Search

Understanding the nature and unravelling the extents of connections between various components in hydrologic systems have always remained a fundamental challenge in hydrologic research community. The complexity of hydrologic system stems from various aspects, including the constitution of numerous components and sub-components, their direct or indirect internal connections, interactions with climate and ecosystem, and the complicated dynamics due to natural evolution and human activities, among others. Furthermore, the impacts of evidenced warming climate and anthropogenic activities on hydrologic system force us to tackle a host of new considerations into detecting, attributing, and predicting hydrologic processes under nonstationary condition. All these factors pose a significant challenge to system modellers, hydrologists and water resource researchers and emphasize the importance of the application of latest and innovative scientific theories for hydrologic research. In catchment modelling and management applications, a proper understanding of the connectivity of different components is important, for example, rainfall, soil type, land use, slope and finally the catchment response, i.e., floods. Considering hydrologic stations or catchments as networks can lead to new insights and is emerging as an attractive field of research within the scientific community. The main objective of this thesis is to present the strength of network theory in solving hydrological problems by developing and implementing network-based numerical models. The research is divided into four main parts. In the first part, a network-based framework is developed to delineate homogeneous neighbours for ungauged catchment to be used with regional flood frequency analysis (RFFA). The developed framework clearly demonstrates the strength of network theory in identifying meaningful homogeneous region thus improving the accuracy of regional flood quantile estimations. In the second part, a network theory based RFFA approach has been integrated with nonstationary climate conditions and its ability in predicting future flood peaks under warming climate is demonstrated. The developed approach shows clear advantages over the existing nonstationary frequency distribution-based approaches. The third part focuses on investigating the temporal and spatial connectivity patterns of hydrologic variables. By assuming individual locations or timesteps as the nodes of a network, constructed time series are used to form network metrics and to define the strength of connections between nodes and the nature of network structure. Our findings indicate the utility and effectiveness of network theory in exploring and analysing different temporal and spatial connections in hydrology. Finally, the fourth part explores the joint dependence of extreme rainfall events in the space based on the theory of networks, where a novel measure to quantify the possibility of concurrence in complex systems is developed. Results show a weaker spatial dependence under warmer temperatures, however, a stronger dependence at El Niño and La Niña periods.