Evapotranspiration is an important water budget component of many large-scale eco-hydrological processes, especially in semi-arid and arid environments. The subsurface hydraulic and root properties, which largely determine partition of plant transpiration and soil evaporation, show a large degree of spatial variability. Large scale eco-hydrological processes are usually quantified using multiple realizations of local moisture flow in independent one-dimensional homogeneous columns. A major problem of this so called stream tube approach is that it ignores the interactions among different stream tubes. Lateral flows might be prominent for long and narrow tubes and heterogeneous hydraulic properties and plant covers. This study is to quantify possible errors of using the stream tube approach in large scale hydrological simulations. Instead of using convenient parallel column models of independent eco-hydrologic processes, this study utilizes a multi-dimensional modeling approach (HYDRUS 2D/3D) to simulate two-dimensional eco-hydrological processes using multiple parallel columns which allow lateral interactions.
Results/Conclusions
We investigated various combinations of plants and bare soils. This study demonstrated that the accuracy of stream tubes approach is controlled by the degree of spatial variability of soil hydraulic properties and initial moisture condition for evapotranspiration. Under the conditions when horizontal moisture flux is significant, the stream tube approach would introduce large unacceptable errors. In particular, for bare soil-plant combination, the stream tube approach may generate up to 50% of relative errors in simulating transpiration at early stage and about 30% of relative errors at late stage of eco-hydrological processes in simulating transpiration. It is also found that the relative errors may become large under wet initial soil moisture conditions; for plant-plant combination, the accuracy of stream-tube is mainly controlled by the differences in hydraulic properties. In general, the error generated by stream tube approach in simulating evaporation is much smaller than that of transpiration.