A technique for estimating ground-water levels at sites in Rhode Island from observation-well data

Estimates of future high, median, and low ground-water levels in Rhode Island are needed for engineering and architectural design decisions and for appropriate selection of land uses. For example, the failure of individual underground sewage-disposal systems due to high ground-water levels can be avoided if accurate water-level estimates are available. Estimates of extreme or average conditions are needed because short-duration re-construction observations are unlikely to be representative. 

The technique described  in this report utilizes a single water-level measurement at a site of interest, in combination with a long-term water-level record at an observation well, to estimate the long-term high, median, and low water levels at the site of interest. The transfer of information to the site of  interest depends on four fundamental assumptions: (1) Water levels will fluctuate in the future as they have in the past, (2) Water levels fluctuate seasonally,  (3) Ground-water fluctuations depend on site geology, and (4) Water levels throughout Rhode Island are affected by similar precipitation and climate.  The technique is based on the equivalent relation between the ratio of potential water-level change to maximum annual water-level range at the site and the ratio of potential water-level change to annual water-level range at the observation well. Equations for estimating high, median, and low water levels, and graphs of probable annual water-level range are given for selecting representative ranges of water levels for sand and gravel and till in Rhode Island. 

The accuracy of the technique is evaluated by use of the equations to estimate water levels at long-term observation wells where high, median, and low water levels are known from monthly measurements over many years. As a test of the estimating procedure,  6,697 estimates each of high, median, and low water levels (depth to water level exceeded 95, 50, and 5 percent of the time, respectively) were compared with measured water levels exceeded 95, 50, and 5 percent of the time at 14 sites unaffected by pumping or other known factors. Mean squared errors (average differences squared, between estimated and measured water levels) for the estimates ranged from 0.34 to 1.53 ft² for high water levels, 0.30 to 1.22 ft² for median water levels, and 0.32 to 2.55 ft² for low water levels. All mean squared errors are less than the State required 3-foot separation between the bottom of the stone underlying the seepage system and the maximum altitude of the water table. This degree of accuracy is acceptable for many design purposes.