Atlantic tropical cyclones in the 20th century: Natural variability and secular change in cyclone count
We show that annual TC counts can be more closely reconstructed from Pacific and Atlantic SSTs than SST of the main development region (MDR) of Atlantic TCs; the former accounting for ~60% of the decadal count variance as opposed to ~30% for MDR SST. Atlantic Multidecadal Oscillation (AMO) dominates the reconstruction, accounting for ~55% of the natural decadal count variance, followed by the ENSO Non-Canonical and Pan-Pacific decadal variability contributions. We argue for an expansive view of the domain of influential SSTs – extending much beyond the MDR.
The additional accounting of count variance by SSTs outside the MDR suggests a role for remotely-forced influences over the tropical Atlantic: The Pan-Pacific decadal mode is linked with decreased westerly wind shear (200-850 hPa) in its warm phase, much as the AMO impact itself. Non-canonical ENSO variability, in contrast, exerts little influence on decadal timescales. Interestingly, the secular but non-uniform warming of the oceans is linked with increased westerly shear, leading to off-setting dynamical and thermodynamical impacts on TC activity!
The early-1930s increase in smoothed counts can be partially (~50%) reconstructed from SST natural variability. The 1950s-60s decrease, in contrast, could not be reconstructed at all, leading, deductively, to the hypothesis that it results from increased aerosols in this period. The early-1990s increase is shown to arise both from the abatement of count suppression maintained by SST natural variability and the increasing SST secular trend contribution; the abatement is related to the AMO phase-change in early-1990s. Were it not for this suppression, TC counts would have risen since the early 1970s itself, tracking the secular change contribution.
The analysis suggests that when SST natural variability begins to significantly augment counts in the post-1990 period – some evidence for which is present in the preceding decade – Atlantic TC counts could increase rapidly on decadal timescales unless offset by SST-unrelated effects which apparently account for a non-trivial amount (~40%) of the decadal count variance.