates, doubled 1992 pumpage from the Devils Lake well field, 1992 irrigation pumpage, and initial water-level altitudes simulated by the October 1992 steady-state simulation. In the simulation using the small storage estimate and doubled pumpage, model cells in the area of the well field went dry after 13 years.

Assumptions made in the design of the model generally are supported by the digital simulation. Except in the area of Warwick Springs and smaller springs, lateral and basal boundaries of the aquifer are impermeable. The flow system is dominated by recharge and evapotranspiration. Recharge rates obtained during the calibration process were lower in topographically high areas than in topographically low areas. Hydraulic conductivity in the area of the Devils Lake well field was larger than that in the rest of the aquifer.