Relations between precipitation, groundwater withdrawals, and changes in hydrologic conditions at selected monitoring sites in Volusia County, Florida, 1995--2010

A study to examine the influences of climatic and anthropogenic stressors on groundwater levels, lake stages, and surface-water discharge at selected sites in northern Volusia County, Florida, was conducted in 2009 by the U.S. Geological Survey. Water-level data collected at 20 monitoring sites (17 groundwater and 3 lake sites) in the vicinity of a wetland area were analyzed with multiple linear regression to examine the relative influences of precipitation and groundwater withdrawals on changes in groundwater levels and lake stage. Analyses were conducted across varying periods of record between 1995 and 2010 and included the effects of groundwater withdrawals aggregated from municipal water-supply wells located within 12 miles of the project sites. Surface-water discharge data at the U.S. Geological Survey Tiger Bay canal site were analyzed for changes in flow between 1978 and 2001. As expected, water-level changes in monitoring wells located closer to areas of concentrated groundwater withdrawals were more highly correlated with withdrawals than were water-level changes measured in wells further removed from municipal well fields. Similarly, water-level changes in wells tapping the Upper Floridan aquifer, the source of municipal supply, were more highly correlated with groundwater withdrawals than were water-level changes in wells tapping the shallower surficial aquifer system. Water-level changes predicted by the regression models over precipitation-averaged periods of record were underestimated for observations having large positive monthly changes (generally greater than 1.0 foot). Such observations are associated with high precipitation and were identified as points in the regression analyses that produced large standardized residuals and/or observations of high influence. Thus, regression models produced by multiple linear regression analyses may have better predictive capability in wetland environments when applied to periods of average or below average precipitation conditions than during wetter than average conditions. For precipitation-averaged hydrologic conditions, water-level changes in the surficial aquifer system were statistically correlated solely with precipitation or were more highly correlated with precipitation than with groundwater withdrawals. Changes in Upper Floridan aquifer water levels and in water-surface stage (stage) at Indian and Scoggin Lakes tended to be highly correlated with both precipitation and withdrawals. The greater influence of withdrawals on stage changes, relative to changes in nearby surficial aquifer system water levels, indicates that these karstic lakes may be better connected hydraulically with the underlying Upper Floridan aquifer than is the surficial aquifer system at the other monitoring sites. At most sites, and for both aquifers, the 2-month moving average of precipitation or groundwater withdrawals included as an explanatory variable in the regression models indicates that water-level changes are not only influenced by stressor conditions across the current month, but also by those of the previous month. The relations between changes in water levels, precipitation, and groundwater withdrawals varied seasonally and in response to a period of drought. Water-level changes tended to be most highly correlated with withdrawals during the spring, when relatively large increases contributed to water-level declines, and during the fall when reduced withdrawal rates contributed to water-level recovery. Water-level changes tended to be most highly (or solely) correlated with precipitation in the winter, when withdrawals are minimal, and in the summer when precipitation is greatest. Water-level changes measured during the drought of October 2005 to June 2008 tended to be more highly correlated with groundwater withdrawals at Upper Floridan aquifer sites than at surficial aquifer system sites, results that were similar to those for precipitation-averaged conditions. Also, changes in stage at Indian and Scoggin Lakes were highly correlated with precipitation and groundwater withdrawals during the drought. Groundwater-withdrawal rates during the drought were, on average, greater than those for precipitation-averaged conditions. Accounting only for withdrawals aggregated from pumping wells located within varying radial distances of less than 12 miles of each site produced essentially the same relation between water-level changes and groundwater withdrawals as that determined for withdrawals aggregated within 12 miles of the site. Similarly, increases in withdrawals aggregated over distances of 1 to 12 miles of the sites had little effect on adjusted R-squared values. Analyses of streamflow measurements collected between 1978 and 2001 at the U.S. Geological Survey Tiger Bay canal site indicate that significant changes occurred during base-flow conditions during that period. Hypothesis and trend testing, together with analyses of flow duration, the number of zero-flow days, and double-mass curves indicate that, after 1988, when a municipal well field began production, base flow was statistically lower than the period before 1988. This decrease in base flow could not be explained by variations in precipitation between these two periods.