UBC Theses and Dissertations
Development, evaluation, and application of dominant runoff generation processes in hydrological modeling Rosin, Klemens
Large scale landuse changes have been making the news throughout the world. However, the assessment of landuse change impacts on the hydrological cycle is still a challenging task. Complex hydrological models cannot be applied to watersheds without detailed climate, vegetation, soil, and runoff data. Simple models do not provide sufficient support for spatially distributed landuse management decisions. Therefore, this study presents parsimonious, process-based, spatially-distributed hydrologic models to assess effects of landuse changes on runoff in ungauged basins. The introduced models were based upon the assumption that storm runoff is predominantly generated on certain areas of a watershed. The most commonly used method to predict the runoff generation areas is the concept of dominant runoff generation processes (DRP), which are channel interception, subsurface storm flow, Hortonian and saturation excess overland flow. In particular, forecasts of saturation overland flow generating areas have been controversial in previous research. Therefore, traditional and new soil saturation prediction concepts were evaluated with field data in the first part of this study. Best predictions were found for combinations of topographical indices and groundwater table depths. For three out of four assessed watersheds, optimized model parameters depended on climate. In the second part of the study, DRP model structures were developed. Established DRP area delineation was extended with dynamic process and connectivity modules. The latter were found to improve model fit and parameter feasibility, particularly in process-based DRP models. Temporal connectivity distributions demonstrated that Hortonian overland flow was more affected by connectivity than subsurface flow. Third, the DRP models were used to predict stream flow and effects of landuse changes on peak flow. Stream flow predictions improved if DRP concepts were added to topographical data; however, the additional DRP-based prediction improvement was marginal if climate data were available. DRP-based snowmelt and peak flow predictions agreed well with observed data. The model was used to predict effects of the mountain pine beetle infestation in British Columbia on peak flow in 290 watersheds. Peak flow increases up to 70% were forecasted, and a strong relationship between peak flow increase and landuse change affected area proportion of a watershed was found.
Item Citations and Data
Attribution 3.0 Unported