Roots, storms and soil pores: Incorporating key ecohydrological processes into Budyko's hydrological model

Due to its role in intercepting and evaporating water, vegetation is a key medium through which catchment water yields can be manipulated and may prove to be an important tool available to managers for mitigating the effects of climatic changes on yields. Understanding of the effects of vegetation on long-term catchment flows is growing, yet incorporation of these effects into hydrological models has typically been empirical and qualitative, or deterministic and quantitative - but at the cost of simplicity. Here we present a new ecohydrological model (the Budyko-Choudhury-Porporato, or BCP, model) that is the combination of two pre-existing models. It is simple, quantitative and process-based. As well as accounting for the well established roles played by average precipitation (P) and potential evaporation (E-p) on long-term catchment flows (R), the BCP model also provides estimates of n - a catchment-specific model parameter that alters the partitioning of P between R and evaporation - which is estimated as a function of plant-available soil water holding capacity (kappa), mean storm depth (alpha) and effective rooting depth (Z(e)). After developing this model and testing results at several spatial scales across Australia, we used the model to analyse the sensitivity of R to changes in Z(e) and alpha (as well as P and E-p,) in the high yield zones of the Murray-Darling Basin. The model performed well overall at both the coarse spatial scale (i.e., the Basin) and at the finer scales of sub-basin regions and of catchments, capturing similar to 90% of the observed variability in R. At the regional scale, BCP-modelled R was more or less the same as when a default (constant) value of n was used to model R (that is, n = 1.8). This is due to the general aridity of these regions (under such conditions R is insensitive to small variations in n). However, at the catchment scale, use of the BCP model in 'wet' catchments (i.e., R > 500 mm y(-1)) increased the accuracy of predictions by approximately 10% compared to the default (n = 1.8) model. Runoff was shown to be highly sensitive to changes in Pin the highest yield zones of the Basin, followed by changes in Z(e) and then in alpha. In contrast, across the whole basin - which is highly arid - the sensitivity of R to changes in Z(e) and alpha, whilst small in absolute terms, are substantial relative to total basin flows. The incorporation of alpha, kappa and - most importantly from a management perspective -Z(e) into the Budyko model provides a simple and transparent tool for aiding the understanding of the long-term ecohydrological behaviour of catchments and assessing the potential hydrological effects of different land management options under a variable climate. (C) 2012 Elsevier B.V. All rights reserved.