Investigating Tropical Cyclones and Global Teleconnections Using High-Resolution Climate Model and Reanalysis Data

It is vital to improve the understanding of tropical cyclone variability, as they are powerful, coupled ocean-atmosphere phenomena that move huge amounts of energy across the globe and regularly cause devastating damage. There is only a short period of observations of tropical cyclones, and therefore major uncertainty in variability from interannual (Atlantic) to multi-decadal (Pacific and Atlantic), as well as in longer-term climate. It is not only important to understand tropical cyclones for their direct impact, but also in the context of climate change. Having a better understanding of the role of tropical cyclones in the global climate system as a whole will potentially allow for more accurate predictions of future climate and extreme weather events that have statistically robust teleconnections.

With developments in computing, high-resolution climate models are now able to produce synthetic tropical cyclones, although they still generally underestimate storm intensity. As these models have a physical basis, analysis of these cyclones helps to better understand the processes operating, and by running the models for decades to centuries creates a large sample that aids in reducing the uncertainty of using the relatively short historic (e.g. Best Track) data alone. These climate models are global and therefore are suitable to assess teleconnections between tropical cyclones and other phenomena, such as ENSO.

Alongside climate model data, reanalyses prove valuable as they provide homogeneous data coverage and act as a bridge to simulated stormology. The reanalysis data is constrained by observations, and can therefore be treated as ‘truth'. By applying a feature-tracking algorithm to the reanalysis data, the tracks can be compared to those created synthetically within the model and any biases in these model tracks, such as storm intensity, can be corrected for.

The ERA-Interim reanalysis and the new JRA-55 reanalysis provide 34 and 55 years of data that can be used to extract storms using a feature-tracking algorithm. These datasets are the longest high-resolution reanalyses available and allow the examination of longer-term decadal variations alongside annual variations, such as ENSO and the AMO. The Atlantic is the best-observed basin, whereas the Pacific record is shorter and the data lacks homogeneity. A huge benefit of using reanalysis data is that it is homogeneous across the globe, so that complete and comparable storms, and their evolution, can be captured.

These reanalysis-derived storm datasets credibly reproduce the variability of tropical cyclone activity that is seen in the observations. Within these reanalysis stormologies it has been possible to identify specific storms that have been well recorded in observations, such as Hurricanes Katrina, Andrew, Ivan and Rita. Many storms extracted from the reanalysis data can be followed after they make extra-tropical transition, however, there are some limitations, for example there are too many storms detected in the East Pacific and the tracking is less skilled once the storms make landfall. Comparing these reanalysis storms to those in historic records allows both the data that is comparable to ‘truth' and the tracking algorithm to be assessed and validated. This comparison and validation occurs first in the basin that is well observed, i.e. the Atlantic, following which it is then possible to compare other basins where there is less historical data, such as the Pacific. This enables the development of a sound understanding of the benefits and limitations of the reanalysis data in relation to tropical cyclones and also an assessment of the tracking algorithm used to extract storms.

The tracking algorithm used is then be applied to climate models, creating a long dataset of simulated tropical cyclones that are analysed with an understanding of uncertainties. The superposition of climate modes, such as ENSO, on tropical cyclone activity is examined using the high-resolution climate model data.