Role of vadose-zone flow processes in regional-scale hydrology: review, opportunities and challenges

At the regional scale, vadose-zone processes are recognized for controlling both short-term dynamics in watershed hydrology and long-term water balances of hydrologic basins. In this paper we explore the various conceptual and mathematical models that have been proposed or could be considered to represent water fluxes in the vadose zone at the catchment, watershed or regional scale. Such models have been in existence in two largely disconnected disciplines: on the one hand, watershed hydrologists and, more recently, climate modelers frequently conceptualize the vadose zone as a zero-dimensional black box represented by lumped parameter models. On the other hand, soil physicists, equipped with tools to measure system and systemstate properties directly in the vadose zone at scales of 10 – 10 m, have relied on Richards’ equation, a physically based, fully parameterized four-dimensional space– time model to represent unsaturated flow at the laboratory or field-plot scale. Over the past thirty years, the modeling efforts of the two disciplines have increasingly converged: hydrologists downscale their models by employing distributed (rather than lumped) models of varying complexity, while soil physicists have employed stochastic methods to upscale from their local-scale measurements and their localscale physical understanding of flow processes to the field and regional scale. The lead question in this work is how the typical small-scale vadose-zone measurements relate to the large-scale representative or ‘effective’ parameter values of variously complex regional vadose-zone models. Recent advances in both, downscaling (from the regional scale) and upscaling (from the laboratory scale) and the use of inverse models have led to promising tools. As a result, at the regional scale, the Richards’ equation and some of its simplifications, but also mass-balance and storage-based bucket models have been employed to represent spatially distributed unsaturated flow. All of these approaches have been employed with some success and under typically rather restrictive assumptions, whereby the least complex models seem to apply exclusively to the largest (and smallest) spatial and temporal scales. Various stochastic analyses have shown that simple averaging of local-scale measurements across the regions is associated with significant errors. Inverse modeling has relied on a priori assumptions about the physical framework that can be tested a posteriori. Both, downscaling and upscaling, regardless of the approach, yield increasingly complex models as they move from their opposing and well-understood starting points towards a unified mathematical representation that appropriately spans the hierarchy of significant process scales. To date, a physically and geostatistically Department of Land, Air, and Water Resources/Hydrology, University of California, Davis, CA 95616-8628, USA. E-mail: [email protected] and [email protected]