Flood analysis in the Wei River Basin

Gai, Lingtong


Floods are considered to be one of the most destructive natural hazards on Earth. Apart from precipitation as the main driving factor for flood occurrence, there is increased recognition that human activities also largely influence the occurrence and impacts of floods. The Wei River Basin in China has suffered from floods for decades due to its monsoon climate with intensive summer rainfall. With the rapid development of the economy and population growth, land use and the water cycle have been altered due to the demands for resources redistribution. Consequently, the Wei River Basis is under pressure from both climate and anthropogenic changes. This thesis aims to bring a comprehensive understanding of the factors causing flood occurrence in the basin and propose an integrated management tool for flood analysis under the complex environment in the Wei River Basin.

Chapter 1 describes the general factors and processes leading to flood occurrence with respect to meteorological, biophysical and anthropogenic aspects. Chapter 2 starts with the introduction of a framework approach for understanding these factors and their importance regarding contribution to flood occurrence in the Wei River Basin, and then presents the results of our application of this approach. The framework approach uses a set of methods to answer the questions why, where, when, and how flooding occurs and includes Conditional Inference Tree (CIT), cross correlation and double mass curves analyses. The results revealed that the dam construction period was the most important factor (why), and the Western upstream regions of the Wei River contributed the most to the flooding of the downstream floodplain (where). The effect of the periods of dam construction on the time lag change (when) and the precipitation-discharge relationship (how) were also analysed by the cross-correlation analysis and double mass curves analysis, respectively. Being able to bring both numeric and non-numeric factors into the analyses, the CIT analysis proved to be a powerful tool for unravelling complex causes leading to flood occurrence. The insights gained in this chapter can be further applied to understanding of flooding schemes in other regions and to derive targeted flood mitigation measures in the Wei River Basin.

LISFLOOD modelling was then calibrated and validated, the results of which are presented in Chapter 3. The urgent need for the development of an integrated approach to evaluating the impacts of climate change, land use change and river alteration as well on the occurrence of hydrological extreme events drove us to choose a physically-based distributed hydrological model as a solid base for accurate discharge simulation. Using globally available land cover, soil, vegetation as well as geographical datasets, combined with local observed meteorological gauging data as input, the application of the LISFLOOD model performed well in simulating the discharge at the outlet station on the floodplain. Being a distributed model, LISFLOOD enabled prediction of discharge at the outlets of 17 tributaries draining into the main river. The simulated discharge from these outlets were analysed regarding their contribution to the total runoff as well as for individual flood events. This study is the first application of the LISFLOOD model for a semi-arid region in China for flood discharge analysis, and has shown to be a sophisticated and reliable tool for catchment scale land and water management planning related to flood occurrence and dynamics.

Chapter 4 reports on the evaluation of the usability of various freely available datasets for discharge simulation with LISFLOOD. The quality of the meteorological data inputs into the hydrological model is of vital importance for understanding the hydrological regime as well as for analysis and prediction of hydrological extremes. Many efforts have been made to develop globally freely available meteorological reanalyses data especially for data scarce regions. We evaluated ten globally freely available datasets for discharge simulation by using them as input for the LISFLOOD model and comparing results between and with observed meteorological data and discharge in Chapter 3. The result was rather disappointing and suggested that none of the evaluated datasets can be applied directly for daily discharge simulation by the LISFLOOD model. This is unfortunate as accurate simulation of discharge is essential for flood analysis under climate change circumstances. An in-depth analysis of the performance of precipitation and temperature data against the observations was then conducted in an attempt to improve the simulation of the datasets.

In Chapter 5 the impact of anthropogenic changes in the basin on flood peak discharge is investigated. It is increasingly recognized that the effects of flood events are greatly influenced by the changes that humans have imposed on the environment and river systems. In the Wei River Basin of China there have been tremendous changes including land use and soil and water alterations under the pressure of population growth and water resource scarcity. To investigate the potential of LISFLOOD for assessing the effects of anthropogenic activities on base flow and peak flow, three categories of scenarios regarding human intervention in the basin were evaluated against a business as usual scenario using the simulated discharge from the LISFLOOD model: 1) natural conditions of the basin, 2) additional reservoirs constructed in different sub-catchments, and 3) water transfer from an adjacent catchment via a pipeline providing a fixed daily inflow. The results of the scenarios are presented for three strategically important cities located on the floodplain. Compared to the business as usual case, the minimum base flow at the three cities increased 54 fold on average with additional dam/reservoir construction, and 41 fold with the pipeline scenario, while with the 1980 land use scenario minimum base flow decreased by 0.8 times. Regarding peak flows, additional reservoirs could reduce them and the water transfer plan would not increase them. The results for the scenarios with the application of the LISFLOOD model cross validated the feasibility of using a modelling approach for catchment flood discharge management as well as providing insight for future policymaking processes.

Last but not the least, Chapter 6 concludes the whole thesis with discussion of the findings from different perspectives as well as provision of outlooks and recommendations regarding the application of hydrological modelling for evaluation of scenarios for regional flood mitigations and related societal impacts. This thesis has increased knowledge and furthered the science regarding flood analysis in the Wei River Basin and how to address the increasing pressure from both climate and anthropogenic changes.