This paper examines the climatological, large-scale, and synoptic-scale asp
ects of South American cold surges using NCEP-NCAR gridded reanalyses for t
he 1992-96 period. Three common cold surge types are identified on the basi
s of a thickness (1000-850 hPa) criteria: type 1-a transient surge associat
ed with weak anticyclone development east of the Andes in the absence of ri
dging aloft, type 2-a strong and persistent surge associated with dynamic a
nticyclogenesis aloft and strong surface anticyclone development east of th
e Andes, and type 3-a surge east of the Brazilian coastal mountains. Cold s
urges are most common during the winter and spring (Jun-Nov), accounting fo
r 189 of the 256 events (74%).
Case studies of two events (19-22 Jul 1992 and 12-14 Apr 1993) are conducte
d from both a conventional isobaric and a potential vorticity (PV) perspect
ive. The upper-air flow pattern in the July 1992 type 2 case is characteriz
ed by the presence of a strong ridge-trough couplet, which amplifies and be
comes quasi-stationary, allowing for a deep layer of equatorward flow over
South America. Dynamically, this flow pattern favors the development of a v
ery strong surface anticyclone to the east of the Andes in response to a co
mbination of differential anticyclonic vorticity advection, low-level cold
advection, and, equivalently, positive PV advection. Because of the associa
ted cold air damming east of the Andes, modified cool air is transported in
to the western part of Amazonia. Cold air damming east of the Brazilian coa
stal mountains is associated with the transition of the July 1992 type 2 su
rge into a type 3 surge.
The cold surge of April 1993 is examined as a rare event that does not fit
the above classification. It is characterized by explosive cyclogenesis clo
se to the coast of Argentina. Unlike the representative type 2 cold surge o
f July 1992, which tends to occur in association with southwesterly flow al
oft, the April 1993 cold surge occurs beneath westerly and northwesterly fl
ow aloft. Cold air penetration into lower latitudes is restricted because t
he geostrophic wind has a component directed away from the Andes equatorwar
d of the cyclone. The dynamical forcing mechanisms associated with the Apri
l 1993 event are of smaller scale than those of the much more common surges
typified by the July 1992 event.