Evaluation of tropical cyclone recurrence rate: factors contributing to epistemic uncertainty

Abstract

In this paper, we identify, assess and characterize epistemic uncertainties associated with the estimation of tropical cyclone storm recurrence rates. To illustrate the potential effects of epistemic uncertainty, we develop a range of storm surge recurrence rates at representative locations along the North Atlantic coast of the USA. We further illustrate how storm rate uncertainty impacts estimates of storm surge hazards. We consider the following factors when evaluating epistemic uncertainty in storm recurrence rate: (1) use of recorded storm track data versus interpolated storm track data, (2) the weighting method used for “counting” historical storms in the vicinity of the representative sites, (3) the kernel size used in the selected weighting method, (4) the intensity parameter used to filter “inconsequential storms” from the historical record, (5) the threshold value associated with the selected intensity filter, (6) the historical time period considered in the evaluation and (7) the impact of coastline geometry definition when partitioning storm populations. We observe that storm rate varies over orders of magnitude when considering these combinations of factors. When considering the effect of epistemic uncertainty in storm recurrence rate on estimated storm hazards, we find that the corresponding range of storm surge elevation spans multiple meters for a given annual exceedance frequency.