HIDDEN MARKOV MODELS OF FUTURE COST OF SEA LEVEL RISE

The total amount of sea level rise (SLR) realized at any particular place on the earth and time in the future is estimated using an array of observations and geophysical models, all of which are subject to uncertainty. A probability density function (PDF) is frequently used to represent future SLR (whether globally or locally), and this PDF includes a spectrum of outcomes including both low impact/high probability and high impact low probability outcomes. The general shape of this PDF is well known, but its specifics are quite uncertain. The cost of any specific future SLR outcome, on the other hand, is studied by economists, but like SLR, it is also modeled and costs attached to any specific SLR outcome can be estimated within certain well-defined contexts. We use Hidden Markov methods to link these two models into a single estimator of future expense, and explore the range of possible outcomes using Monte Carlo methods, with a particular focus on land ice contributions to SLR. Our principle objective is to determine where future SLR research resources should be directed to achieve the most beneficial (least costly) outcome for society: should resources be directed towards better understanding of high-impact but low-probability components of SLR (e.g. collapse of the West Antarctic Ice Sheet) or towards low-impact but much higher probability components (e.g. declining mass balance of global small glaciers)?