Landscape-based hydrological modelling: Understanding the influence of climate, topography, and vegetation on catchment hydrology

Abstract

In this thesis, a novel landscape-based hydrological model is presented that was developed and tested in numerous catchments around the world with various landscapes and climate conditions. A landscape is considered to consist of a topography and an ecosystem living on it. Firstly, the influence of climate on hydrological process was studied. It was assumed that an ecosystem organizes its root zone storage capacity to overcome droughts with a certain return period to survive, but not larger than needed, so as to save energy and nutrients. It was found that ecosystems organize their root zone storage capacity to overcome droughts with a return period of about 20 years. This indicates that the root zone storage capacity is the optimal result of an ecosystem adjusting to the climate. The size of the root zone storage capacity in turn has great influence on virtually all hydrological processes. Subsequently, in order to understand the influence of topography on hydrological behavior, we selected a cold-arid catchment in the upper Heihe River in China as the study site. The influence of topography was explicitly considered in the FLEX-Topo model. Firstly, topography data was applied to make a landscape classification. It is interesting to note that also the land cover map can be derived from topography information, indicating the great influence of topography on natural land cover. And then, within the framework of FLEX-Topo modelling approach, we applied a semi-distributed model structure to describe the different runoff generation mechanisms in different landscapes. We found that FLEX-Topo, allowing for catchment heterogeneity, performs much better than lumped models while transferring both model and optimized parameter sets to two nested catchments. To test the influence of vegetation and topography information on model transferability separately, a stepwise modelling approach was applied in 6 catchments in Thailand. Each model was calibrated on one catchment and then transferred with its optimized parameter sets to the other catchments. During calibration, all models exhibited similar skill to fit the hydrographs in all catchments. However, when transferred, the performance of the lumped model reduced dramatically, because it did not explicitly consider vegetation and topography. It was shown that individually, vegetation and topography helped to improve model transferability, especially in catchments with heterogeneous landscape composition. The likely reason is the co-evolution of topography, vegetation, soil and hydrology. Finally, a glacier subroutine was added to the FLEX-Topo modelling framework to improve its applicability in glacierized catchments, where glacier melt water is an essential water resource for downstream. The simulated results were not only validated by hydrographs as in other studies, but also by the long-term glacier mass balance and snow water equivalent data. Furthermore, after considering the different proportion of different landscapes, the glacier model could be successfully upscaled and transferred to a larger catchment. This further validates our proposed modelling concept, the process equations and parameterization.