|
WELL-YIELD PROBABILITIES RELATED to SITE and WELL CHARACTERISTICS in the FRACTURED-BEDROCK AQUIFER of NEW HAMPSHIRE
National Ground Water Association
March 13-15, 2002, Denver, Colo.
Richard Bridge Moore, U.S. Geological Survey
Gregory E. Schwarz, U.S. Geological Survey
Stewart F. Clark, Jr., U.S. Geological Survey
Gregory J. Walsh,
U.S. Geological Survey
James R. Degnan, U.S. Geological Survey
Thomas J. Mack, U.S. Geological Survey
Abstract
The USGS investigated the yields of wells in New Hampshire's bedrock aquifer . The data set used for the study was compiled by the State of New Hampshire, and includes 20,308 accurately located wells with information on yield, depth, and construction. The analysis produced a statewide GIS grid of well-yield probabilities, which may be used to assess potential water supplies and characterize bedrock aquifers for the development of regional ground-water flow models.
Multiple regression, using instrumental variables, was the primary method of analysis. The dependent variable was the natural log of the reported well yield. With over 400 explanatory variables examined, results indicate that well yield is generally lower in wells located on step hill slopes and hilltops, and at a greater distance from surface-water bodies. Yields were found to be greater in wells located in valleys, in sites with large upgradient topographic drainage areas, and within 100 feet of some types of lineaments. Well yields also correlate with 29 mapped geologic units. For example, the Frontenac and Rye Formations have high yields.
Quadrangle-scale investigations were done to determine the degree to which predictive well-yield statistical relations can be improved by local geologic mapping. Additional geologic, fracture, and lineament data were collected for the Pinardville and Windham, N.H. quadrangles, where data from numerous wells (1,682 and 1,504 wells, respectively) were used to test the value of the additional information. The statistical model was applied with and without the additional data. Yield-probability maps were produced for each of the two quadrangles, with and without the added data. These maps clearly demonstrate the advantage of including detailed geologic map units and fracture-correlated lineaments when predicting well yield. The improved characterization may improve simulation of ground-water flow and assessment of ground-water supplies.
U.S. Geological Survey  NH-VT Water Science Center Home Page |