AEW vorticity centres (via stream-function calculation,
essentially eliminating non-divergent flow) at 700 hPa for
West Africa (0–30◦N, 30◦E–30◦W). Berry et al. (2007)
developed a method to decompose the calculated streamfunction
vorticity into shear and curvature components,
and to use the westward advection of curvature vorticity
to identify AEW trough and ridge regions. In this
methodology, trough or ridge axes were identified where
westward advection of curvature vorticity is equal to zero.
The same curvature vorticity threshold (0.5 × 10−5 s
−1)
suggested in Berry et al. (2007) was used to distinguish
between ridge and trough classifications (following manual
inspection of multiple individual time steps, not shown).
Removal of pseudo-troughs (-ridges) resultant from local
minima (maxima) of non-divergent wind curvature was
accomplished via a second thresholding mask. Points
retaining trough or ridge classification following the masking
procedures were identified as such. An example of a troughclassified
system is shown in Figure 2.
Mean meridional winds at each longitude were calculated
for each year to establish background flow. Criteria
for southerly (northerly) phase designation followed the
reasoning that negative (positive) curvature vorticity
advection occurred east (west) of a trough, along with
meridional wind components greater (less) than the
calculated mean. Points that did not meet any of the criteria
above were designated as being not associated with an AEW
phase (no-wave). Bain et al. (2011) discussed limitations of
the composite view of AEWs, noting that there is a spectrum
of wave structures possible over West Africa. With this in
mind, a specific distance from trough or ridge axes was
not used as a classification criterion in this study. Instead,
the algorithm employed can be used to identify regions
consistent with dynamics of each wave phase. Analysis of
these maps (not shown) revealed that the identified regions
were slightly smaller than the 500 km distance associated
with triggering and maintenance of convection (Berry, 2009;
Nicholls and Mohr, 2010). Since regions are compared
against one another, the large PF sample size should smooth
out any natural variability of convective location within an
AEW phase region; this should not impact the results of this
study.