Click here to snag a copy to read (Erin M.K. Haacker, Kayla A. Cotterman, Samuel J. Smidt, Anthony D. Kendall, David W. Hyndman). Here’s the abstract:
We use an 82-year record of water table data from the High Plains Aquifer to introduce a new application of segmented regression to hydrogeology, and evaluate the effects of droughts, crop prices, and local groundwater management on groundwater level trajectories. Across the High Plains, we find discernable regional cycles of faster and slower water table declines. A parsimonious Classification And Regression Tree (CART) analysis details correlations between select explanatory variables and changes in water table trajectories, quantified as changes in slope of well hydrographs. Drying relative to prior-year conditions is associated with negative changes in slope; in the absence of drying conditions, steep declines in commodity price are associated with positive changes in hydrograph slopes. Establishment of a groundwater management area is not a strong predictor for change in water table trajectories, but more wells tend to have negative changes in around the time of management areas are formation, suggesting that drought conditions are associated with both negative deflections in water table trajectory and enactment of management areas. Segmented regression is a promising tool for groundwater managers to evaluate change thresholds and the effectiveness of management strategies on groundwater storage and decline, using readily available water table data.