Analysis of water-level fluctuations in Wisconsin wells

More than 60 percent of the residents of Wisconsin use ground water as their primary water source. Water supplies presently are abundant, but ground-water levels continually fluctuate in response to natural factors and human-related stresses. A better understanding of the magnitude, duration, and frequency of past fluctuations, and the factors controlling these fluctuations may help anticipate future changes in ground-water levels.

This report presents the results of statistical analyses of historical ground-water level fluctuations in Wisconsin. Short- and long-term fluctuations are discussed in terms of their relation to components of the hydrogeologic system.

Water-level measurements from 124 wells with at least 20 years of record each were used in the study. The mean, highest and lowest monthly mean, median, and selected quantiles were obtained using the SAS Proc Univariate procedure. The frequency values from the Univariate table were used to construct stage-duration graphs. Pearson Type ill frequency analyses were used to obtain probabilities of exceedance of particular water levels. The data were divided into seasonal data sets for each well. The stage-duration and Pearson Type III analyses are based on past fluctuations but may be useful for estimating future water-level changes under similar conditions.

Multiple-regression analyses were run on data from groups of wells representing different ground-water districts. The SAS Proc Stepwise method was used. The regression model used average annual amplitude as the dependent variable and mean water level, mean annual precipitation, standard deviation of the seasonal mean precipitation, aquifer type, and topographic setting as the independent variables.This procedure produced different regression equations for each hydrogeologic district. Regression analyses also were done on data from groups of wells representing different aquifers. These regression models used average annual amplitude as the dependent variable and mean water level, topographic setting, and standard deviation of seasonal mean precipitation as the independent variables.

Because of the many factors influencing ground-water level fluctuations, it was difficult to obtain a regression model that accurately reproduced average annual amplitude. The results of the regression analyses are helpful in recognizing the important variables; however, the equations are not effective in predicting the amplitude of a particular well because local conditions were omitted in the regional analyses.

Hydrographs of average annual water level and frequency distribution analyses of annual maximum and minimum water levels were inspected for possible long-term trends. Analysis of annual maximum and minimum water levels indicates several periods in the annual cyclic fluctuations-two periods of recession (winter and summer), and two periods of rising levels (spring and fall). Usually, water levels are lowest in late winter and highest in spring for every annual cycle. The summer-fall minimum and the fall maximum are less distinct and do not occur every year.

A composite frequency analysis of extreme annual water levels on 71 of the wells shows that the lowest levels most frequently occur in December, February, or March. However, the record low usually occurs in August, September, or October during drought. Ground-water levels most often peak in May, April, or June. In the fan they may peak from September through December, depending upon complexities of meteorological, geomorphological, and geological factors.

The long-term cyclicity of ground-water level fluctuations is shown on hydrographs of wells Sw-7, Ln-25a, Mt-7, Ju-8, and Ju-98. Seasonal variations that tend to obscure the long-term trends are eliminated by plotting the average annual water levels. The hydrographs are similar even though the wells are 80 to 100 miles apart and constructed in different geologic materials. The long-term trends and the duration of the cycles apparently depend little on the location and on the lithologic composition of the aquifers, but rather on precipitation. The hydrographs show several periods of welldefined peaks and lows. The ground-water levels reached peaks in 1946, 1952, 1960, 1966, 1973, 1979. The average interval between these peaks is 6.6 years. The low levels occurred in 1949, 1955-59, 1964, 1970, and 1977; average interval between the low levels is 7.0 years, which is similar to that for the high levels.

Long-term trends are apparent on hydrographs of wells Br-46, Mr-2S, Pt-276, Ro-3, and Ve-8. The trend of average annual water levels has been generally increasing since the late 1950's and is in general agreement with the increasing trend of precipitation. Hydrographs of well Ve-8, which has the longest period of record in Wisconsin, indicate that the generally rising trend started even earlier at the end of an extensive drought period in the 1930's.