Climate is changing. It is widely accepted that irrespective of the emission reductions efforts, adaptation to the already committed climate change is a must in the coming decades. Flooding is one of the most devastating climate-induced hazards, calling for adaptation across scales: from government-led adaptation (e.g. dikes), to personal household-led adaptation. There are numerous private adaptation measures, both structural and non-structural, that households can autonomously take to reduce damages and speed up own recovery, in case of the adverse event does occur. Worldwide household surveys provide the empirical evidence suggesting that the appraisal of fear, perceptions of own coping abilities and social influences affect individual decision to adapt. Especially the latter appear as a strong determinant of household-led adaptation. Yet, the role of social networks has not been studied systematically, let alone the exploration of the interplay between private households' adaptation and public policies in the presence of social influences. To increase the understanding of social networks' impact on private flood adaptation under various public policies, an empirical agent-based model is built. In addition to empirical flood maps, I employ Protection Motivation Theory and households' survey data from Houston (Texas, USA) to capture the household’s adaptation decisions, which are subject to social influence. To represent a range of social networks, I explore the effect of three random networks - Erdös-Renyi random network, Barabasi-Albert scale-free network, Watts-Strogatz small-world network -- on the diffusion of flood preparedness. These three networks could serve as a proxy of diverse types of social relationships that vary across cultures, with some being hierarchical and others egalitarian. Furthermore, I test how the diffusion of private adaptation among households evolves under four generic public policy strategies: protection by publicly-funded infrastructure, market-based policy such as subsidy, information policy and regulatory policy.
The thesis findings reveal that the type of social networks influence success of private adaptation decisions significantly. Notably, the effect of social norms depends on the network configuration, where more homogeneous networks lead to higher private adaptation uptake. Furthermore, the thesis quantifies the interaction effects between social networks and policies designed to support private climate change adaptation, for example communication campaigns affecting individual risk perceptions that in turn drive household-led adaptation. Therefore, information policies can be very effective in steering public opinions and, hence, the uptake of households' adaptation measures. Moreover, I find that in the presence of social networks, public adaptation policies interact with each other: a combination of subsidies with communication exhibit synergistic effects on the flood damage reduction from private adaptation. Policies that do not interact with households' perceptions, like infrastructure projects or laws, can be considered more robust - less sensitive to social networks. To conclude, networks can shape the success of climate change adaptation policies and should be accounted for in their design. Further research should focus on the endogenous evolution of social networks and the model behavior under changing flooding conditions and what this complex adaptive system's dynamics means for future cross-scale adaptation efforts.

Climate change intensifies the frequency and severity of floods - the most devastating and costly climate-induced hazard. Government adaptations such as dikes or beach nourishments are important, yet insufficient in the face of worsening hazards. They reduce the hazard probability, but the local actions at the household level determine the extent of damages and inequalities in its distributional impacts on various societal groups. To design effective policies and risk reduction strategies it is critical to quantify the speed and scope of the household adaptation uptake and the resulting damage prevention. This is especially the case for distributional impacts (i.e., how different societal household groups adapt), which are often neglected. Quantifying both aggregate and distributional impacts of household climate change adaptation (CCA) on flood risk fosters the design of tailored flood risk management policies which allocate resources to the societal groups that need them most. This thesis presents a state-of-the-art agent-based flood model in downtown Shanghai, China to understand the role of household CCA in reducing coastal flood risk both in its aggregate and distributional impacts. To model the households’ exposure to climate-induced floods we overlayed the geolocations of 18.039 residential buildings with 21 inundation maps that depict dike failures and dike overtopping under different climate-change scenarios. We further parameterized households using context-specific micro-level survey data from Shanghai and depicted households’ adaptation decisions using an extended version of the Protection Motivation Theory. Our results show that autonomous household adaptation (adaptation without government policy) to climate change plays an essential role in reducing flood damage in downtown Shanghai. However, despite the considerable adaptation uptake, the residual damages increase over time due to the effects of sea level rise and land subsidence. This shows that autonomous household adaptation alone is insufficient to keep pace with the increasing severity of climate-related flooding, as it is constrained by barriers in the form of adaptation measure costs and regulations. Thus, additional policies are needed to overcome these barriers. These policies should take into account differences in adaptation behaviour and damage prevention among societal household groups. Specifically, our results indicate that households with lower worry, self-efficacy, and income adapt measurably slower to climate-induced floods, which makes these household groups significantly more vulnerable to flooding.