This thesis aims to further investigate rare natural disasters and studies adaptation decisions under uncertainty by solving several computational economic models. The modeling of rare natural disasters depends on the treatment of catastrophic outcomes with a low probability. The impact of rare disasters on economic activities has been explored in the literature. This thesis is dedicated to answering a question: How do we optimally adapt or alleviate the risk of uncertain rare natural disasters? Uncertainty is central to this dissertation. I overview the sources of uncertainty and present how these uncertainties are addressed in the literature of environmental economics. I demonstrate why rare natural disasters are worth studying and present the advantages that can be gained by investigating them. Subsequently, I present my choice of adaptation measures, spatial adaptation and adaptive capital stock, to better situate my study in the existing literature of adaptation. I investigate adaptation decisions by using an approach based on numerical methods. I employ the existing computational general equilibrium model GENESwIS in Chapter 1. I propose a new simulation approach, the hazard myopia, for a better consideration of spatial adaptation for future high impact floods in Switzerland. The hazard myopic agent solves an intertemporal optimization problem by adopting his subjective belief of the risk of flooding. Capital stocks are recursively updated based on the actual damage. However, my simulation results contradict the real Swiss situation. To correctly handle rare but catastrophic events, one promising approach is to convert the model from deterministic to fully stochastic so that the precautionary saving is endogenized. Why does an economic agent save for a less productive but risk-free capital or a non-productive but adaptive capital stock? Chapter 2 aims to address this question by solving dynamic stochastic equilibrium models. I present the optimal policy functions for spatial adaptation and adaptive capital stock in a stylized but reproducible setting. I quantitatively show that the certainty-equivalent deterministic model underestimates the risk of rare natural disasters. Chapter 3 discusses the qualitative argument established by Schelling (1992, Section IV). I quantitatively demonstrate that the developing economy’s best strategy to adapt to future disasters is to advance its economic development. A developed economy bolsters non-productive yet adaptive capital stock to prepare for future uncertainties. Throughout Chapters 2 and 3, my implementations are massively parallelized on a high- end computation cluster to speed up the solving processes. The efficient usage of parallel computing is an emerging field in economics. This dissertation demonstrates the applicability of exploiting the massive power of modern computing in economics.