全球变化知识资源中心

This research investigates the influence of decision-maker behavior on policies that are likely to be adopted for the protection of highway infrastructure against inundations resulting from sea level rise. We analyze two different types of games to represent decision-maker behavior, and use the San Francisco Bay Area shoreline with a scenario of a 0.5 m sea level rise as a case study, which is expected in 2054. In our model, the objective of the decision-makers (the counties bordering the SF Bay Area) is to minimize the traffic delay caused by inundations in the transportation network that lies in the geographical boundaries of their counties.

Our model considers hydrodynamic interactions, traffic flow patterns changes as a result of inundations, and budget constraints on the protection costs. The hydrodynamics in the Bay Area are affected by the shoreline protection strategy: protection of the shoreline of a county may lead to increased inundations in another, unprotected, county. Furthermore, closure of a highway link in one county affects traffic delays in other counties due to traffic re-routing. Thus, protection decisions made by a county have potential impacts on several other counties, and therefore counties must take into account other counties' actions.

Both competitive (Nash) and cooperative games are analyzed. A Nash equilibrium is achieved among the counties when each acts to minimize its objective without the possibility of collaborating with other counties. We compare the results of Nash games with possible cooperation strategies for a range of funding scenarios. It is shown, through several examples, that cooperation among counties increases benefits (reduction of Vehicle Hours Traveled) for all participants in most cases. In some cases, cooperation also reduces protection costs. We have identified that there are counties that are not beneficial to protect, in terms of traffic delay minimization, such as Napa, while other counties for which protection is critical, such as Marin. However, these results depend on the formation of the collaborations and the available budget, and therefore our conclusions cannot be generalized.