|Thesis abstract: |
Water resources management in a changing environment, due to ongoing changes in climate and the society, is a major challenge to modern water science. Most of the hydrological models developed in the last decades and ordinarily used to inform decision-makers and operators focus on the natural component in the water cycle whilst human activities are predominantly regarded as external forces that slightly impact the hydrological system. Yet, the human signature is the main signal component in many river basins worldwide and strongly impacts our ability to predict the future evolution of these systems. In other word, human beings are no more exogenous players with respect to the hydrological system, but co-evolve with it (Sivapalan, Savenije, & Blöschl, 2012; Wagener et al., 2010). Characterizing and modelling this co-evolution is key to building reliable medium-to-long term projections of the main hydrological variables and ultimately to design management and adaptation strategies to mitigate water stress and crises. My major research focuses on developing a decision-analytic framework (Water Management under Change (WMuC)) to analyse and characterize co-evolutionary water resources systems (descriptive component) and to design and assess alternative management strategies (prescriptive component). The objective is to build a mathematical model of the co-evolving natural and human derived processes under changing climate and socio-economic conditions. The model will integrate traditional hydrological process equations for the natural component and behavioural models to characterize the human decision-making processes and their effects on the water cycle. Different climate and socio-economic scenarios will be used as boundary conditions. The behavioural modeling component is the core topic of the research. Multi Agent Systems (MAS) are the state-of-the-art tools to characterize heterogeneous human agents, where agents¿ behaviors can be modeled explicitly or implicitly. In explicit model, agents make decisions by following the predefined rules with ¿if¿then¿ alike structure without necessarily approaching certain optimal assessment, e.g. maximizing revenue (Berger, 2001; Ducrot, Le Page, Bommel, & Kuper, 2004; Manson, 2005). A potential limitation of explicit models is that any change in the rules cannot be captured by the model due to the fixed decision-rule structure.
In this research we will explore implicit behavioural models, where each agent¿s rule is obtained by solving an optimization problem, where the agent behavior is formalized by an operating rule mapping key information (e.g. expected income, expected water availability) into the agent¿s decisions and the objective function is intrinsically a utility function. Reciprocal influence between different classes of agents (e.g. farmers and water regulators) may also occur due to the interaction, communication and coordination behaviors.
The decision-analytic framework will be demonstrated on one or more case studies depending on resources and data availability. One case study will be surely represented by the Lake Como water systems. Here water supply and water demand are determined by the interaction between the lake operator and multiple farmers. By implicitly modeling farmers¿ decisions and activating the reciprocal information exchange, we are allowed to analyse holistically the system behavior as well as its evolution under changing environment.