Threshold behavior of stormflow response is an emergent pattern observed in several studies demonstrating subsurface storage controls on catchment rainfall‐runoff dynamics. These studies demonstrate a distinct transition from negligible stormflow discharge response to rapid, linearly increasing stormflow identified by a single, uniquely defined threshold as a basic catchment attribute that relates to geophysical properties. Utilizing precipitation, streamflow, and soil moisture data spanning 15 years from three catchments at the Coweeta Hydrologic Laboratory (CHL), we analyze how threshold behavior forms and varies at several timescales. We pose three hypotheses: (1) stormflow thresholds form at CHL as a function of antecedent soil moisture and gross precipitation, (2) thresholds vary seasonally and interannually, and (3) threshold variation through time implies greater long‐term complexity of runoff controls beyond catchment geophysical properties, including forest canopy ecohydrologic feedbacks. We isolate threshold behavior of stormflow using piecewise regression analysis in short to long‐term data sets with respect to antecedent soil moisture index and gross precipitation. We use this to investigate threshold variation over seasonal, interannual, and decadal timescales that encompass hydroclimatic extremes. Seasonal analysis reveals that thresholds are more variable between growing seasons than between dormant seasons. In growing seasons with greater water stress, stormflow thresholds are lower after controlling for soil moisture storage suggesting more complex, long‐term rainfall‐runoff relationships as a result of forest canopy response to water stress. We present a conceptual model of how vegetation‐climate interactions influence long‐term rainfall‐runoff relationships creating interannual variability of stormflow thresholds and linear stormflow response.

In headwater catchments, streamflow recedes between periods of rainfall at a predictable rate generally defined by a power–law relationship relating streamflow decay to streamflow. Research over the last four decades has applied this relationship to predictions of water resource availability as well as estimations of basin‐wide physiographic characteristics and ecohydrologic conditions. However, the interaction of biophysical processes giving rise to the form of these power–law relationships remains poorly understood, and recent investigations into the variability of streamflow recession characteristics between discrete events have alternatively suggested evapotranspiration, water table elevation, and stream network contraction as dominant factors, without consensus. To assess potential temporal variability and interactions in the mechanism(s) driving streamflow recession, we combine long‐term observational data from a headwater stream in the southern Appalachian Mountains with state and flux conditions from a process‐based ecohydrologic model. Streamflow recession characteristics are nonunique and vary systematically with seasonal fluctuations in both rates of transpiration and watershed wetness conditions, such that transpiration dominates recession signals in the early growing season and diminishes in effect as the water table elevation progressively drops below and decouples with the root zone with topographic position. As a result of this decoupling, there exists a seasonal hysteretic relationship between streamflow decay and both evapotranspiration and watershed wetness conditions. Results indicate that for portions of the year, forest transpiration may actively compete with subsurface drainage for the same water resource that supplies streamflow, though for extended time periods, these processes exploit distinct water stores. Our analysis raises concerns about the efficacy of assessing humid headwater systems using traditional recession analysis, with recession curve parameters treated as static features of the watershed, and we provide novel alternatives for evaluating interacting biological and geophysical drivers of streamflow recession.

Linking quickflow response to subsurface state can improve our understanding of runoff processes that drive emergent catchment behaviour. We investigated the formation of non‐linear quickflows in three forested headwater catchments and also explored unsaturated and saturated storage dynamics, and likely runoff generation mechanisms that contributed to threshold formation. Our analyses focused on two reference watersheds at the Coweeta Hydrologic Laboratory (CHL) in western North Carolina, USA, and one reference watershed at the Susquehanna Shale Hills Critical Zone Observatory (SHW) in Central Pennsylvania, USA, with available hourly soil moisture, groundwater, streamflow, and precipitation time series over several years. Our study objectives were to characterise (a) non‐linear runoff response as a function of storm characteristics and antecedent conditions, (b) the critical levels of shallow unsaturated and saturated storage that lead to hourly flow response, and (c) runoff mechanisms contributing to rapidly increasing quickflow using measurements of soil moisture and groundwater. We found that maximum hourly rainfall did not significantly contribute to quickflow production in our sites, in contrast to prior studies, due to highly conductive forest soils. Soil moisture and groundwater dynamics measured in hydrologically representative areas of the hillslope showed that variable subsurface states could contribute to non‐linear runoff behaviour. Quickflow generation in watersheds at CHL were dominated by both saturated and unsaturated pathways, but the relative contributions of each pathway varied between catchments. In contrast, quickflow was almost entirely related to groundwater fluctuations at SHW. We showed that co‐located measurements of soil moisture and groundwater supplement threshold analyses providing stronger prediction and understanding of quickflow generation and indicate dominant runoff processes.

Preferential flow in hillslope systems through subsurface networks developed from a range of botanical, faunal and geophysical processes have been observed and inferred for decades and may provide a large component of the bulk transport of water and solutes. However, our dominant paradigm for understanding and modelling hillslope hydrologic processes is still based on the Darcy–Richards matric flow framework, now with a set of additional methods to attempt to reproduce some of the aggregate function of the two‐phase system of network and matrix flow. We call for a community effort to design and implement a set of well planned experiments in different natural and constructed hillslopes, coupled with the development of new theory and methods to explicitly incorporate and couple the co‐evolution of subsurface flow networks as intrinsic components of hydrological, ecological and geomorphic systems. This is a major community challenge that can now benefit from new experimental infrastructure, renewal of older infrastructure and recent advances in sensor systems and computational capacity but will also require a sustained and organized interdisciplinary approach. Copyright © 2014 John Wiley & Sons, Ltd.

No abstract