Accurate tropical cyclone forecasting faces a number of challenges. On the shortest time-scales, changes in the environment that a storm travels through can result in dramatic changes in storm track and intensity. Examples of these vortex/synoptic-scale interactions include variability in local vertical shear and/or SST, trough interactions, and changes in the ambient steering flow. Additionally, changes in the internal vortex-scale structure can cause dramatic changes in the evolution of intensity and the distribution of surface winds. Examples of this are eyewall replacement cycles, the formation of annular hurricanes, potential vorticity mixing events, and the formation of eyewall mesovortices. Accurate numerical forecasting of the storm's environmental conditions has been quite successful, but simulating and forecasting internal storm variability requires very fine model resolution and initialization fields, which presents challenges to our operational capabilities.

On seasonal time-scales, forecasting overall tropical cyclone activity relies on a number of documented relationships with local and non-local patterns of environmental variability. These relationships vary between the ocean basins. Examples of this are ENSO, which relates to an increase in activity in the Pacific basins and a decrease in the Atlantic, and the strength and positions of the regional subtropical highs. Other modes of variability that can be related to tropical cyclone activity are the AMO, Tropical Multidecadal Modes (TMM), NAO, and the Atlantic Meridional Mode (AMM). All of these modes may provide some skill to seasonal forecasts of tropical cyclone activity, but most of the variance is left unexplained, and much of the inter-annual variability is likely to be random noise.

On decadal and longer time-scales, forecasting tropical cyclone activity is challenged on a number of fronts. Marked variability among differing numerical models introduces uncertainties in our ensemble-based forecasts, parameterization of physical processes can unphysically prejudice the model results, and uncertainties in the long-term tropical cyclone data records can reduce the relevance of the model hindcasts that we rely on to gain trust in what the models are telling us. Additionally, there is little theoretical guidance with which to interpret the model results. The theory of Potential Intensity is constrained to local thermodynamic conditions, but much of the variability of tropical cyclone activity results from larger-scale circulation variability that does not have such a clear theoretical underpinning.

This talk will focus on some of the details of these forecasting challenges, and some of the research being done to mitigate them.