A graphical method that uses continuous water-level and barometric-pressure data was developed to estimate barometric efficiency. A plot of nearly continuous water level (on the y-axis), as a function of nearly continuous barometric pressure (on the x-axis), will plot as a line curved into a series of connected elliptical loops. Each loop represents a barometric-pressure fluctuation. The negative of the slope of the major axis of an elliptical loop will be the ratio of water-level change to barometric-pressure change, which is the sum of the barometric efficiency plus the error.
The negative of the slope of the preferred orientation of many elliptical loops is an estimate of the barometric efficiency. The slope of the preferred orientation of many elliptical loops is approximately the median of the slopes of the major axes of the elliptical loops. If water-level change that is not caused by barometric-pressure change does not correlate with barometric-pressure change, the probability that the error will be greater than zero will be the same as the probability that it will be less than zero. As a result, the negative of the median of the slopes for many loops will be close to the barometric efficiency.
The graphical method provided a rapid assessment of whether a well was affected by barometric-pressure change and also provided a rapid estimate of barometric efficiency. The graphical method was used to assess which wells at Air Force Plant 6, Marietta, Georgia, had water levels affected by barometric-pressure changes during a 2003 constant-discharge aquifer test. The graphical method was also used to estimate barometric efficiency. Barometric-efficiency estimates from the graphical method were compared to those of four other methods: average of ratios, median of ratios, Clark, and slope. The two methods (the graphical and median-of-ratios methods) that used the median values of water-level change divided by barometric-pressure change appeared to be most resistant to error caused by barometric-pressure-independent water-level change. The graphical method was particularly resistant to large amounts of barometric-pressure-independent water-level change, having an average and standard deviation of error for control wells that was less than one-quarter that of the other four methods.
When using the graphical method, it is advisable that more than one person select the slope or that the same person fits the same data several times to minimize the effect of subjectivity. Also, a long study period should be used (at least 60 days) to ensure that loops affected by large amounts of barometric-pressure-independent water-level change do not significantly contribute to error in the barometric-efficiency estimate.