Study On Spatial Averaging Watershed Hydrological Model At Macro-scale Based On Probability Approach

The current generation of physically based distributed hydrological models is based on the point scale equations. They have distinct advantages compared with the conceptual models, but also suffer from serious shortcomings, of which the essential one is the inconsistence between the equation applicable scale and model application scale. Such scale inconsistency makes the physical meaning of the equations as well as the parameters uninterpretable. To address this problem, Kavvas et al. proposed a new hydrological modeling approach which upscales the point scale equations to macro scale through the spatial averaging based on the probability description of landscape heterogeneity.The mathematical equations controlling the hydrological processes serve as the basis of watershed hydrological modeling. As the first attempt of hydrological modeling based on spatial averaging approach, the equations and hydrological model developed by Kavvas et al. have obvious deficiencies such as ignoring of soil properties variance in vertical direction and over-demanding of data. In this thesis, the equations controlling the major hillslope processes have been improved: the spatial averaging infiltration model for layered soil (SAI model) and spatial averaging interflow model (SAIF model) are developed, which incorporating spatial heterogeneity of soil properties in both horizontal and vertical aspects. Besides, the new surface flow model for the hillslope with rill-interrill configurations is developed by simplifying Yoon's model so that the parameters can be derived from digital elevation model directly.Based on macro-scale equation controlling hydrological processes, a new watershed hydrological model, Spatial Averaging Hydrological Model (SAWH model), is developed. The mathematical description of watershed heterogeneity is improved and the parameters estimation is significantly simplified and can be easilier applied compared with former models based on spatial averaging approach, i.e. WEHY model. In the SAWH model, the watershed is divided into model computation units (MCU) according to topograph, and the MCU is then partitioned to several sub-regions according to land use, the heterogeneity of soil hydraulic conductivity and micro topograph is descriped by propobility density function and incorporated into the upscaled equations. The inconsistency between the equation applicable scale and the model application scale is avoided through using the upscale equations, while the physically meaning of point scale equations and their parameters are preserved.The SAWH model is applied at two typical watersheds with different climate, one is Qin River basin with semi-arid climate and the other is Baron Fork River basin with humid climate. The model can capture the observed streamflow pattern with high computation efficiency which suggests the macro-scale spatial averaging approach can serve as an alternative approach for continuous hydrological modeling in large watershed.The analysis of model parameter sensitivities in Qin River basin shows the heterogeneity of soil hydraulic conductivity have significant influence to the watershed hydroligcal responses.The uncertainty of SAWH model is further explored by analyzing two typical flood events in Baron Fork River Basin. The radar based rainfall data is compared with the rain gauge based data for analysis of temporal uncertainty of rainfall data. Also, the spatial uncertainty of rainfall data and its influence on SAWH predictibility are analyzed by comparing different rainfall data resolution. Finally, the model uncertainty caused by model parameters is explored by GLUE approach.