Identifying Subsurface Connectivity From Observations: Experimentation With Equifinality Defines Both Challenges and Pathways to Progress

Abstract

Linkages between landscapes and streams are increasingly described in terms of hydrological connectivity. The ability to effectively distinguish different patterns of water movement through catchments makes connectivity particularly interesting to both scientists and practical water managers. Hydrometric data (groundwater levels, soil moisture and streamflow) are often employed to infer the connection between the landscape and its drainage network. Such observational data, however, are insufficient to infer subsurface connectivity in humid settings with perennial stream flow, due to the risk of equifinality. To quantify how much subsurface flow patterns can differ and still be consistent (equifinal) with comprehensive observations of hillslope groundwater levels and stream runoff (the hydrometric data), this study used a modelling experiment based on a well-characterised field site. Particle-tracking simulations at different flow rates defined the water flow paths and transit times of two virtual hillslopes that differed profoundly in the vertical distribution of the saturated hydraulic conductivity. Even though the simulated weekly stream flows and groundwater levels were similar (i.e., the hillslopes were hydrometrically equifinal) particle velocities and water ages at specific locations along these hillslopes differed by orders of magnitude. Flow path lengths and catchment transit times varied up to several 100%. The hillslope- and stream-based metrics used to describe connectivity also varied with stream flow rates. These results underline the need to recognise the risks for equifinality when inferring subsurface connectivity from hydrometric observations alone, even when those observations are comprehensive. The results also highlight the value of model simulations for quantifying the uncertainty in the inferred connectivity, targeting the best sampling locations/times to reduce this uncertainty with tracer data and better understanding the way connectivity influences stream chemistry.