Numerical simulation of unsaturated flow along preferential pathways: implications for the use of mass balance calculations for isotope storm hydrograph separation

An objective common to many watershed studies is to separate storm hydrographs into two components: water that was present in the watershed prior to a storm event (soil moisture and groundwater) and water which fell on the watershed during the storm. To use this approach, a number of assumptions must be made including that the composition of water in the soil moisture and groundwater reservoirs are constant and known. The objective of this paper is to show that in settings where flow and transport are dominated by preferential pathways for flow, steady state mass balance calculations for quantitative hydrograph separation may be in error. We present field data from a site where flow and transport are dominated by preferential pathways (relict fractures in saprolite of sedimentary rocks) which indicate that the delta 18O content of the water in the unsaturated and shallow saturated zones is not constant over the course of a storm event. We use a numerical model to further explore the interactions between the fractures and surrounding matrix. Both the field data and modeling results indicate that the delta 18O of the previous storm event(s) has a strong influence on water in the fractures. On the time scale of a storm event, only the water in the matrix immediately surrounding the fracture mixes with water in the fracture, while the bulk of the matrix is isolated from fracture flow. The spatial and temporal heterogeneity of the delta 18O in the subsurface and the isolation of the most of the matrix water from flow in fractures make the measurement of a singular delta 18O value for subsurface reservoirs problematic and the assumption of a constant value doubtful. Since most near-surface geologic materials have preferential flow paths, we suggest that quantitative hydrograph separation using mass balance techniques is not possible in most situations. Future field and modeling investigations using the approach outlined here could be designed to explore the important temporal and spatial scales of variability in watersheds, and lead to a more quantitative approach to storm hydrograph separation.