Preferential flow through weathered bedrock leads to rapid rise of the water table after the first rainstorms and significant water storage (also known as "rock moisture") in the fractures. We present a new parameterization of hydraulic conductivity that captures the preferential flow and is easy to implement in global climate models. To mimic the naturally varying heterogeneity with depth in the subsurface, the model represents the hydraulic conductivity as a product of the effective saturation and a background hydraulic conductivity Kbkg, drawn from a lognormal distribution. The mean of the background Kbkg decreases monotonically with depth, while its variance reduces with the effective saturation. Model parameters are derived by assimilating into Richard's equation 6 years of 30 min observations of precipitation (mm) and water table depths (m), from seven wells along a steep hillslope in the Eel River watershed in Northern California. The results show that the observed rapid penetration of precipitation and the fast rise of the water table from the well locations, after the first winter rains, are well captured with the new stochastic approach in contrast to the standard van Genuchten model of hydraulic conductivity, which requires significantly higher levels of saturated soils to produce the same results. "Rock moisture," the moisture between the soil mantle and the water table, comprises 30% of the moisture because of the great depth of the weathered bedrock layer and could be a potential source of moisture to sustain trees through extended dry periods. Furthermore, storage of moisture in the soil mantle is smaller, implying less surface runoff and less evaporation, with the proposed new model.

The amount of moisture transpired by vegetation is critically tied to the moisture supply accessible to the root zone. In a Mediterranean climate, integrated evapotranspiration (ET) is typically greater in the dry summer when there is an uninterrupted period of high insolation. We present a 1-D model to explore the subsurface factors that may sustain ET through the dry season. The model includes a stochastic parameterization of hydraulic conductivity, root water uptake efficiency, and hydraulic redistribution by plant roots. Model experiments vary the precipitation, the magnitude and seasonality of ET demand, as well as rooting profiles and rooting depths of the vegetation. The results show that the amount of subsurface moisture remaining at the end of the wet winter is determined by the competition among abundant precipitation input, fast infiltration, and winter ET demand. The weathered bedrock retains (Formula presented.) of the winter rain and provides a substantial moisture reservoir that may sustain ET of deep-rooted (>8 m) trees through the dry season. A small negative feedback exists in the root zone, where the depletion of moisture by ET decreases hydraulic conductivity and enhances the retention of moisture. Hence, hydraulic redistribution by plant roots is impactful in a dry season, or with a less conductive subsurface. Suggestions for implementing the model in the CESM are discussed.