-
Herein, we describe several meteorological characteristics associated with
the climate of tropical South America, focusing on the Brazilian State of
of Rondonia in particular. We begin with a description of the regional
geomorphology (e.g., topography, hydrology, vegetation) of Rondonia, a
primary climatological "control" that should be considered in
addition to Rondonia's near-equatorial latitude. Next, the
regional characteristics of several meteorological parameters are
described including rainfall, dry and wet-bulb temperature, wind, and humidity.
Following the aforementioned overview, a summary of previous planetary boundary layer
observations collected in Rondonia during the ABRACOS/RBLE (Gash et al., 1996) are described. The next section summarizes
synoptic scale phenomena observed over tropical South America. We conclude
this section with a synopsis of previous convective classifications developed
from observations collected during the Amazonian Boundary Layer Experiment (ABLE;Garstang et al., 1990).
2. Geomorphology
varies from a height of 50-1000 m, but is largely confined to elevations of
300 m or less. Areas where the elevation exceeds 300 m, and in a few cases
500 m, are oriented linearly from (Fig. 1)
1) northwest to southeast between 10.5 S 64.5 W and 13 S 61 W; and
2) north to south, approximately parallel to 61.5 W and between 10 S and 11 S.
The terrain exhibits its greatest slope as it approaches the extreme
southeastern corner of the region, near the city of Vilhena, where elevations
exceed 700 m near the headwaters of the Ji Paran River and the Serra Dos
Parecis plateau.
Hydrologically, Rondonia can be subdivided into several basins (e.g., Fig.
1) that serve as catchments for the north-south flowing Madeira River and one
of it's tributaries, the Ji Pirana River. The Madeira River is a primary
tributary of the Amazon River, and reaches its confluence with the Amazon
approximately 700 km north of Rondonia, near the city of Manaus. The Guapore
River flows along the extreme southern edge of the region (becoming the border
between Brazil and Bolivia), and is supplied by catchments located south of
the northwest to southeast oriented terrain feature (Fig. 1) discussed in the
previous paragraph.
For quick-look LANDSAT images of Amazonia go to INPE- Amazonia from Space.
3. Meteorological Variables
Trends in the monthly mean dry bulb temperature are relatively flat during
the wet-season with means of approximately 24.6-25.0 C, monthly mean maximums
of 28-29 C and monthly mean minimums near 22C (see Ratisbona, 1976; Culf et al.,
1996). Relative peaks in monthly mean dry-bulb temperature occur in April and
October, but are only a few tenths of a degree warmer than in the wet-season.
Consistent with its location in the tropics, the annual variation of mean
monthly dry-bulb temperature in Rondonia is 2-3 C. Differences in
in the surface radiation budget are associated with a slight tendency for
mean dry bulb temperatures to be O(0.1-0.3C) higher over the canopy
(e.g., Culf et al., 1996) relative to pasture (deforested) locations. Monthly
mean and monthly mean maximum wet-bulb
temperatures exhibit similar annual trends (e.g., Figs. 2a-b
respectively; data courtesy of
ABRACOS).
Fig. 2a
Note that the mean daily variability of the wet-bulb in Fig. 2 is very small
(e.g., max variations of 2C or less) during the wet-season and much more
pronounced during the dry season (e.g., variations of 6-8 C).
Diurnally, both dry and wet-bulb temperatures exhibit peaks in the afternoon
hours (e.g., Fig. 3; wet bulb) over forest and pasture locations with the
mean diurnal variation of the dry bulb temperature being approximately 7C
(e.g., Culf et al. 1996) during the wet season.
Fig. 3a
The amplitude of the wet-bulb temperature diurnal variation is only
2C (Fig. 3), illustrating the added control of specific humidity variation
on the wet-bulb temperature. Wet bulb temperatures at the top of the forest
canopy are several tenths of a degree warmer than those of the pasture site in
Fig. 3, perhaps due to slightly higher specific humidities (see, Culf et al.
1996).
b. Wind
Winds at the surface in Rondonia are typically very light, with
speeds averaging less than 2-3 m/s. Regionally, the direction of the wind
is controlled by the combined migration of the south Atlantic anticyclone, the
Equatorial trough, and the development of a thermal continental low pressure
region to the south. During the wet-season, windspeeds are typically 1-2 m/s
and the direction is from the north-northeast in the northern portion of the domain and north-northwest in the southern portion of the domain. The wind
direction during the dry season is north-northwest in the mean, but
exhibits marked departures to a more southwesterly direction in association
with cool frontal passages ("Friagens"). In general, significantly lower
dewpoints (e.g., Fig. 2) and enhanced westerly winds occur episodically during
both the wet and dry season when the region is influenced by frontal passages
to the south (e.g., note large
negative deviations in the wet-bulb temperate in Fig. 2 during the dry season-
most of these departures occurred in association with enhanced westerly wind).
There appears to be little relationship between surface wind direction/speed
and wet-bulb temperature during the wet season (at least not during the 1995
wet season).
Secondary circulations associated with local influences such as topography
and land surface heterogeneity can also affect the winds locally. Mesoscale
simulations of the dry season wind field in Rondonia (Silva Dias et al., 1996)
suggest that topographically enhanced flows (e.g., drainage at night and
upslope flow during the day) of 1-3 meters per second
occur along a northwest to southeast transect of elevated terrain
located southwest of Ji Parana and extending southeast to northwest through
Rondonia (e.g., Fig. 1). The degree to which land surface heterogeneity
influences similar local wind circulations during the wet season is unknown.
Winds aloft (200 mb) over Rondonia are light and southeasterly during the
wet season, reversing to stronger westerly flow during the dry season
(austral winter).
c. Humidity
Surface mean Relative Humidity (RH) during the wet-season (Nov-Mar) is
approximately 80%. Mean values of RH dip to ~50% during the dry season
(May-September), with values of 70-80% typical of the transition months
(April and October) (cf.Climate Diagnostics Center).
ABRACOS/RBLE Specific Humidity (SH) data presented in Culf et al. (1996)
indicate seasonal monthly mean SH's of 16.5 g/kg and 17.2 g/kg over pasture
and forest locations respectively. The maximum SH over the pasture
occurred in the December-February time frame, while maximums over the forest
canopy occurred in April. During the wet-season, the SH's measured by Culf et
al. exhibited a peak to trough variation of only 0.4 g/kg over both forest and
pasture sites. Forest SH values were ~0.3-0.6 g/kg higher than
the pasture SH during the wet season. During the dry season, SH
values decreased
to minimums of 11.6 g/kg and 13.4 g/kg over the pasture and forest
sites respectively. SH values over the canopy were approximately 1-2 g/kg
larger than pasture SH values during the transition/dry seasons. Increased
humidity values over the forest canopy are the result of enhanced
evapotranspiration relative to deforested areas (e.g., Salati et al., 1996).
It's interesting to note that the Amazon serves as a relative "ocean"
for areas south of Rondonia. Northerly winds export moisture and latent
heat from the Amazon basin to regions of southern Brazil, much as the
North American continent experiences monsoonal intrusions of moisture
from both the Gulf of Mexico and the eastern Pacific Ocean during
northern hemisphere summer.
d. Rainfall
Rondonia's near-equatorial location places the region under the influence
of several "quasi-permanent", but seasonally oscillating circulation features.
During Austral summer, southward migration of the near equatorial trough,
coupled with a southeastward retreat of the south Atlantic anticyclone
and development of a thermal low pressure region over the continent (15S to
to 20S), results in moisture convergence and deep convection over Rondonia.
Onset of this "wet-season" in Rondonia generally occurs in mid to late October
and continues into April. Available rainguage (Fig. 4a) and OLR (Fig. 4b) data
(cf. Horel et al., 1989; Hastenrath, 1997) suggest that the heaviest rains of
the wet season occur in the months of December to February.
Fig. 4a: Monthly precipitation totals for a 2.5 x 2.5 Degree box
centered on Rondonia. Totals are courtesy of the Global Precipitation Climatology Center
and were computed using available gauge data.
The GIF image selections below were created for the box area shown in the
"instrumentation" graphic (FIG 1) [8 S - 13 S; 60 W - 65 W]. The images
were produced by Brad Newton, (Ph.D student in Prof. Tom Dunne's group)
University of California Santa Barbara. In each image, an "x" marks the
location of a raingauge. There are on the order of 40 gauges in the
region. The center of the figures (I have not edited them) is located
within the S-POL and TOGA 150 km coverage radii.
Nov. 94
Nov. 95
Nov. 96
Dec. 94
Dec. 95
Dec. 96
Jan. 94
Jan. 95
Jan. 96
Feb. 94
Feb. 95
Feb. 96
Mar. 94
Mar. 95
Mar. 96
Rainfall during the peak in the wet season typically exceeds 250 mm/month.
During the dry season, convection retreats toward the northwest
and reintensifies over Central America,
and the eastern Pacific ITCZ region; precipitation totals subsequently
decrease to less than 50 mm/month in Rondonia (Fig. 4). Annual precipitation
totals are approximately 2-2.5m, the majority of which clearly falls during
the wet-season (e.g., Fig. 4).
Diurnally, satellite studies suggest that convection and rainfall peak in the late
afternoon-early and evening hours (e.g., Figs. 5a-b) in Rondonia (Garreaud and
Wallace, 1997; Negri et al., 1998), though guage data collected during ABRACOS suggest that the
diurnal cycle in rainfall may be more variable.
Fig. 5: GOES-IR composite image for March, 1998. (a) 1200 UTC
(0800 L), and (b) 2100 UTC (1700 L).
**NOTE** Real-time and recent archived GOES images for sectors over
Brazil can be viewed Here.
4. Boundary Layer
Differences in the depth of the CBL between pasture and forested
locations may induce secondary thermally driven circulation features.
Based on mesoscale modeling studies, Silva Dias et al. (1996) suggest that
secondary circulation features induced by spatial differences in vegetation
type and coverage may provide the requisite trigger for convection if the
proper thermodynamic instability is present. During the dry season satellite
observations suggest that area coverage by cumuliform clouds is enhanced over
deforested regions (Cutrim et al., 1995). It remains to be seen how, or even
if, secondary circulation features affect the initiation and/or development
of convection during the wet-season.
Of importance to the moisture budget of the boundary layer are
fluxes of latent heat associated with evapotranspiration. Over the forest
canopy evapotranspiration can account for appoximately 50% of the budget
(i.e., of the same magnitude as the transport terms). Hence recycling of
precipitation intercepted by the canopy surface is an important source of
of latent heat in the PBL.
5. Synoptics
A review of the
literature suggests that one of the most frequently discussed synoptic features
in this region of the world is the upper level (e.g., 200 mb) Bolivian high.
This anticyclone develops coincident with increasing convection over Amazonia
during the wet season. Several studies attribute formation of the Bolivian
high to latent heat release and associated mass fluxes provided by Amazonian deep convection.
Observational and
modeling studies suggest that the Bolivian anticyclone develops to the
southwest of the Amazonian convection center in association with
a Rossby wave response to an isolated heat source (i.e., an adjustment
process; Silva Dias et al., 1983).
Horel et al. (1989) found that the onset of the wet season (Oct-Nov.),
accompanied by development of deep convection and the Bolivian high, is rapid,
occurring over a time period of less than one month. Coincident with the wet
season onset, the south Atlantic anticyclone (a prominant surface feature)
shifts to the southeast, permitting a southward intrusion of the equatorial
trough into northern Amazonia. The Amazon convection center (AMZ) (e.g.,
Hastenrath, 1997) subsequently develops as the combined response to enhanced
moisture convergence, solar heating (austral summer), and latent heat fluxes
associated with enhanced evapotranspiration. Interestingly, Horel et al.
found that the transition from wet to dry season is not as abrupt, taking
approximately two months to complete.
During the end of the wet season (Mar-Apr.), there is an increasing
probability that Rondonia can be affected by synoptic
scale boundaries originating from decayed coastal convection along the northeastern coast of South America (e.g., coastal squall lines; see Sec. 6), orginally forced
by sea breeze convergence. These boundaries can exceed length scales of
1000 km and can propagate up to 2000 km inland. Convection regenerates
along the boundaries in an oscillating fashion, driven in part by the
diurnal cycle of solar heating.
In addition to synoptic scale disturbances generated along the northeastern coast
of South America, frontal systems associated with mid-latitude cyclones moving
across southern Brazil and Argentina during austral fall, winter and spring
can also act to initiate and organize convection in the region (e.g., Molion,
1987). As the wet season progresses into the March transition, the possibility
of cool frontal intrusions into the tropics from the south becomes a very
real possibility. Indeed, a more recent study by Garreaud and Wallace (1998)
suggests that frontal intrusions can even modulate climatological convection
during austral summer. Garreaud and Wallace suggest that these
incursions of modified mid-latitude air may ultimately be responsible for
~20-30% of the total precipitation occurring in the western Amazon basin
during the months of December to February. Time-latitude diagrams
of OLR presented by Garreaud and Wallace
suggest that enhanced convection associated with northward moving
disturbances may occur in the southwestern portion of the Amazon basin
with a period of 7-15 days.
Though not as pronounced an influence as that observed in the W. Pacific,
the tropical 30-60 day oscillation (ISO) may also modulate the occurence
of convection over western Amazonia. The ISO can be detected over Amazonia
as a slight enhancement in the positive and negative OLR anomaly (Hendon and
Salby, 1994[HS94]; Salby and Hendon, 1994[SH94]), but is more readily observed
in the 200 mb zonal-wind anomolies. This is consistent with the
"radiating response" behaviour discussed in HS94/SH94. Kousky and
Kayano (1994) found evidence for an intraseasonal oscillation in their
analysis of OLR and wind data over South America. In the Kousky and
Kayano study, the ISO produced its strongest signal in OLR and 250 mb
winds over tropical northeastern South America during austral summer.
Kousky and Kayano found that passage of the ISO was associated with
positive (negative) OLR and westerly (easterly) 250 mb zonal wind anomalies
over northeastern South America. Kousky and Kayano did not examine the
upper level divergence field, however, the results of HS94 suggest the
presence of enhanced upper level divergence over tropical South America
during coincident periods of negative OLR anomaly. The HS94 study
also suggests a tendency for 1000 mb convergence to lead negative
OLR anamolies by approximately 5-10 days. The SH94 study suggests
that the 30-60 day oscillation is amplified directly by convection in
the eastern hemisphere, but that the propagating Kelvin response influences
convection over western Hemisphere locations such as South America. The
dynamical forcing associated with the Kelvin wave is strong enough to
produce a coherent signal in the OLR over South America, but the resultant
signal in OLR is much weaker than that observed over the Indian or western
Pacific Oceans. In essense, the effect of the MJO on convection over tropical
South America could be viewed as that of a weak modulator of climatological
climatological convection (apparent only if the OLR is bandpass filtered).
Recent studies utilizing atmospheric electricity data (lightning flash rate
and Schuman Reasonance) also suggest a modulation of the convection over
tropical South America on ISO timescales (e.g., Petersen et al., 1997; Anyamba
et al., 1997).
Paegle and Mo (1997) found that the SASS may be linked to the
ISO in some fashion since the SASS occurs with a similar period. The SASS is
manifested as an oscillation in convection and rainfall between the SACZ and
subtropical/tropical regions of central and eastern South America. The SASS
appears to exert only a weak influence (speculation based on the Paegle and
Mo analysis)
on convection in the vicinity of Rondonia.
Rondonia is located on the southwestern side of the Amazon River basin
(Fig. 1) and is encompassed by the area between 8 S - 13S, and 60 W - 65 W.
Vegetation in the region is dominated by tropical rainforest, deforested
pastureland, and sparsely forested savannah. A heterogeneous grid
of alternating forested and deforested grassland lies in the region between
10 S - 12 S and 61 W - 63 W (e.g., see LBA-Images at ORNL). The terrain relief (Fig. 1)
Fig. 1: Relief map of Rondonia region with a draft of the LBA-TRMM
instrumentation network indicated (LBA Home Page). Note that tethersonde operations will also likely
be conducted at the TOGA radar site. Several hydrological catchment
basins are indicated in italics. Cities and towns are indicated by bold
letters.
a. Surface Dry and Wet-Bulb Temperature
Fig. 2b
Fig. 2: (a) Daily mean wet-bulb temperatures (1995) over Fazenda Nossa
Senora Aparecida, a pasture location approximately 50 km northwest of Ji Parana.
A seven day running mean (red line) is applied to the point data (yellow).
Time in Julian-Days is plotted on the abscissa. (b) Monthly mean maximum
wet-bulb temperatures (1995) over the pasture site (NS), light red line
and also over
Reserva Jaru, a forest site RJ, bold red line.
Fig. 3b
Fig. 3: Diurnal cycle of hourly mean wet-bulb temperature for
pasture (top: same site as Fig. 2) and forest site (bottom: Reserva Jaru)
partitioned by month of the 1995 wet-season (Nov. - Mar.). Time in UTC is
plotted on the abscissa.
Fig. 4b: January, 30-year mean OLR over tropical South
America. Provided by the Climate Diagnostics Center.
Previous field experiments conducted in the Amazon such as ABLE have
observed significant variability in the diurnal cycle of rainfall. Diurnal
cycle variability observed during ABLE was apparently a function of
the dominant convective system type (e.g., Greco et al., 1990). Over Rondonia,
recent observations (personal comm., M. Silva Dias) suggest that small
MCS's are the dominant mode of convection. Forcing of these systems
can originate via diurnal heating and/or interactions between propagating
gravity waves/outflow boundaries emmanating from convection originally
located over elevated terrain (e.g., eastern slopes of the Andes)
west of Rondonia earlier in the day. During ABLE, smaller local convective
systems were the most frequently observed convective type and were
strongly modulated by the diurnal cycle of solar heating. Rainfall in these
systems was well correlated to the maximum observed wet-bulb temperature
measured prior to rainfall events (Eltahir and Pal, 1996). However, less
frequent, but larger propagating, organized convective systems produced the
majority of the total measured rainfall.
To the extent that larger propagating systems (e.g., Amazon coastal
squall lines) do not affect the timing/initiation of convection over
Rondonia, it seems reasonable to expect afternoon peaks in convective rainfall
as the general rule. 
Fig. 5a
Fig. 5b
Observations from the ABRACOS/RBLE (Nobre et al., 1996) have provided
important insights to our understanding of the structure of nocturnal and convective
boundary layers (NBL and CBL respectively) over both forested and
deforested regions. Land surface heterogeneity (e.g., forest vs. pasture)
exerts a very strong control over the development and subsequent depth of the
NBL and the CBL.
Development of the NBL over both forested and pasture locations begins at
approximately 2000 L, and the NBL reaches a maximum depth near 0500 L.
However, the peak NBL depth (385 m) over the forest canopy is 140 m deeper
than that of the pasture NBL (240 m). Nobre et al. (1996) attribute
this difference in depth to a relative increase in mechanical turbulence
generation over the forest canopy at night. As a result of the increase in
turbulence and the increase in depth, the forest NBL exhibits a much weaker
thermal inversion/lapse rate (22 K/km) then the pasture NBL (48 K/km).
Erosion of the NBL over both forest and pasture sites begins at
approximately 0800 L, signaling the development of the CBL. Growth of the
CBL over the pasture locations is the most rapid, with the mixed layer depth
increasing to nearly 1600m at a rate of ~500 m/hr between 0800 L and
1100 L. In the same time interval the forest mixed layer depth increases
to approximately 600m. Between 1100 L and 1700 L, the CBL over the pasture
grows to a depth of 2200 m, while that of the forest extends to a depth of
only 1200 m. The virtual potential temperature gradient across the top of
the pasture CBL is much stronger (8 K/km) than that of the forest CBL (3 K/km).
Conventional boundary layer modeling algorithms apparently cannot
accurately simulate the growth of the CBL over the pasture, and only
do a mediocre job of simulating CBL growth and depth over the forest,
indicating weaknesses in our understanding of PBL physics for this region.
Prominant synoptic, and to some extent climatological, features occurring and/or affecting tropical South
America include the Equatorial trough, Bolivian high, S. Atlantic anticyclone,
and migrating frontal systems which pass through southern Brazil but sometimes
extend into the deep tropics (e.g., the Rondonia region). To some extent, the S.
Atantic Convergence Zone (SACZ) and associated South Atlantic See-Saw (SASS)
(Paegle and Mo, 1997) pattern may also exert some effect on the region.
