This thesis presents a combined experimental and numerical modelling investigation on the mechanism of rainfall runoff generation in a terrace hillslope system. The research focused on the runoff generation caused by saturation excess under the influence of terrace-induced slope variations. The objective was to provide better understanding of the involved hydrological processes by examining the relationships amongst the key factors, including the development of the subsurface saturation area, subsurface flow, infiltration and runoff initiation. A series of laboratory scale experiments with a high conductive medium and numerical modelling, using the MODHMS model, were conducted to achieve the goal.
The experimental and numerical results show that at a small scale, there was no significant difference in the way runoff was produced between the terraced and uniform (non-terraced) slope cases. The response of soil water pressure head indicated two stages of wetting. After the rainfall started, the soil water pressure increased initially but remained below zero. This was then followed by a period of constant pressure level. In the third stage, the soil water pressure increased to a new, positive constant value, which indicated a fully saturated soil condition. The condition lasted until the cessation of the rainfall event. The length of the first constant soil water pressure period was reduced with the soil depth, suggesting that the soil was saturated from below. The tensiometers deployed at the slope base showed that the wetting front moved up-slope from the down-slope end. The data showed that the soil was initially wetted (but not to saturated level) from the surface, the water accumulated and formed a saturated zone at the slope toe, and then this saturated zone expanded in a radial fashion upwards the slope. The laboratory observations were confirmed by the numerical results, which revealed further that the surface water ponding occurred firstly on the lower part of the uniform slope, where the subsurface saturated zone intersected the ground surface, but not from the slope outlet at the lower boundary where more water infiltrated the soil. In the case of terraced slope, the water started ponding on the terrace bed. The inter-flow was the dominant process, accounting for 81% and 82% of the total discharge from the terraced and uniform slope experiments, respectively. A relative large amount of water was stored in the soil. The subsurface flow strongly controlled the infiltration and consequently affected the surface runoff generation. Thus, the resulting surface runoff production was similar in both slope cases.
Although the overall runoff response was identical in both terraced and uniform slope systems, the change in the surface topography caused by terrace clearly produced localised effects on the subsurface flow field. The numerical simulations showed that as the result of an increase in the slope of the terrace riser, the underneath groundwater flow was diverted more toward the down-slope direction whereas under the horizontal terrace bed, the flow was vertical. This flow pattern created in the near surface soil layer two distinct zones: the soil in the inner part of the terrace was wetter due to 'flow concentration' while the soil under the outer part of terrace can be characterised as 'flow divergence'. Such effect was not transmitted down to the lower soil profile. For relative short slopes adopted in the current experiments, the infiltration was not uniform along the slope length. Particularly in the case of terraced slope, the infiltration showed a complex pattern, including exfiltration on some parts of the terrace where the slope was interruptedly changed, i.e. from the terrace riser to terrace bed, and then to terrace riser, and back to the original slope. The overland water depth profile demonstrated that more water was temporally held on the terrace bed during runoff production; however this did not make any significant difference in the runoff compared with the uniform slope experiments. The experimental results also showed the effects of the initial moisture content and its distribution on the responses of soil water pressure and runoff. Under a wetter condition, the responses were more rapid. In addition, the vertical boundary at the down-slope end was shown to have a strong effect on the subsurface flow and thus the infiltration.
While future investigations are needed to examine further the effect of terrace on runoff and discharge from a hillslope system, the localised effects of the terrace have been clearly demonstrated and should be taken into account in future studies on the same system, including nutrient and pollutant transport. The dominance of through-flow, which was caused by the high conductivity of the medium as well as the large storage capacity of the soil (related to the soil porosity and thickness), can be potentially important factors that led to lessening the effect of terrace on surface runoff generation. These factors are in need of further investigation. The artefact of the vertical boundary should also be eliminated in future experiments by setting more realistic conditions or considering a longer slope. Improvements can be made in future works to include vegetation and better representation of spatial variation of soil hydraulic properties.