GSA Annual Meeting, November 5-8, 2001

Paper No. 0
Presentation Time: 11:20 AM

RELATING WELL YIELDS TO SITE CHARACTERISTICS IN FRACTURED BEDROCK OF NEW HAMPSHIRE


MOORE, Richard Bridge1, SCHWARZ, Gregory E.2, CLARK Jr, Stewart F.3, WALSH, Gregory J.4, DEGNAN, James R.5 and MACK, Thomas J1, (1)U.S. Geol Survey, 361 Commerce Way, Pembroke, NH 03275, (2)U.S. Geol Survey, National Center, 12201 Sunrise Valley Drive, Reston, VA 20192, (3)U.S. Geol Survey, P.O. Box 628, Montpelier, VT 05601, (4)U.S. Geol. Survey, P.O. Box 628, Montpelier, VT 05601-0628, (5)U.S. Geol Survey, 331 Commerce Way, Pembroke, NH 03275, rmoore@usgs.gov

The USGS investigated well yields of fractured bedrock in New Hampshire using 20,308 accurately located wells. The data set, collected by the State of New Hampshire, includes yield, depth, and construction information reported by the driller of public and private wells drilled in the past 18 years. The analysis produced a statewide GIS grid of well-yield probabilities, which can be used to characterize bedrock aquifers for the development of regional ground-water-flow models and assessment of water supplies.

Instrumental-variables regression was used as the primary method of analysis. This estimation method facilitates the prediction of yield contingent on many predictor variables and provides an unbiased estimate of the yield/depth relation - a relation that is not directly estimated using standard least squares methods because of the effects of yield targeting in the drilling data. The results show that well yield is significantly increased in those areas characterized by flatter slopes, adjacent waterbodies, valleys, large topographic drainage areas, proximity to specific types of lineaments, or various geologic map units such as the Frontenac and Rye Formations.

A quadrangle-scale investigation of yield was done to determine the degree to which predictive well-yield relations can be improved by incorporating additional geologic, fracture, and lineament data that are not available statewide. A large amount of well data in the Pinardville and Windham, N.H. quadrangles (1,682 and 1,504 wells respectively) provided an opportunity for assessing the effects of these additional variables on the accuracy of yield estimates. The improved characterization, resulting from detailed geologic mapping, may be suitable for quadrangle-scale ground-water-flow-model development and assessment of ground-water supplies.