The water storage capacity of the root zone determines whether plants survive dry periods and controls the partitioning of precipitation into streamflow and evapotranspiration. It is currently thought that top-down, climatic factors are the primary control on this capacity via their interaction with plant rooting adaptations. However, it remains unclear to what extent bottom-up, geologic factors can provide an additional constraint on storage capacity. Here we use a machine learning approach to identify regions with lower than climatically expected apparent storage capacity. We find that in seasonally dry California these regions overlap with particular geologic substrates. We hypothesize that these patterns reflect diverse mechanisms by which substrate can limit storage capacity, and highlight case studies consistent with limited weathered bedrock extent (melange in the Northern Coast Range), toxicity (ultramafic substrates in the Klamath-Siskiyou region), nutrient limitation (phosphorus-poor plutons in the southern Sierra Nevada), and low porosity capable of retaining water (volcanic formations in the southern Cascades). The observation that at regional scales climate alone does not ‘size’ the root zone has implications for the parameterization of storage capacity in models of plant dynamics (and the interrelated carbon and water cycles), and also underscores the importance of geology in considerations of climate-change induced biome migration and habitat suitability.

Water age and flow pathways should be related; however, it is still generally unclear how integrated catchment runoff generation mechanisms result in streamflow age distributions at the outlet. Here, we combine field observations of runoff generation at the Dry Creek catchment with StorAge Selection (SAS) age models to explore the relationship between streamwater age and runoff pathways. Dry Creek is a 3.5 km2 catchment in the Northern California Coast Ranges with a Mediterranean climate, and, despite an average rainfall of ~1,800 mm/yr, is an oak savannah due to the limited water storage capacity. Runoff lag to peak—after initial seasonal wet-up—is rapid (~1-2 hours), and total annual streamflow consists predominantly of saturation overland flow, based on field mapping of saturated extents and an inferred runoff threshold for the expansion of saturation extent beyond the geomorphic channel. SAS modeling based on daily isotope sampling reveals that streamflow is typically older than one day. Because streamflow is mostly overland flow, this means that a significant portion of overland flow must not be event-rain but instead derive from older, non-event groundwater returning to the surface, consistent with field observations of exfiltrating head gradients, return flow through macropores, and extensive saturation days after storm events. We conclude that even in a landscape with widespread overland flow, runoff pathways may be longer and slower than anticipated. Our findings have implications for the assumptions built into widely used hydrograph separation inferences, namely, the assumption that overland flow consists of new (event) water.

Predicting the behavior of overland flow with analytical solutions to the kinematic wave equation is appealing due to its relative ease of implementation. Such simple solutions, however, have largely been constrained to applications on simple planar hillslopes. This study presents analytical solutions to the kinematic wave equation for hillslopes with modest topographic curvature that causes divergence or convergence of runoff flowpaths. The solution averages flow depths along changing hillslope contours whose lengths vary according hillslope width function, and results in a one-dimensional approximation to the two-dimensional flow field. The solutions are tested against both two-dimensional numerical solutions to the kinematic wave equation (in ParFlow) and against experiments that use rainfall simulation on machined hillslopes with defined curvature properties. Excellent agreement between numerical, experimental and analytical solutions is found in all cases. The solutions show that curvature drives large changes in maximum flow rate qmax and time of concentration tc, predictions frequently used in engineering hydrologic design and analysis.