AbstractsPhysics

A high-resolution global atmospheric model is used to disentangle the relationship between African easterly waves (AEWs) and Atlantic tropical storms (TCs) from the large-scale environmental factors that may obscure their connection. Since the two most cited references on AEW interannual variability in relation to TC activity draw conflicting conclusions about the historical relationship, and the AEW counts in each study do not show agreement on historical variability, novel analysis procedures are developed to produce consistent AEW and TC count statistics for the historical record using reanalysis products. This reanalysis-derived historical record is used to legitimize the model for the study of AEWs, which is subsequently utilized to investigate the relationship between AEWs and TCs. The internal variability of the relationship between AEW and TC count, including the sensitivity to ENSO phase and annual trends, and the interplay between environmental factors, AEW activity, and TC activity are probed using three sets of simulations: 1) climatological simulations, consisting of three ensemble members forced with historical seasonally and annually varying SST; 2) simulations with interannually invariant forcing, including a control simulation with climatological mean SST and a perpetual La Nina simulation with composite SST from strong La Nina years; 3) perturbed simulations, in which the large-scale environment is drastically altered through the manipulation of African albedo. Since variability exists in AEW count that is unexplained by known indicators of large-scale environmental favorability, across all simulations and multiple timescales, it is unlikely that the ubiquitous covariance between AEW and TC count is simply a response to environmental factors. The statistically significant correlations between AEW and TC statistics suggest that AEW variability accounts for a portion of the observed variability in TC count not due to known environmental factors, since there is unexplained variance in AEW count, and both individual years and aggregated model runs with more (fewer) AEWs also tend to have more (fewer) TCs. It is argued that while half of the covariance between AEW and TC count interannually is mediated by the large-scale environment, the other half can be attributed to stochastic AEW variability.