干旱区生态环境知识资源中心

The effectiveness of water harvesting systems collecting surface runoff form rangeland hillslopes in semi-arid regions is difficult to predict, since the hydrological response at the outlet depends on the heterogeneity of hydrological processes. The lack of continuous runoff pathways, due to the irregular spatial patterns of soil properties and the variety of antecedent soil moisture conditions directly influence runoff generation and control discharge into the water harvesting cisterns. The aim of this paper is to evaluate the effectiveness of semi-arid hillslopes in generating runoff for water harvesting systems. Runoff was estimated by the STREAM expert-based model which was applied to three semi-arid hillslopes (0.4 to 6 ha). On the one hand the STREAM model rules were adapted to the regional conditions i.e. an antecedent precipitation index was adjusted to local soil moisture conditions and the rainfall duration was defined as the total rainfall event quantity and the effective rainfall duration (P(tot)/t(eff)). On the other hand, the distribution of rock outcrop and vegetation cover along the slopes was used to define homogeneous hydrological units. Final infiltration capacities were attributed to these hydrological units based on values found in the literature. The prediction performances are acceptable for the three water harvesting systems with an RMSE of 13.9 m(3). It was shown that the rainfall/runoff model was more sensitive to the duration of the storm than to the antecedent soil moisture conditions. The use of a unique set of hydrological parameters for the three water harvesting systems on representative hillslopes allows the runoff prediction from any rangeland hillslope within the same region. Furthermore, the spatial patterns of soil surface characteristics are crucial for collecting runoff at the outlet of the system. Model runs demonstrated that degradation of vegetation and sealing of very small areas within flow paths can lead to an increase of annual runoff by as much as a factor two. (C) 2009 Elsevier B.V. All rights reserved.