An updated version of time series analysis, SeriesSEE (Halford 2006), was applied to evaluate the test data. Component series in the models for each well consisted of built-in earth tides and various moving averages of five general time-series: a shallow well nearest the deep well to be modeled, a background deep well, hourly atmospheric pressure and precipitation records from a nearby airport, and a summation series of the precipitation consisting of the sum of hourly precipitation for the preceding 12.5 days. Pumping responses were then modeled with a Theis approximation. SeriesSEE fits the pumping component to the water level data along with the other components by optimizing the values for storativity and transmissivity. Five criteria were developed for accepting a given pumping model result: 1) models incorporating a pumping component should provide a definite improved fit measured by reduced RMSE, 2) the raw data should contain an indication of recovery after pumping was stopped, 3) model drawdowns should exceed three times the “noise” prior to starting pumping, 4) hydraulic conductivity from fitted transmissivity should be consistent with previous slug test results, and 5) multiple model realizations should provide consistent results. Model drawdown results from multiple wells were utilized further in a distance-drawdown analysis. The time-series analysis allowed use of the entire dataset from the test by accounting for the overwhelming effectiveness of the extreme precipitation event. Model results from individual wells and the distance-drawdown analysis were in general agreement.
