Using stormwater hysteresis to characterize variations in quick and diffuse flowpaths within a conduit dominated karst spring

Abstract

Groundwater quality in karst systems is difficult to monitor because the extreme heterogeneity within the recharge area and complex subsurface flow network makes flowpaths and travel rates difficult to predict. Understanding how flowpaths vary during storm events is important because water transmitted through conduit flowpaths can travel fast, may come from long distances, and has little filtration of contaminants. The hypothesis tested in this project is that ion ratios in spring discharge will show the timing of changes from diffuse to quick flow depending on storm intensity and antecedent conditions and provide more detail than total ion conductivity. Cedar Run Spring is located in the Cumberland Valley of south-central Pennsylvania. The valley is part of the larger Great Valley Section and is composed of Cambro-Ordovician aged carbonate units, collectively known as the Cumberland Valley Sequence. Initial background monitoring with data loggers and monthly samples indicated that Cedar Run Spring had a conduit component within the flow network. An automated stormwater sampler was installed at the spring and collected twenty-four water samples for major-ion analysis. Storm-intensity conditions ranged from high to low for the four storm events collected. In addition, the antecedent conditions varied from wet to dry. The Mg/Ca ratio characterizes the flowpath through which the water moves. A higher ratio indicates more diffuse flow because slower flow paths are needed to dissolve dolomite (which contains Mg), while a lower ratio indicates more conduit flow because calcite (Ca dominant) dissolves more readily. Hysteresis loops of conductivity versus discharge rotated counterclockwise because conductivity decreased on the rising limb of storm response, followed by an increase on the falling limb for all but the winter storm, which was influenced by road salt. In contrast, hysteresis loops for Mg/Ca versus discharge rotated in a clockwise direction for all but one of the storm events because of an increase in Mg/Ca that indicated a flush of older matrix water. The storm event that did not display in initial increase in Mg/Ca was apparently flushed by a recent previous storm event. Mg/Ca hysteresis for the storm events that were diffuse displayed several sharp increases and decrease in addition to several smaller hysteresis loops in response to multiple slugs of recharge water. These variations were not indicated in overall conductivity. High intensity events displayed a quick switch in flowpaths, as indicated by the increase in Mg/Ca early on the rising limb, and a single hysteresis loop. The rapid change in Mg/Ca suggested that during storm events water was able to enter the karst system through sinkholes, then activated flowpaths with older matrix water. Mg/Ca proved to be better at tracking the variability in flowpaths during storm events than the overall conductivity, because Mg/Ca is directly related to water-rock interactions.