A NEW HYDROGRAPH-DECOMPOSITION METHODOLOGY FOR KARST SPRINGS WITH RECESSION COEFFICIENTS THAT INCREASE WITH TIME

Traditional karst-spring hydrograph decomposition techniques using exponential curve fitting can identify up to three straight-line segments. In some instances, the intermediate slope (or recession coefficient) is steeper than the initial slope, producing a convex form on the master depletion curve. Inasmuch as discharge is proportional to head, the discharge is expected to decline the quickest (i.e. the slope to be steepest ) immediately following the peak when there is a greater water volume to provide a driving force. Additionally, in semi-log space, the quickflow component is expected to decay in a linear manner, and the baseflow component is expected to decay in an exponential manner, producing a concave shape on the master depletion curve. A few models proposed for recession coefficients that increase with time include: spring basins with two lithologies of very contrasting permeabilities, which creates a lag-time for the inflow of water to the spring orifice (Soulios, 1991); changes in storage-outflow relations created by temporarily flooded cave passages that are drained by smaller diameter conduits (known as reservoir-constricted flow) (Bonacci, 1993); and water exchange and pressure transfer between the matrix and conduit when antecedent water levels in the aquifer matrix are medium to high compared to those of the conduits prior to rainfall (Bailly-Comte et al., 2010). Whereas there are ample recession methodologies described in the literature, almost all of these approaches are designed for springs that exhibit zero- or first-order decay and are not applicable to recession coefficients that increase with time. To the knowledge of the authors, there has been only one other method proposed for hydrographs that exhibit a convex form. Bonacci (1993) proposed a transfer function for recession coefficients that increased with time. However, this function required either the collection (or assumption) of many hydrological and climatological parameters that are typically not available or cost-effective for most spring-basin studies. A new method is proposed herein that only requires data about discharge over time to simulate the recession. This methodology relies on an adaptation of the recession functions published by Padilla et al. (1994) to handle convexity and requires no assumed variables.