VOLUME 149 MONTHLY WEATHER REVIEW NOVEMBER 2021

Convective Cloud Clusters and Squall Lines along the Coastal Amazon

A. C. SOUSA,a L. A. CANDIDO,b AND P. SATYAMURTYb

a
PG Program in Climate and Environment (CLIAMB), National Institute of Amazon Research (INPA),
Manaus/Amazonas, Brazil
b
National Institute of Amazon Research (INPA), Manaus/Amazonas, Brazil

(Manuscript received 8 March 2021, in final form 12 August 2021)

ABSTRACT: Mesoscale convective cloud clusters develop and organize in the form of squall lines along the coastal Amazon
in the afternoon hours and propagate inland during the evening hours. The frequency, location, organization into lines, and
movement of the convective systems are determined by analyzing the ‘‘precipitation features’’ obtained from the TRMM
satellite for the period 1998–2014. The convective clusters and their alignments into Amazon coastal squall lines are more
frequent from December to July, and they mostly stay within 170 km of the coastline. Their development and movement in the
afternoon and evening hours of about 14 m s21 are helped by the sea breeze. Negative phase of Atlantic dipole and La Niña
combined increase the frequency of convective clusters over the coastal Amazon. Composite environmental conditions of 13
large Amazon coastal squall-line cases in April show that conditional instability increases from 0900 to 1200 LT and the wind
profiles show a jet-like structure at low levels of the atmosphere. The differences in the vertical profiles of temperature and
humidity between the large-squall-line composites and no-squall-line composites are weak. However, appreciable increase in
the mean value of CAPE from 0900 to 1500 LT is found in the large-squall-line composite. The mean mixing ratio of the mixed
layer at 0900 LT in La Niña situations is significantly larger in the large-squall-line composite. Thus, CAPE and mixed-layer
mixing ratio are considered to be promising indicators of the convective activity over the coastal belt of the Amazon basin.
KEYWORDS: South America; Atmosphere-land interaction; Convection lines; Mesoscale processes; Sea breezes; Squall lines;
Interannual variability

1. Introduction years under the combined influence of El Niño (EN) and nega-
tive phase of AD.
The north coast of South America is affected by mesoscale
Betts et al. (1976) observed tropical squall lines over
convective cloud clusters (CCCs) and their organization into
Venezuela and commented that the formation of an SL is a
squall lines (SL) parallel to the coast (Garstang et al. 1994;
highly organized convective process. Houze (1977) studied
Cohen et al. 1995, 2009). These systems contribute nearly 40%
tropical squall lines and showed the structure of an SL con-
of the annual precipitation of northern coastal Brazil and ad-
sisting of a convective region, formation of new convective
joining regions. Heavy afternoon showers caused by these
elements in the direction of its displacement and a stratiform
systems disrupt the society in the coastal cities of Brazil
region where the air descends. Maddox (1980) presented a
(Loureiro et al. 2014). Flash floods not only damage infra-
description of mesoscale convective complexes in general and
structure like roads and bridges and the industrial production
explained their structure and development. His work was
but also cause forest tree mortality (Negrón-Juarez et al. 2010).
mostly about the extratropical and subtropical convective
The formation and intensity of the convective systems de-
systems. Kousky (1980), Greco et al. (1990), Garstang et al.
pend on the synoptic situation, i.e., the prevailing winds and
(1994) and Cohen et al. (1995) studied SLs in the Amazon
thermal and moisture characteristics of the atmosphere over
basin. Greco et al. (1990) estimated that 40% of the precipi-
the continent and over the adjacent seas. However, the pre-
tation in the Amazon basin is caused by squall lines during the
diction of the convective systems depends heavily on the
months of April and May 1987. Cohen et al. (1995) found that
knowledge of their frequency, diurnal and seasonal variabil-
45% of the precipitation in the state of Pará in Brazil is caused
ities, the effects of El Niño–Southern Oscillation (ENSO) and
by squall lines during the rainy season December through May
the tropical Atlantic dipole (AD) on their frequency, their
(Rao et al. 2015). Cohen et al. (2009) gave a summary of the
climatological movement and an understanding of the mech-
characteristics of the squall lines over north and northeast
anisms of their formation. Numerical weather prediction
coasts of Brazil. They found that the months of May and
models are helpful. However, global models as well as regional
January present the highest and lowest frequencies, respec-
models by themselves are not fully efficient in mesoscale pre-
tively, of the squall lines. In La Niña (LN) years they are more
diction, especially in the equatorial latitudes. The effects of
frequent and tend to form southward of their mean position in
ENSO and AD on the precipitation over the Amazon basin were
EN years. The squall lines forming in JFM are more intense
analyzed by Souza et al. (2000) and found negative anomalies of
and those forming in OND are of medium intensity. The fre-
precipitation in the austral summer and autumn seasons of the
quency and propagation of Amazon squall lines and the asso-
ciated vertical profiles are studied by Alc^ antara et al. (2011)
Corresponding author: Aline Corr^ea de Sousa, alinecorrea.acs@ and found that they are more frequent in April, May, and June
gmail.com season and 54% of them propagated less than 170 km inland

DOI: 10.1175/MWR-D-21-0045.1
Ó 2021 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright
Policy (www.ametsoc.org/PUBSReuseLicenses).
3589
3590 MONTHLY WEATHER REVIEW VOLUME 149

from the coast. They also found an associated low-level jet systems produce copious amounts of precipitation, moisture
structure in the wind profile. parameters like precipitable water and moisture conver-
In a study of potentially severe convection (PSC) over six gence become important. As the region of study is in the
subregions of the Amazon basin, Nunes et al. (2016) found equatorial latitudes, the intensity and profile of the easterly
maximum frequency over the Guyana Shield in the 3-month trade winds are important. And, as the systems develop near
period from July to September and over the northeastern the coast, the land surface characteristics and the sea breeze
Amazon in the period October–December. A PSC was defined become important (Kousky 1980; Cavalcanti 1982). During
by the authors as an area of 2000 km2 with polarization-corrected late morning and afternoon the land surface warms up more
temperature (PCT) , 250 K and the volume of precipitation in than the sea surface producing thermal contrast. Warmer air
excess of 1000 mm h21 km2 and the area must contain at least over the land surface rises creating a relative low pressure
one pixel with PCT , 100 K. Diurnal variability of PSC over area and the sinking of cooler air over the water surface
these two regions showed maximum frequency during the produces a relative high pressure. The air is forced by the
afternoon and evening hours, between 1200 and 1700 LT. pressure gradient near the surface to move onshore, and
Oliveira and Oyama (2015, 2020) have analyzed the atmo- compensatory offshore movement occurs around 1.5 km
spheric conditions preceding squall-line initiation over the above the surface, completing a thermally direct circulation
north coast of Brazil. Their results show the importance cell. The surface onshore movement is sea breeze (Parker
of lower-tropospheric humidity and the formation of cold 2003). For a difference of 28C between land and sea sur-
air pools. The study of Oliveira and Oyama (2020) over the faces and considering the cell width to be 100 km and no
Brazilian north coast (more exactly near the Alcantara frictional effects, the acceleration of sea breeze is about
Rocket Launching Center) showed that 47% of the March 2.0 3 10 23 m s 22 (Holton 2004). That is, the sea breeze can
precipitation is caused by squall-line activity. assume a velocity of about 7 m s21 in an hour, in principle. In
Most of the observational studies about the north coast reality the breeze remains around 3–4 m s21. The breeze rein-
squall lines in Brazil were based on visual identification of forces the easterly trades over the coastal belt.
convective systems in satellite cloud imagery. An SL is an The present study is undertaken to obtain a deeper un-
elongated convective region in the sense that its length is much derstanding of the frequency, spatial and temporal variability
larger than its width. Cohen et al. (2009) and Machado et al. of the CCCs and their organization into Amazon Coastal
(2014) affirmed that, normally, a squall line forms and develops Squall Lines (ACSLs) near the north coast of Brazil to help
in the afternoon and early night hours. meteorologists to improve weather forecasts in the region.
Garstang et al. (1994) proposed a conceptual model Although some qualitative and quantitative descriptions of
structure of the coastal SL, essentially showing warm air the systems can be found in earlier publications (Kousky
rising and gliding over the relatively cooler air in the wake, 1980; Cavalcanti 1982; Garstang et al. 1994; Cohen et al. 1995;
with convergence in the lower troposphere and divergence Nunes et al. 2016), the spatiotemporal variability was not
in the upper troposphere. They also showed their life cycle presented. The past studies have not explored the potential
passing through several stages: coastal genesis, intensifica- indicators of the development of these systems and their
tion, maturation, weakening, reintensification and final usefulness for operational meteorologists.
dissipation of convection. During the late afternoon hours Thus, we revisit the observational aspects of the distri-
the convective clusters intensify and the cumulonimbus bution of CCCs within the length and width of the ACSLs
clouds attain heights of up to 17 km. Some of the SLs move near the north coast of Brazil with the help of TRMM data.
westward for several hours, while the individual cells em- It is also necessary to categorize the CCCs and the ACSLs
bedded in them (CCCs) go through a regenerative process. according to the strength or intensity of convection and size.
In the last stages of their life cycle the embedded mesoscale Above all, it is important to compare the physical charac-
cells in the SL spread laterally, weaken and lose their me- teristics and thermodynamic characteristics of the back-
soscale structure. At this stage bright patches of convection ground atmospheres, one in which ACSL develops and the
as seen in IR satellite imagery become blurred and practi- other in which ACSL does not develop. This comparison may
cally the SL does not show displacement. offer clues for the operational meteorologists of the region in
The environmental conditions associated with the tran- forecasting work.
sition from shallow convection to deep convection in the The objectives are 1) to conduct an observational analysis
Amazon basin is studied by Zhuang et al. (2017) using of the interannual, seasonal, and diurnal variabilities and
GOAmazon data for a 12-month period in 2014/15 and spatial variability of CCCs and ACSLs along the north coast
found that deep convection cases are associated with lower of Brazil and to examine if the interannual variability bears
surface temperature, higher specific humidity, shallower any relation to the phases of ENSO and AD and 2) to find
mixed layer, smaller sensible heat flux and weaker surface what environmental factors provide (or do not provide) clues
wind. The transition is observed around noon. There can for the detection of the development of ACSLs for the benefit
be differences between the interior Amazon and coastal of the meteorologists of the region.
Amazon cases. The article is organized as follows. Section 2 gives a de-
One can think of a few factors that play important roles scription of the datasets and the method of analysis. Section 3
in the development of the CCCs over the north coast of gives the spatial, interannual, seasonal, and diurnal variabil-
the Amazon basin and their organization into SL. As the ities. Section 4 is about the atmospheric conditions associated
NOVEMBER 2021 SOUSA ET AL. 3591

FIG. 1. (a) Location map. Rectangular area with dashed border: CCCs are identified.
Oblique area with continuous black border: ACSLs are identified. Dotted rectangle: Oceanic
area. Brazilian states are identified with two letters: Amapá (AP), Pará (PA), Amazonas
(AM), Mato Grosso (MT), Maranhão (MA), Ceará (CE), Piauí (PI), and Bahia (BA).
Topography is shown in gray shades. (b) Mean monthly variability of temperature and
precipitation over the black dashed rectangular area from ERA5 reanalysis data (R 5 rainy
season; D 5 dry season).

with the development and movement of the systems. Section 5 temperature and monthly precipitation characteristics, ob-
gives a summary and discussion of results. tained from ERA5 datasets, of the region are shown in
Fig. 1b. It is interesting to note that the temperature is about
28C higher in the dry season (August–November), than in the
2. Data and method
rainy season (December–May). In the rainy season the mean
a. Study area and datasets monthly precipitation exceeds 200 mm. The rectangular area
bounded by 28 and 128N and 558 and 358W is considered for
1) STUDY AREA
obtaining the atmospheric conditions over the seas adjoining
The area considered here is part of the eastern Amazon the coastal region of the study. The present study area ap-
basin situated between 68S and 48N latitudes and 568 and 428W proximately corresponds to the combined northern halves
longitudes in equatorial South America and is shown in Fig. 1a. of the Guyana Shield and Eastern Amazonia subregions
The dominant physical characteristic of the region is the lower considered in the study of Nunes et al. (2016).
Amazon River surrounded by tropical forest. The topography is
2) SATELLITE IMAGERY
almost uniform, except for the uplands of over 500 m above sea
level in the northwestern Pará and southern parts of the Guyanas Examples of ACSL over the north coastal region of Brazil
known as Guyana Shield. The coastline is inclined from northwest are given in Fig. 2. In this, the enhanced GOES 13 cloud images
to southeast and therefore the easterly or northeasterly equatorial show the cloud top temperatures. The cooler the temperature
winds cross the coast into the continent. The mean monthly the higher is the cloud top and hence the deeper the convection
3592 MONTHLY WEATHER REVIEW VOLUME 149

FIG. 2. Brightness temperature (color shades) observations on (top) 3 Apr 2014, (middle) 7 Apr 2013, and (bottom) 8 Apr 2014,
representing Small ACSL, Medium ACSL, and Large ACSL, respectively, for (left) 1815 UTC (1515 LT), (center) 2115 UTC (1815 LC),
and (right) 2315 UTC (2015 LT). Squall lines are delineated by the green-outlined polygon. Cases are examples of enhanced IR images.

(Satyamurty and Rosa 2020). The lines of coastal convection For B . 176, TB 5 418 2 B; for B # 176, TB 5 300 2 (B/2);
are circumscribed by green rectangles. The identification of (1)
ACSLs and their categorization procedures are given below in
section 2b. TB is given in kelvins.
The InfraRed (IR) geostationary satellite images produced
3) PRECIPITATION FEATURES AND IDENTIFICATION
by AWC/NOAA (Aviation Weather Center/National Oceanic
OF CCCS
and Atmospheric Administration) and supplied by GHRC
(Global Hydrology Resource Center) of NASA (National A precipitation feature (PF) is an area, of size ;75 km2 or
Aeronautics and Space Administration) are examined for larger, comprising at least four radar pixels over which rain is
semiobjective classification of the ACSLs into Small, Medium, identified. Its characteristics such as the area of precipitation,
and Large ACSLs. The global mosaics are available at intervals volume of precipitation, reflectance, and minimum brightness
of 30 min. with spatial resolution of 14 km for the period of our temperature are determined objectively (Liu 2007; Liu et al.
study 1998–2014. In these images high cloud tops are cold and 2008). The University of Utah generated and distributed
appear bright and the surface appears dark. The digital values the PF data obtained from the Tropical Rainfall Measuring
B of the pixels in grayscale are converted into ‘‘brightness Mission (TRMM) satellite data. More details about PF data
temperatures’’ TB of the corresponding pixels using Eq. (1), are available at http://trmm.chpc.utah.edu.
below. The instruments for measuring brightness B are somewhat In the present study a CCC is defined as a contiguous
different in different satellites, but they are basically Advanced group of PFs forming a cell 100 km long or composing an area
Very High Resolution Radiometers (AVHRR): greater than or equal to 2000 km2. The CCCs over the north
NOVEMBER 2021 SOUSA ET AL. 3593

and northeast coastal region of the Amazon basin are iden-

tified objectively from orbital samplings on daily basis for
the period 1998–2014 in this study. The objective method is
based on Mohr and Zipser (1996) who utilized microwave
brightness temperature. The present method uses PR radar
for defining PFs.
4) PCT AND MIN85PCT
The spatial distribution of TB [Eq. (1)] is used for the iden-
tification and classification of CCCs and ACSLs. Interpretation
of TB leaves ambiguities resulting from lower emissivity of
water masses. Spencer et al. (1989) developed an algorithm for
the determination of polarization corrected temperatures (PCT)
FIG. 3. Relation between mean values of volume of rain and
that largely removes the ambiguities. This parameter is given by Min85PCT of TRMM for the whole period of study (1998–2014).
bTBh 2 TBy See the criteria for Very intense, Intense, and Weak convection
PCT 5 , (2) in Eq. (3).
b21
where TBh and TBy are the horizontally and vertically polar- minimum value of PCT found over a PF or contiguous PFs,
ized TB that are partially affected by combinations of cloud designated ‘‘Min85PCT,’’ is less than 250 K. The PFs with ice
and precipitation. The ‘‘b’’ parameter is adjusted to produce scattering include features that do not meet all the criteria for
background PCT values in the range of 275 and 290 K. PCT mesoscale convective systems, but at least one data bin with
values less than 250 K represents precipitation (Nesbitt et al. PCT # 250 K is included. This assures the existence of signif-
2000). PCT can separate ocean and river surfaces from cold icant ice scattering, which means significant optical depth of
cloud tops in the 85-GHz frequency because of differences in the ice particles [more likely graupel than snow, according to
the horizontal and vertical emissivities of wet surfaces. Thus, Vivekanandan et al. (1990) and Smith et al. (1992)] are included
the CCCs are better identified from PCT, the corrected in the feature. The size of the PFs with these characteristics
brightness temperature. The parameter Min85PCT, the mini- varies from 75 km2 up to the size of mesoscale convective clus-
mum 85-GHz PCT, is the minimum temperature found over a ters. A characteristic size of PFs is around 200 km2.
PF or a CCC. A CCC is considered present only when the CCCs are categorized according to the following criteria:

Min85PCT , 125 K / very intense (CCCV),

125 # Min85PCT , 175 K / intense (CCCI), and
175 # Min85PCT , 250 K / Weak convective element (CCCW) . (3)

5) OBTAINING CCC figure is the mean of all observations of precipitation for the
value of Min85PCT in the interval 65 K. It is interesting to see a
The criteria for obtaining CCCs (a contiguous area of
significant correlation between the two parameters, especially in
2000 km2 with Min85PCT # 250 K) include scattering by ice
the minimum PCT (brightness temperature) range between 100
crystals and with at least one data bin having Min85PCT #
and 250 K. The correlation breaks down for PCT values around
225 K. McGaughey et al. (1996) indicated that Min85PCT
100 K because such events are very rare. The cooler the tem-
with 225 K guarantees the presence of convective elements, and
perature the higher is the volume of rain. Therefore, when the
225 K is selected here as the limit for detecting ice scattering
convective systems are categorized according to Min85PCT, in
inside the convective core. Mohr and Zipser (1996) used the
effect, we obtain a categorization of the systems in terms of their
criterion of Min85PCT # 175 K for intense convection. In the
precipitation potential. The three categories of Min85PCT [de-
same manner, Min85PCT # 125 or 100 K can be interpreted as
fined in Eq. (3)] are separated by the vertical lines in this figure.
extremely intense convection. The relationship between lower
passive-microwave PCTs and precipitation is due to the scat-
6) RADIOSONDE DATA AND ERA5 REANALYSIS DATA
tering of upwelling microwave radiation by ice particles within
the cloud. Therefore, more ice in the cloud means more scat- Radiosonde data from the meteorological station Belém
tering, which in turn means lower PCT. (1.388S, 48.488W), Brazil, observed two times per day, 0000 UTC
Figure 3 shows the scatter diagram of the ‘‘volume of rain’’ (2100 LT) and 1200 UTC (0900 LT), for the period of study are
and the Min85PCT. Volume of rain is the total rainwater pro- retrieved from the site of the Department of Atmospheric
duced in a PF or a CCC given in units of the rate of precipitation Science of the University of Wyoming. The 0900 LT data are
(mm h21) multiplied by the area (km2) over which precipitation used to characterize stability, moisture, and wind profiles be-
occurred. The graph in the figure represent the whole period of fore the onset of coastal convection and the 2100 LT data to
study. The volume of rain for a given Min85PCT shown in the characterize the postconvective atmosphere.
3594 MONTHLY WEATHER REVIEW VOLUME 149

FIG. 4. Schematic diagram illustrating the semi-objective method for identification of ACSLs;
d1 and d2 are the initial and final positions.

Because of scarcity of observational atmospheric data over Other morphological characteristics such as the life dura-
the study area, reanalysis data ERA5 (Copernicus Climate tion, dissipation time, and mean velocity of propagation are
Change Service 2017) of the European Centre for Medium- also obtained. The dissipation time is when TB . 240 K and
Range Weather Forecasts (ECMWF) with a resolution of 0.258 the cloudiness dispersed. Duration of an ACSL is the time
latitude 3 longitude for the period of study are employed for interval between the formation and the dissipation. The mean
the analysis of atmospheric conditions. velocity is calculated from the positions of the central point at
the times of formation and dissipation by visual examination.
b. Semi-objective identification of Amazon coastal
A schematic figure to explain the identification of ACSL is
squall line
given in Fig. 4. The dark spots are the CCCs embedded in the
Enhanced cloud top temperature images, as exemplified in ACSL bounded by the oblique quadrilateral. The arrows show
Fig. 2, are used to identify ACSLs. The TB isotherms are drawn the displacement of the ACSL from point d1 to point d2.
at intervals of 10 K between the values of 273 and 178 K. The Days with no ACSL development in the study area are also
enhanced images are examined visually, and ACSLs are identified in order to obtain the synoptic and thermodynamic
identified according to the following criteria: conditions on such days. These days are labeled No ACSL. The
conditions associated with ‘‘Large ACSL’’ are contrasted with
(i) elongated cloud patches (identified between the values
conditions of ‘‘No ACSL’’ situations to identify the parameters
mentioned above) of linear dimension greater than
and their values that might be associated with the formation
500 km with compact cells near or along the coast are
of ACSL.
candidates for ACSL and
(ii) elongated cloud patches with sparsely distributed or shal- c. Physical parameters associated with squall lines
low cells (TB . 273 K) or those very far from the coastal
The following five types of variables are considered.
region are not counted as ACSL.
(i) Stability indices and thermodynamic parameters shown
When an ACSL candidate is identified, its characteristics (day,
in Table 1 are obtained over the coastal land area by
month, year, time, position, length, and width) are estimated and
using the standard formulas (Galway 1956; George 1960;
registered. Based on its length and width, it is categorized as
Prosser and Foster 1966; Miller 1967; Sadowski and Rieck
(i) Small ACSL if its length is less than 800 km and/or its 1977; Emanuel 1994). For this, the radiosonde data at
width is less than 80 km, Belém (1.388S, 48.488W) and the ERA5 data at the grid
(ii) Medium ACSL if its length is less than 1400 km and its point nearest to the station are used. The K and totals
width is less than 170 km, and indices are based on the lapse rate and humidity in the
(iii) Large ACSL if its length exceeds 1400 km and its width lower troposphere. The K index expresses the potential
exceeds 170 km at its maximum size during its life. for the occurrence of storms with heavy precipitation
(George 1960) and is given by
This classification is based on the idea that a larger ACSL
produces a higher volume of rain. The dimensions of the K5[T(850) 2 T(500)] 1 fTd (850) 2 [T(700) 2 Td (700)]g ,
ACSLs are estimated from the clusters with cooler cloud tops,
(4)
TB , 210 K, that is, the size of the 210-K closed isotherm.
Examples of ACSLs of the three categories are shown in Fig. 2 in which T is temperature and Td is the dewpoint
in which the top, middle and bottom panels show Small ACSL, temperature, both in 8C. The numbers 500, 700, and
Medium ACSL, and Large ACSL, respectively. 850 indicate the isobaric levels in hectopascals; K is
NOVEMBER 2021 SOUSA ET AL. 3595

TABLE 1. Mean thermodynamic indices at 1200 UTC (0900 LT) and 1800 UTC (1500 LT) for Large-ACSL cases in April. The indices
are calculated for the grid point nearest to the city of Belém (1.388S, 48.488W). Dif: difference from No-ACSL mean; M: mean for
17 Apr months (510 days); SD: Standard deviation of the 510 days of data; (Tc 2 To)1000: thermal contrast between land and sea at
1000 hPa from ERA5 data; (Tc 2 To)sfc: thermal contrast at the surface; K: K index; Totals: totals index; CAPE: convective available
potential energy; CINE: Convective inhibition energy; um: Mean mixed layer potential temperature; rm: mean mixed layer mixing
ratio; w: precipitable water.

El Niño La Niña Climatology

1200 UTC 1800 UTC 1200 UTC 1800 UTC 1200 UTC 1800 UTC
(0900 LT) (1500 LT) (0900 LT) (1500 LT) (0900 LT) (1500 LT)
Thermo-dynamics index Large Large Large Large
and variables ACSL Dif ACSL Dif ACSL Dif ACSL Dif M SD M SD
K (8C) 34.85 20.51 35.06 0.17 36.39 1.69 35.36 1.43 34.96 2.82 34.37 1.90
Totals (8C) 42.52 0.04 42.13 0.15 43.27 0.51 42.55 0.00 42.87 1.32 42.39 1.15
CAPE (J kg21) 576.36 178.68 830.02 130.35 348.93 230.14 954.21 296.53 408.17 306.98 647.08 383.33
CIN (J kg21) 59.62 216.94 52.92 17.61 54.55 10.85 64.85 27.65 61.79 61.63 60.26 57.57
um (K) 299.27 0.08 300.97 0.57 298.10 0.04 299.95 0.18 298.54 0.75 300.09 1.25
rm (g kg21) 20.32 0.16 18.04 20.08 19.07 3.48 17.42 0.26 19.35 0.80 17.26 0.71
w (mm) 59.28 21.04 61.35 20.20 57.90 0.56 57.94 0.77 56.01 3.82 57.53 3.08
(Tc 2 To)sfc 21.20 0.07 1.44 0.32 21.72 0.27 0.98 0.20 21.10 0.76 1.66 0.97
(Tc 2 To)1000 20.47 20.01 1.90 0.26 20.80 0.21 1.32 0.14 20.27 0.57 2.04 0.88

generally considered to indicate scattered thunderstorms Tc 2 To is considered to represent the thermal contrast
when it is within the range of 318–358C and numerous and hence the sea breeze.
thunderstorms when it is greater than 358C. The totals
The variables in i–iv are calculated using the twice-daily
index is given by
radiosonde observations at 0000 UTC (2100 LT) and
TT 5 [T(850) 1 Td (850)]2 2T(500), (5) 1200 UTC (0900 LT). The variables are also calculated
using the ERA5 data for 1500, 1800, 2100, and 0000 (next
and has an interpretation similar to K. When TT $ 428C, day) UTC. The parameter in v is calculated for 1200, 1500,
vigorous thunderstorm activity is expected. and 1800 UTC. The idea is that some of them may serve as
(ii) Vertical profiles of equivalent potential temperature ue, indicators for the development of the convective systems
zonal (u) and meridional (y) wind components, and mixing studied here.
ratio of water vapor r are obtained by using standard d. Usefulness of meteorological indices
formulas (Bolton 1980; Emanuel 1994). Precipitable water
w is calculated by integrating the specific humidity from the It is necessary to evaluate the usefulness (or dependability)
surface to 300 hPa. of a dynamic or a thermodynamic index to distinguish extreme
(iii) Vertically integrated convergence of water vapor flux Cw meteorological situations. For an index I, such as CAPE, to
is obtained from be a useful indicator it is necessary that distinctly different
numerical values of I are observed in distinct meteorological
ð 300
dp situations. Two situations, one in which a Large ACSL de-
Cw 5 2=  (qV) , (6) velops and the other in which no convective activity is ob-
Ps g
served, that is, No ACSL, are distinct situations. If, on average,
where P s is surface pressure, q is specific humidity the value of the index I in one situation is not very different
(very approximately equal to mixing ratio r), V is from its value in the other situation, the forecasters will not find
vector wind, p is pressure, and g is acceleration due the index helpful. Moreover, a given index can be a useful
to gravity. guide in one season and not in another season. Also, the use-
(iv) Convective available potential energy (CAPE) and con- fulness of an index can vary from region to region.
vective inhibitive energy (CINE) are also calculated using The usefulness or dependability of a given Index is judged in
standard formulas given in Emanuel (1994). this work in the following simple manner. Let us consider, for
(v) Ocean–continent thermal contrast (Tc 2 To): Sea breeze instance, April month situations. The index I is calculated for
develops from thermal contrast between the atmospheric all April days in the 17-yr period (510 days) of the study. Then,
columns over the continent and over the adjoining ocean its climatological mean Im and the standard deviation s(I) are
in the lower troposphere. Here, the 1000-hPa temper- calculated. Next, the mean value of I for the Large-ACSL days
ature is considered to represent the lower troposphere. during April months is calculated: I(ACSL). Similarly, the
The spatially averaged 1000-hPa temperatures over the mean I value for No-ACSL days in April, I(No ACSL), is also
rectangular areas shown in Fig. 1, Tc over land, and To over obtained. The criteria for judging the usefulness of I are
ocean, are obtained from the ERA5 data. The difference given as
3596 MONTHLY WEATHER REVIEW VOLUME 149

TABLE 2. Changes from 1200 UTC (0900 LT) to 1800 UTC (1500 LT) in the mean values of the indices listed in Table 1 for the Large-
ACSL cases for April. Symbols and units are as in Table 1. Remarks are about the daytime changes relative to climatology.

El Niño La Niña Climatology Remark

K (8C) 0.21 21.03 20.59 Decrease in LN situation
Totals (8C) 20.39 0.72 0.48 Increase in LN situation
CAPE (J kg21) 253.66 605.28 238.91 Large change in LN situation
CIN (J kg21) 26.70 10.30 1.53
CAPE 2 CIN (J kg21) 260.36 594.98 237.38 Large change in LN situation
um (K) 1.70 1.85 1.55 All three values are comparable
rm (g kg21) 22.28 21.65 22.09 All three values are comparable
w (mm) 2.07 0.04 1.52 No change in LN situation
(Tc 2 To)sfc 2.64 2.70 2.76 All three values are comparable
(Tc 2 To)1000 2.37 2.12 2.31 All three values are comparable

jI(ACSL) – I(No ACSL)j , s(I) / I is not a useful indicator,

2s(I) . jI(ACSL) – I(No ACSL)j $ s(I) / I is a useful indicator, and
jI(ACSL) – I(No ACSL)j $ 2s(I) / I is a very useful indicator . (7)

These criteria can be applied to any of the indices shown in precipitation is related to the reduction of CCCs over the
Table 1 to judge if it is a useful indicator of the occurrence of north and northeast coastal belt of Brazil in the strong EN
ACSL in the region of study. year (Fig. 5a). In the strong La Niña (LN) years, 1999, 2000,
However, when the difference in the values of I before and 2008 and 2011, the numbers of CCCs were high and even
during or after intense convective activity (ACSL) is large attained maxima in CCC frequency in three of the four LN
relative to the difference in its climatological values at the years (Fig. 5a). This graph indicates a strong influence of
corresponding hours, the index is also considered useful. As we ENSO on the occurrence of CCCs in the study region. We
will see in section 3c below, maximum convective activity is also observe that in 1999, 2000 and 2008 the sign of AD was
observed at 1800 UTC (1500 LT), and therefore the changes in negative (Fig. 5a). That is, negative phase of AD and LN
the values of the indices from 1200 UTC (0900 LT) to combined increase the frequency of CCCs over coastal
1800 UTC (1500 LT) are considered for this purpose and are Amazon (Fig. 5a). The year 2010 was anomalous during
shown in Table 2. which the number of CCCs increased despite EN and pos-
itive AD conditions. However, we cannot attribute a direct
3. Interannual, seasonal, diurnal, and spatial variability relation between AD phase and the frequency of CCCs,
of CCCs and ACSLs contrary to Nunes et al. (2016) who attributed higher frequency
of potential severe convection (PSC) cases in 2005/06 to a warmer
a. Interannual variability
North Atlantic.
The CCCs are identified and categorized from the PF The frequency of ACSLs (Fig. 5b) was also low in the strong
data as described in section 2a(3). The ACSLs are identified EN year 1998. However, in the LN years 1999 and 2000 also the
and categorized as described in section 2b. In the 17 years of frequencies of ACSLs were low, and in the EN years 2007 and
our study 7520 CCCs and 2115 ACSLs over land are found 2010 the frequencies were high. Influence of weak or moderate
in the dashed rectangular region shown in black in Fig. 1a. EN on the ACSL frequency is not clear. The correlation be-
The interannual distributions of CCCs and ACSLs are tween the graphs in Figs. 5a and 5b is very low. That is, higher
shown in Figs. 5a and 5b, respectively. The ENSO phases number of CCCs does not mean higher number of ACSLs.
(colored stripes) and Atlantic Dipole (AD) signs are also Figure 5c shows the scatterplot of yearly occurrences of ACSLs
marked in the figure to check their influence on the con- and CCCs. This plot also shows that more CCCs does not mean
vective systems. more ACSLs.
Approximately 440 CCCs and 124 ACSLs per year on
b. Seasonal variability
average have occurred over the region. That is, on average,
four CCCs compose an ACSL. The year with least number Year-by-year monthly frequencies of the CCCs and ACSLs
of occurrences of CCCs was 1998, which was a strong El during the 17-yr period are shown in Figs. 6a and 6b, respec-
Niño (EN) year (Fig. 5a). It is known that during the EN tively. For example, in the month of March 2005 the number of
years the precipitation over the northeastern Amazonia CCCs (Fig. 6a) was over 80 while the number of ACSLs (Fig. 6b)
presents negative anomaly (Ropelewski and Halpert 1987, was about nine. That is, in that month there were about nine
1989; Grimm 2003; Andreoli et al. 2016; Viegas et al. 2019) CCCs per one ACSL, on average. The figure shows that the
due to anomalous subsidence caused by the Walker circu- frequency of occurrence of CCCs and ACSLs is higher during
lation. The present result indicates that the reduction of the rainy season, December through May (Rao and Hada 1990).
NOVEMBER 2021 SOUSA ET AL. 3597

FIG. 5. Interannual variation of (a) CCCs and (b) ACSLs, with red shades for EN and blue shades for LN. The
brighter the color is, the stronger is the ENSO phase. The plus and minus signs on the bottom line indicate the
positive and negative phases, respectively, of the Atlantic dipole. (c) Scatterplot of the number of ACSLs against
the number of CCCs.

The months of March and April are clearly affected by the characteristics of CCCs, inherent to the database used,
highest frequencies of CCCs, mainly after 2006, similar to on a more refined scale, that is, areas of occurrence on a
the results of Rehbein et al. (2017) who have studied the subregional scale, while Nunes et al. (2016) make an
occurrence of mesoscale convective systems (MCS) along amalgamation of intense systems that integrates the results
the Amazon basin. In these months there is the highest over a large area.
intensity of sunlight in the region. During the dry season, A simple estimate gives us an idea of the importance of
and especially during the 4-month period August– ACSLs for the area of our study. If there are five ACSLs in a
November (A, S, O, and N in Fig. 6) (Rao et al. 2015), rainy month, say April, and if the precipitation in each case is
the frequency of the convective systems is very low despite approximately 40 mm, the monthly precipitation due to ACSLs
increased solar heating. That is, solar radiation is not the accumulates to 200 mm, which is a substantial part of the
only factor for increased convective activity. The intensity monthly climatological precipitation shown in Fig. 1b. Thus,
of convection is influenced by total precipitable water, the ACSLs, which are formed from the alignment of CCCs, are
convective available potential energy, wind shear and important contributors to the total rainfall of the coastal
moisture flux convergence that contribute to a significant northern Brazil.
fraction of the total amount of rain in the Amazon, asso- In this study, the dimensions of organized convective sys-
ciated with CCCs and ACSLs. However, during atypical tems are also estimated and the resulting data are analyzed to
years 2006 and 2007 there are more ACSLs even in the dry identify the frequencies of occurrence of Small-, Medium-,
season (Fig. 6b). An increase in the frequency of dry- and Large-ACSL categories. Small ACSLs occur in larger
season ACSLs is seen after 2004, indicating a change in numbers in the dry season (figure not shown). Medium and
their variability. The preferential season of the CCCs Large ACSLs peak in the month of April. In other words,
found here is different from the seasonal preference of the many of the ACSLs occurring in the March–May (MAM)
Potential Severe Convection (PSC) cells over the Guyana season are either Medium or Large. In this season, in addition
Shield, June through September, and over the northeast- to abundant solar radiation, the intertropical convergence
ern Amazon, October through December, found by Nunes zone (ITCZ) is closer to the equator (Chiang et al. 2002;
et al. (2016). These differences are due to the differences Hastenrath 2011; Schneider et al. 2014). The evaporation
in the definitions of the convective system and the subre- from the water-logged Amazon basin supports the convective
gion of study. In addition, the present study preserves the activity.
3598 MONTHLY WEATHER REVIEW VOLUME 149

FIG. 6. Monthly distribution of (a) CCCs and (b) ACSLs during 1998–2014. The darker shade
indicates abundance.

c. Diurnal variability On the whole, the density is high around the Marajó Island
and eastern parts of Amapá state in Brazil. Except in the
Kousky (1980), Cavalcanti (1982), and Cohen et al. (2009)
SON season, the density decreases inward from the coast.
said that the convective systems over the coastal Amazon
This, along with the maximum CCC frequency in the after-
basin develop preferentially in the afternoon hours. To
noon hours, suggests the influence of the coastal sea breeze
quantitatively appreciate this characteristic, the mean diur-
on the development of convection in the study area. In SON
nal variability of CCCs against month of the year is given in
there is CCC activity in some interior pockets, not on the
Fig. 7. The numbers are mean values for the 17 years studied.
coast. This is perhaps due to lack of sufficient moisture
The figure shows larger frequency around 1800 UTC (1500 LT),
convergence in the lower troposphere near the coastal belt.
when the sea breeze is established, than at 1200 UTC (0900 LT)
In the interior over the wet or inundated plains, however, the
and 1500 UTC (1200 LT) in the 4-month season February–
atmosphere has enough moisture in the lower troposphere to
May. In March and April months, the convective systems
support some convection.
are present until late in the night. This indicates that the
unstable conditions persist longer in these months. Figure 7 e. Spatial distribution of CCCs according to their intensities
gives a broad idea of the variability of convective activity
Figure 9 presents a 3-hourly distribution of locations of
along the day during any given month. This type of infor-
CCCVs, CCCIs, and CCCWs on the topographic map of the
mation is considered to be useful for the operational mete-
region from 1500 UTC (1200 LT) to 0300 UTC (0000 LT) for
orologists. (Total number per month is obtained by totaling
the four seasons. The CCCs mostly cluster on the coastal belt of
horizontally. Total number of systems per year for a given
about 300 km as was seen in Fig. 8. One can also see that the
hour of the day is obtained by totaling vertically.) Sodré
number of CCCVs (red) is largest in MAM at all hours of the
et al. (2015) also found the preferential interval of formation
day presented. Another feature is an increase of the numbers
of the circular mesoscale convective systems to be mid-
of CCCIs from 1500 UTC (1200 LT) to 0000 UTC (2100 LT) and a
afternoon near the coast and to be late afternoon away from
decrease afterward. That is, the development and maintenance of
the coast.
d. Spatial variability
Figure 8 presents the spatial density of the CCCs for the
four seasons. The density is the number of CCCs (total of all
the three categories) per 18 latitude by 18 longitude area. The
3-month period MAM presents the highest densities and
September–November (SON) has the lowest densities, con-
trary to the result of Nunes et al. (2016). The quick change in
the density from SON to December–February (DJF) tells us
that the transition from dry season to wet season is sudden.
This can be likened to the onset of the South American
monsoon (Wanzeler da Costa and Satyamurty 2016). The
return to the dry season is slow as we can see that convection
over the northern part of the region remains active in July–
August (JJA). It is known that the retreat of the monsoon is FIG. 7. Hourly distribution of CCCs as function of month, for a
slower than the onset. mean of 17 years. The darker shade indicates higher frequency.
NOVEMBER 2021 SOUSA ET AL. 3599

FIG. 8. Seasonal maps of spatial density of CCCs for (a) December–February (DJF), (b) March–May (MAM),
(c) June–August (JJA), and (d) September–November (SON). Color shading gives the number of occurrences in
the 3-month season.

CCCs in the region continues throughout the afternoon and These effects seem to be necessary for the development and
night hours. Alignments of the CCCs parallel to the coast, i.e., sustenance of convective activity over the region. However,
the formations of ACSLs, are apparent in the figure. The because of the many rivers in the study region, river breezes
mean orientations and locations of the ACSLs in different also develop and influence the convective activity (Matos and
seasons are schematically shown in Fig. 10, and are similar Cohen 2016).
to the result of Cavalcanti (1982). In DJF and MAM sea-
sons the ACSLs extend into the state of Maranhão (MA). 4. Atmospheric conditions associated with
In JJA season they penetrate westward of the state of ACSL occurrence
Amapá (AP).
a. Synoptic situation
Close to the coast there is a tendency for reduction in the
number of CCCVs (red crosses in Fig. 9) after 2100 UTC (1800 Figure 12 shows the ACSL composites of surface pressure,
LT), indicating a movement inland of the very intense con- 925-hPa winds, and the convergence of moisture flux fields
vective region, in all seasons. The ACSLs, as seen from the over South America and the tropical Atlantic for LN con-
alignments of CCCVs, move inward from the coast after ditions in April at 1200 UTC (0900 LT) and 1800 UTC (1500
around 1800 UTC (1500 LT) with a speed of 14 m s21 in good LT). The top line is for the Large-ACSL composites and the
agreement with the estimate given by Cohen et al. (2009) of bottom line is their differences from No-ACSL composites.
16 m s21. The propagation is faster than the speed of the As was mentioned in the last paragraph of the previous
easterly trades (5–10 m s21). This observation indicates that section, the lower-tropospheric winds approaching the study
the propagation is helped by downstream development of region are northeasterly and of the order of 10 m s21 over the
convection (Houze 1977) and the sea breeze (Germano seas. The winds are retarded on the land providing conver-
et al. 2017). gence as schematically shown Fig. 11. The equatorial belt
The roles of sea breeze perpendicular to the coastline, from the Amazon basin all the way up to the coast of West
northeasterly equatorial winds, coastline inclination, surface Africa shows moisture convergence at 1200 UTC (0900 LT)
friction over land and the blocking of surface winds by the (Fig. 12a). Along a narrow belt of the northeastern Amazon
uplands in the northwestern parts of the study region are coast the convergence intensifies (dark blue shade) at
schematically shown in Fig. 11. The sea breeze accelerates the 1800 UTC (1500 LT) (Fig. 12b) when the convection attains
northeasterly equatorial winds over the sea and the surface maximum intensity. The surface pressure and wind fields at
friction over the land surface decelerates the winds to provide the two hours are similar. The differences between the Large-
convergence. The uplands in the western parts of the region ACSL and No-ACSL composites (Figs. 12c,d) are small, al-
also contribute to convergence by blocking near-surface winds. most imperceptible, in the tropics. In the middle latitudes,
3600 MONTHLY WEATHER REVIEW VOLUME 149

FIG. 9. Three-hourly spatial distribution of CCCs for the four seasons DJF, MAM, JJA, and SON. The intensity is according to
Min85PCT values: yellow circles 5 convective elements (CCCW), black plus signs 5 intense convection (CCCI), and red crosses 5 very
intense convection (CCCV).

however, the surface pressure differences east of the South the Florida coast with convective activity over the north coast
American coast around 358S in the Atlantic show anomalous of Brazil.
high pressure area and associated anomalous anticyclonic
b. Vertical structure
winds directed into the Atlantic ITCZ providing additional
convergence. The composites of the vertical profiles of T and u (not
Figure 13 shows the composite synoptic situation for the shown) for the Large-ACSL and No-ACSL situations do
Large-ACSL cases under EN conditions. In this situation the not show appreciable differences. This behavior of the
equatorial convergence is stronger at 1200 UTC (0900 LT) equatorial atmosphere, not presenting significant differ-
than at 1800 UTC (1500 LT). Here also the differences be- ences in temperature profiles between stormy and non-
tween the Large-ACSL and No-ACSL composites are small stormy conditions in the Amazon basin was observed by
in the tropics. In the subtropical South Atlantic near South Tanaka et al. (2014). The north coastal belt is similar to the
American coast there is also an anomalous high pressure Amazon basin in this respect. However, the composite
center in the Large-ACSL situation. That is, for the forma- profiles of ue, u, y, and r present interesting characteristics
tion of a Large ACSL sufficient convergence of moisture is and some appreciable differences between Large-ACSL
necessary irrespective of the phase of ENSO. In the North and No-ACSL days.
Atlantic an anomalous high pressure center near the Florida Figure 14 shows the composite vertical profiles for Large-
coast is seen in the Large-ACSL composite (Figs. 13c,d). ACSL cases in April under LN conditions and their dif-
However, it is difficult to connect the high pressure center off ferences from the No-ACSL composites. High values of ue
NOVEMBER 2021 SOUSA ET AL. 3601

FIG. 11. Schematic representation of the action of continental

surface friction and sea breeze on convergence of wind and
FIG. 10. Schematic representation of seasonal positions of ACSLs. moisture.
Colors indicate seasons.

transport of moisture into the coastal northern Brazil needed

(Fig. 14a) at the surface on the order of 360 K at 1500 UTC for convective activity. The difference in the northerly com-
(1200 LT, red profile) are due to high values of both temper- ponent in the lower troposphere decreases in magnitude from
ature and mixing ratio. The ue decreases from the surface up 2 m s21 at 0900 LT to zero at 1200 LT and becomes 1 m s21
to the middle troposphere indicating conditional instability. southerly at 1800 LT. This indicated that a stronger northerly
Near the surface the lapse rate of ue is stronger, especially at component in the near surface layer in the early stages of the
1500 UTC (1200 LT) (red profile), which makes the surface development of convection favors ACSL formation. In the
layer more unstable. The lapse rate of ue decreases gradually upper troposphere, compensatory southerly winds, outflow
by 2100 UTC (1800 LT) and 0000 UTC (2100 LT). The from the convective tops, are stronger in ACSL environment.
difference from No-ACSL composites (Fig. 14e) shows Figures 14d and 14h show the mixing ratio profiles and dif-
clearly that the Large-ACSL situation is warmer by 2 K at ferences, respectively. It seems the vertical profiles in the
the surface and the anomaly becomes negative by 2100 UTC Large-ACSL situations (Fig. 14d) remain almost unchanged
(1800 LT) and 0000 UTC (2100 LT) when the convection is de- during the day. One interesting characteristic is high values of
intensified. Cooler ue at the surface at 2100 UTC (1800 LT) and mixing ratio (.8 g kg21) even at the 500-hPa level. However,
0000 UTC (2100 LT) indicates cold pool formation after the difference profiles (Fig. 14h) show about 1 g kg21 more
the peak convective activity, as was observed by Oliveira and humidity in the Large-ACSL situation than in the No-ACSL
Oyama (2015). situation in the layer between 850 and 500 hPa, especially at
The composite zonal wind profiles u (Fig. 14b) show east- 2100 UTC (1800 LT). This is mainly due to saturation in the
erlies throughout the troposphere. The profiles at 1200 UTC cloud layer in the Large-ACSL case.
(0900 LT) (blue) and at 1500 UTC (1200 LT) (red), before the
c. Moisture and stability parameters
ACSLs attained their peak activity, show maxima in the lower
troposphere near 800-hPa level. This is similar to the finding of Since most of the CCCs and ACSLs occur in the MAM
Alc^antara et al. (2011) of a low-level jet associated with season, the moisture and stability parameters for the repre-
Amazon coastal squall lines (abbreviated ASL in their study). sentative month of April are presented and are compared with
At 1800 (orange) and 2100 (green) UTC the strength of the the 17-yr climatological mean and standard deviation for the
lower-tropospheric easterlies reduces. This is due to upward same month. Table 1 shows the mean values for the Large
momentum transport by the convective eddies. Thus, after the ACSL and the differences from the means for No-ACSL cases.
peak convective activity the low-level maximum easterly wind They also show the April means (17 April months) and stan-
regains its strength. In the difference profiles (Fig. 14f) the dard deviations with which the differences between Large-
westerly anomalies observed in the lower troposphere in- ACSL and No-ACSL days (Dif) can be compared for judging
dicate deceleration of easterlies in the Large-ACSL cases, the usefulness of the parameters. Table 2 shows the changes in
as if strong convective activity applies breaks to the flow. In the mean values of the indices for Large-ACSL cases from
the upper troposphere the easterlies are accelerated due to 1200 UTC (0900 LT) to 1800 UTC (1500 LT) along with the
eddy momentum flux. respective changes in the mean values of the 510 April-days.
The meridional wind profiles y (Figs. 14c,g) show a near- The K index obtained here in the morning (0900 LT) and in
surface maximum in northerly wind, predominantly after the afternoon (1500 LT) in both EN and LN situations sat-
2100 UTC (1800 LT). This is perhaps required for larger isfies the condition for convective activity. The deviations from
3602 MONTHLY WEATHER REVIEW VOLUME 149

FIG. 12. Composites of regional synoptic features for (top) Large ACSL and (bottom) difference from No-ACSL
situations in April under La Niña conditions. Isolines give surface pressure (hPa), vectors give 925-hPa wind
(m s21), and color shades give divergence of moisture flux (kg m22 s21).

No-ACSL composites (Dif) are very small relative to their activity. That is, Large-ACSL environment in terms of totals
respective climatological standard deviations, perhaps on the and K indices is not appreciably different from No-ACSL en-
order of the observational errors. The standard deviations vironment and the value of the indices do not vary appreciably
of K index for the whole 510 April days are 2.828 and 1.908C, from morning to afternoon hours. Therefore, both totals
respectively at 1200 UTC (0900 LT) and 1800 UTC (1500 LT). and K indices are not useful for forecasting ACSL occurrences.
That is, the K index is not able to distinguish the Large-ACSL A heuristic analysis of the reasons as to why the K and totals
and No-ACSL situations and the Difs are far less than their are always large is as follows. In the expression for K [Eq. (4)],
standard deviations. Only in LN situation at 1800 UTC (1500 the equatorial troposphere has smaller lapse rate [T(850) 2
LT) the Dif is comparable to the standard deviation. The totals T (500)] but Td(850) is larger than in extratropical environ-
index in all situations is around 428C in both the Large-ACSL ment, because of near saturation of the lower troposphere.
and No-ACSL cases, at both the hours and in both the phases Also, the dewpoint depression at 700 hPa is small. Therefore,
of ENSO, satisfying the condition for vigorous thunderstorm the numerical values of K remain large and cannot distinguish
NOVEMBER 2021 SOUSA ET AL. 3603

FIG. 13. As in Fig. 12, but for El Niño conditions.

the situations of intense convection and no convection. Thus, the In general, the mean mixed layer potential temperature rises
index is not useful in the equatorial coastal regions. A similar by 1 K from 1200 to 1800 UTC in all situations. The difference
analysis can be made with the totals index in the equatorial regions. between Large-ACSL and No-ACSL days is near zero at
The CAPE differences between Large ACSL and No ACSL 1200 UTC (0900 LT) and rises to about 0.6 K in EN conditions.
are less than the respective standard deviations in the morning Once again, the differences are less than the standard deviation
as well as in the afternoon hours (Table 1), as with the case of making this index not useful for prediction. Mean mixed layer
totals and K indices. However, the CAPE in the Large-ACSL mixing ratio, rm, also does not show usefulness for prediction in
composites increases considerably from 1200 UTC (0900 LT) EN situation. In the LN situation, however, at 1200 UTC (0900
to 1800 UTC (1500 LT), especially in the LN situations the LT) the difference (Dif) is 4 times the standard deviation and is
difference is about 600 J kg21. Thus, the CAPE, although does considered to be a very useful indicator according to our
not distinguish clearly Large ACSL days from No-ACSL days, premise [Eq. (7)].
its increase from morning to afternoon (Table 2) is considered Precipitable water content w is larger in the EN conditions
useful. The convective inhibition does not show appreciable than in the LN conditions, perhaps due to higher temperatures,
changes and differences. because at higher temperatures the saturation vapor pressure is
3604 MONTHLY WEATHER REVIEW VOLUME 149

FIG. 14. Vertical profiles of (left) equivalent potential temperature ue, (left center) zonal (u) and (right center)
meridional (y) components of wind, and (right) mixing ratio r at 3-h intervals for (a)–(d) Large-ACSL occurrences
and (e)–(h) the difference from No-ACSL occurrences in April under La Niña conditions. Synoptic hours are
indicated with colors.

higher. Surprisingly, its value is somewhat lower on changes from 1200 UTC (0900 LT) to 1800 UTC (1500 LT).
Large-ACSL days than on No-ACSL days at both the Especially, in the LN situations the changes are more than
hours presented, but higher than the climatological mean. double the climatological value. These indices show some
This can be attributed to depletion of moisture due to promise in forecasting work.
precipitation on Large-ACSL days. The difference from
d. Continent–ocean contrast
No-ACSL is smaller by an order of magnitude than its
standard deviation. The thermal contrast between the land surface and the
The afternoon changes (Table 2) are, in general, larger in adjoining sea surface, (Tc 2 To), negative at 1200 UTC
the La Niña ACSL composites relative to the climatological (0900 LT), becomes positive by 1800 UTC (1500 LT), at
values than in the EN composites. The changes in um, rm , both the surface and the 1000 hPa level (Table 1). This
and (Tc 2 To) in Large-ACSL cases are comparable to happens in both the ENSO phases and for Large-ACSL and
changes in their respective climatological values; um in- No-ACSL days and the differences between the two ex-
creases by more than 1.58C, rm decreases by nearly 2 g kg21, treme weather conditions are small, smaller than respective
and the thermal contrast between the land surface and the climatological standard deviations (Table 1). This shows
adjoining sea surface (Tc 2 To) increases by more than 28C. that conditions for sea breeze in the afternoon hours are
The K index and totals index also do not show large changes always present. The afternoon change in (Tc 2 To) at the
relative to the climatological changes, except that in the LN surface in the Large-ACSL cases in April is about 2.78C,
situation K decreases by 18C and totals increases by 0.728C. comparable to the climatological value (Table 2). The
Somewhat weaker and opposite behavior is observed in the contrasts are stronger in the dry season (October) (not
EN situations. The precipitable water changes in the EN shown) than in the rainy season (April) because solar ra-
situation and in climatology are positive and are larger than diation does not suffer reflection by cloudiness. This pa-
1.5 mm. This is essentially due to evaporation during the rameter also does not give a clue for the formation of
day. However, in the LN situation the change is negligible. ACSL. However, since the convective activity reaches
This can be attributed to an approximate balance between maximum intensity and the aligned CCCs move inland
the combined effect of evaporation and convergence of in the afternoon hours, the participation of sea breeze in
moisture and the precipitation in the ACSL during the af- the formation and displacement of CCCs and ACSLs is
ternoon hours in LN situations. The combined changes in considered important.
CAPE and CINE, i.e., changes in (CAPE 2 CINE) values, Wright et al. (2017) showed that the precipitation from
in the ACSL composites are larger than the climatological shallow clouds increases the relative humidity in the free
NOVEMBER 2021 SOUSA ET AL. 3605

troposphere, reduces convective inhibition and increases The atmospheric conditions associated with the Medium-
CAPE, thus increasing instability of the atmospheric column. and Large-ACSL development show wind maxima in the lower
This makes the convection to attain greater horizontal and and middle portions of the troposphere accompanied by higher
vertical extension and to produce higher volume of rainfall in humidity and instability in the layer between 800 and 400 hPa,
the wet season (Zhuang et al. 2017). especially in March and April months. The development of
deep convection and the organization of CCCs and ACSLs is
favored by the strengthening of the northeasterly trade winds
5. Summary and discussion
associated with the southward intrusion of the ITCZ during the
The mesoscale convective cloud clusters (CCCs) and March and April months (Schneider et al. 2014; Chakraborty
their alignments into Amazon Coastal Squall Lines (ACSLs) et al. 2020). The convergence in the ITCZ is enhanced due to
along the coastal belt from the state of Maranhão in Brazil up frictional convergence as schematically depicted in Fig. 11. The
to the Guyanas region are important weather systems. Nearly northerly component of wind due to the combined effect of
40% of the annual rainfall in this belt is attributed to these trades and sea breeze advects the convective region inland.
systems. The convective activity sustains itself through regeneration of
The analysis of TRMM data performed in the present study cumulus convection downstream even after the sea breeze
utilizing a semiautomatic method is able to provide wide- ceases (Houze 1977).
ranging statistics of the diurnal and seasonal frequencies of Small ACSLs are more frequent in JJA months when the
occurrence of CCCs and their geographic distribution along solar heating is more intense. In JJA cooler and stable atmo-
the coastal northern Amazon. The characterization of the sphere above 800 hPa extends up to the 400-hPa level and the
ambient meteorological conditions favorable for the evolu- atmospheric column has less moisture. Support for the devel-
tion of CCCs into Large ACSLs, shown in Figs. 12–14 is opment of Medium and Large ACSL is absent mainly because
considered important. Convection over the Amazon region of weakening of northerly flow in the lower troposphere.
causes increase in the rate of entrainment into the clouds in Atmospheric conditions such as precipitable water exceeding
low levels (Becker and Hohenegger 2021) and as the rate of 55 mm, CAPE nearing 1000 J kg21, and vertical shear larger
entrainment grows, convection becomes deeper and orga- than 2 m s21 per km are determining factors for the onset of
nizes into squall lines. Strong mesoscale convergence as- convection in the late-morning and afternoon hours and for its
sociated with squall lines intensifies vertical motion inside deepening in the evening and night hours in the Amazon re-
the clouds and generates turbulence favoring clustering of gion (Chakraborty et al. 2020).
convection. The monthly distribution of ACSLs obtained here is dif-
The CCCs populate the coastal belt of about 300 km width ferent from the monthly frequency of the squall lines discussed
(Figs. 8 and 9). They are mostly found in the 8-month period in Cohen et al. (2009). Their figure shows an almost uniform
December–July (Fig. 6a) and preferential regions of oc- distribution of squall lines throughout the year. In the present
currence are observed over the Marajó Island and coastal study there is a significant preference for the 3-month period
parts of Amapá state in Brazil east of the Guyana Shield March–May (Fig. 6b) for the ACSLs. The visual technique
uplands. This result disagrees with Nunes et al. (2016) who utilized by earlier authors was purely subjective and the
found maximum frequency in October–December over the statistics presented here with higher spatial resolution of
northeastern subregion of the Amazon and in June–August TRMM data and for longer period (17 years) are considered
over the Guyana Shield. more robust.
Along the day the intense and very intense convective Interannual variabilities (Fig. 5) show an increase in the
clusters increase their numbers from 1200 to 1800 LT, during CCC count and a decrease in the ACSL count in LN years.
the time the sea breeze is established. Their diurnal frequency That is, an ACSL in an LN year contains more CCCs on av-
is highest around 1500 LT. However, they persist through the erage and fewer CCCs constitute an ACSL in EN years. The
early night hours and move inward at speed of ;14 m s21, decrease in rainfall in EN years and the increase in LN years in
faster than the speed of the equatorial easterlies (5–10 m s21), the coastal regions of eastern Amazon basin (Andreoli et al.
indicating that the propagation is helped by downstream 2016; Viegas et al. 2019) can be attributed to changes in the
development (Houze 1977) and sea breeze. As a result of frequency of CCCs. In EN years the subsidence limb of the
continental surface friction and uplands in the northwestern Walker circulation affects the northeastern Amazon basin in-
portion of the study region, there is velocity and moisture hibiting or reducing the formation of CCCs.
convergence, as shown in the schematic diagram (Fig. 11), to The analysis of the composite vertical profiles shows that the
support convection. surface meteorological conditions of wind and temperature,
The ITCZ is closest to the Brazilian coast in the MAM the moisture content in the free atmosphere, instability and
season and, hence, the frequencies of CCCs and Medium wind shear in the coastal belt are essential for controlling the
and Large ACSLs are the highest in this season. The fre- convection and the organization of CCCs into Large, Medium,
quency of Large ACSLs is 5 per month in March and April and Small ACSLs as was mentioned in Zhuang et al. (2017),
and diminishes to less than 1 per month in September. This Chakraborty et al. (2020) and Becker and Hohenegger (2021).
result is slightly different from the result of Alc^antara et al. The thermodynamic variables and instability parameters as-
(2011) where the frequency was highest in the April to sociated with Large ACSLs (Table 2) show the importance of
June season. mixed layer mixing ratio rm and CAPE, especially in the LN
3606 MONTHLY WEATHER REVIEW VOLUME 149

situation. The CAPE increases by more than 600 J kg21 from PAPAC program, for financial support (062.00859/2019-
0900 to 1500 LT when the convective activity attains maximum PAPAC/2019). The third author thanks CAPES for the
intensity. The rm in Large-ACSL composites is larger than in PVNS grant (Process: 23038.019802/2018-07) and CNPq for
No-ACSL composites by nearly 3.5 g kg21 in LN years. The ue the Productivity in Research grant (Process: 301370/2017-6).
profiles show large lapse-rates indicating conditional instability
necessary for the development of convection. The lapse rates Data availability statement. All of the datasets used in this
decrease gradually from 1200 to 2100 LT. Vigorous convection study are available at the sites or from the institutions men-
is understandably sustained in late-afternoon and night hours tioned in section 2.
by liberation of latent heat. Profiles associated with Large
ACSLs show jet-like structure in both the u and y (Fig. 14) REFERENCES
components of wind. The meridional wind maximum near the Alc^antara, C. R., M. A. F. Silva Dias, E. P. Souza, and J. C. P.
925-hPa level is northerly and increases from 1200 to 2100 LT. Cohen, 2011: Verification of the role of the low level jets in
This suggests that, as the convective activity strengthens, more Amazon squall lines. Atmos. Res., 100, 36–44, https://doi.org/
humid air from the north is drawn. The zonal wind maximum is 10.1016/j.atmosres.2010.12.023.
easterly and is around the 800-hPa level, agreeing with Andreoli, R. V., S. S. de Oliveira, M. T. Kayano, J. Viegas, R. A. F.
Alc^antara et al. (2011). The easterly maxima are larger at 0900 de Souza, and L. A. Candido, 2016: The influence of different
LT, at the starting time of convection. When the atmospheric El Niño types on the South American rainfall. Int. J. Climatol.,
column is not absolutely unstable (i.e., ›u/›z . 0), larger ver- 37, 1374–1390, https://doi.org/10.1002/joc.4783.
tical wind shear means smaller Richardson number, which fa- Becker, T., and C. Hohenegger, 2021: Entrainment and its depen-
dency on environmental conditions and convective organization
vors triggering of instability.
in convection-permitting simulations. Mon. Wea. Rev., 149, 537–
Other parameters such as K and totals (Table 1) do not show
550, https://doi.org/10.1175/MWR-D-20-0229.1.
neither appreciable contrasts between Large-ACSL days and Betts, A. K., R. W. Grover, and M. W. Moncrieff, 1976: Structure
No-ACSL days nor appreciable afternoon changes. Tanaka and motion of tropical squall-lines over Venezuela. Quart. J. Roy.
et al. (2014) also obtained small differences in the vertical Meteor. Soc., 102, 395–404, https://doi.org/10.1002/qj.49710243209.
profiles of temperature, humidity mixing ratio and related Bolton, D., 1980: The computation of equivalent potential tem-
parameters between heavy rainfall and no-rainfall days in the perature. Mon. Wea. Rev., 108, 1046–1053, https://doi.org/
Amazon basin. Both rm and precipitable water w show higher 10.1175/1520-0493(1980)108,1046:TCOEPT.2.0.CO;2.
values in EN situations than in LN situations. This is due to lack Cavalcanti, I. F. A., 1982: Um estudo sobre interações entre sis-
of convection that could deplete atmospheric moisture through temas de circulação de escala sinótica e circulações locais (A
study of interactions between synoptic-scale circulation systems
precipitation.
and local circulations). Dissertation, Department of Meteorology,
The results presented indicate that some subtle changes in
Instituto Nacional de Pesquisas Espaciais (INPE), São José dos
the flow and vertical structure trigger the intense convection Campos, São Paulo, Brazil, 140 pp.
associated with CCCs and their alignment into ACSLs. An Chakraborty, S., J. H. Jiang, H. Su, and R. Fu, 2020: Deep convective
experimental campaign to study the sea breeze and the lower- evolution from shallow clouds over the Amazon and Congo
tropospheric mixed layer over the sea and over the continent is rainforests. J. Geophys. Res. Atmos., 125, e2019JD030962,
necessary for understanding the characteristics accompanying https://doi.org/10.1029/2019JD030962.
and preceding the convective activity and to find robust pre- Chiang, J. C. H., Y. Kushnir, and A. Giannini, 2002: Deconstructing
diction indicators of these systems. Figures 6–14 along with the Atlantic intertropical convergence zone variability: Influence of
geostationary satellite imagery and radar imagery can be of the local cross-equatorial sea surface temperature gradient and
some help. remote forcing from the eastern equatorial Pacific. J. Geophys.
Res., 107, 4004, https://doi.org/10.1029/2000JD000307.
For the observational datasets, a semiobjective metric for
Cohen, J. C. P., M. A. F. Silva Dias, and C. A. Nobre, 1995:
the precipitation potential is created that utilizes the dif-
Environmental conditions associated with Amazonian squall line:
ferences of brightness temperature with finer resolution and A case study. Mon. Wea. Rev., 123, 3163–3174, https://doi.org/
more recent data to produce a 17-yr climatology of the 10.1175/1520-0493(1995)123,3163:ECAWAS.2.0.CO;2.
CCCs and ACSLs over the North-Northeast coastal belt of ——, I. F. A. Cavalcanti, R. H. M. Braga, and L. Santos Neto, 2009:
South America. This metric permits not only identification Linhas de Instabilidade na costa N-NE da América do Sul
of the CCCs and ACSLs but also categorization of the sys- (Lines of instability on the N-NE coast of South America).
tems according to their intensity and size, which was not Tempo e Clima no Brasil, I. F. A. Cavalcanti et al., Eds.,
done before. Oficina de Textos, 75–93.
Copernicus Climate Change Service, 2017: ERA5: Fifth gen-
Acknowledgments. Material presented here is part of the eration of ECMWF atmospheric reanalyses of the global
climate. Copernicus Climate Change Service Climate Data
doctoral thesis of the first author supervised by the second
Store (CDS), accessed 11 June 2019, https://cds.climate.copernicus.eu/
author. The first and second authors thank CAPES for the fi-
cdsapp#!/home.
nancial support (Process: PDPG-Amaz^ onia-Legal-CAPES- Emanuel, K. A., 1994: Atmospheric Convection. Oxford University
88887.510154/2020-00). The first author thanks Fundação de Press, 580 pp.
Amparo à Pesquisa do Estado do Amazonas (FAPEAM), Galway, J. G., 1956: The lifted index as a predictor of latent in-
for a Ph.D. scholarship (POSGRAD-018/2015/FAPEAM). stability. Bull. Amer. Meteor. Soc., 37, 528–529, https://doi.org/
The second author thanks FAPEAM within the scope of the 10.1175/1520-0477-37.10.528.
NOVEMBER 2021 SOUSA ET AL. 3607

Garstang, M., H. L. Massie, J. Halverson, S. Greco, and J. Scala, and intensity comparison over tropical oceans and conti-
1994: Amazon coastal squall lines. Part I: Structure and ki- nents. Mon. Wea. Rev., 124, 2417–2437, https://doi.org/
nematics. Mon. Wea. Rev., 122, 608–622, https://doi.org/ 10.1175/1520-0493(1996)124,2417:MCSDBT.2.0.CO;2.
10.1175/1520-0493(1994)122,0608:ACSLPI.2.0.CO;2. Negrón-Juárez, R. I., and Coauthors, 2010: Widespread Amazon
George, J. J., 1960: Weather Forecasting for Aeronautics. Academic forest tree mortality from a single cross-basin squall line
Press, 673 pp. event. Geophys. Res. Lett., 37, L16701, https://doi.org/10.1029/
Germano, M. F., M. I. Vitorino, J. C. P. Cohen, G. B. Costa, J. I. de 2010GL043733.
Oliveira Souto, M. T. C. Rebelo, and A. M. L. de Sousa, 2017: Nesbitt, S. W., E. J. Zipser, and D. J. Cecil, 2000: A census of
Analysis of the breeze circulations in Eastern Amazon: An precipitation features in the tropics using TRMM: Radar,
observational study. Atmos. Sci. Lett., 18, 67–75, https:// ice scattering, and lightning observations. J. Climate, 13,
doi.org/10.1002/asl.726. 4087–4106, https://doi.org/10.1175/1520-0442(2000)013,4087:
Greco, S., and Coauthors, 1990: Rainfall and surface kinematic con- ACOPFI.2.0.CO;2.
ditions over central Amazonia during ABLE 2B. J. Geophys. Nunes, A. M. P., M. A. F. Silva Dias, E. M. Anselmo, and C. A.
Res., 95, 17 001–17 014, https://doi.org/10.1029/JD095iD10p17001. Morales, 2016: Severe convection features in the Amazon
Grimm, A. M., 2003: The El Niño impact on the summer monsoon in Basin: A TRMM-based 15-year evaluation. Front. Earth Sci.,
Brazil: Regional processes versus remote influences. J. Climate, 4, 37, https://doi.org/10.3389/feart.2016.00037.
16, 263–280, https://doi.org/10.1175/1520-0442(2003)016,0263: Oliveira, F. P., and M. D. Oyama, 2015: Antecedent atmospheric
TENIOT.2.0.CO;2. conditions related to squall-line initiation over the northern
Hastenrath, S., 2011: Exploring the climate problems of Brazil’s coast of Brazil in July. Wea. Forecasting, 30, 1254–1264, https://
Nordeste: A review. Climatic Change, 112, 243–251, https:// doi.org/10.1175/WAF-D-14-00120.1.
doi.org/10.1007/s10584-011-0227-1. ——, and ——, 2020: Squall-line initiation over the northern coast
Holton, J., 2004: An Introduction to Dynamic Meteorology. 4th ed. of Brazil in March: Observational features. Meteor. Appl., 27,
Elsevier Academic Press, 535 pp. e1799, https://doi.org/10.1002/met.1799.
Houze, R. A., 1977: Structure and dynamics of a tropical squall-line Parker, D. J., 2003: Mesoscale meteorology: Overview. Encyclopedia
system. Mon. Wea. Rev., 105, 1540–1567, https://doi.org/ of Atmospheric Sciences, J. R. Holton, Ed., Academic Press,
10.1175/1520-0493(1977)105,1540:SADOAT.2.0.CO;2. 1237–1243, https://doi.org/10.1016/B0-12-227090-8/00478-4.
Kousky, V. E., 1980: Diurnal rainfall variation in the northeast Prosser, N. E., and D. S. Foster, 1966: Upper air sounding analysis by
Brazil. Mon. Wea. Rev., 108, 488–498, https://doi.org/10.1175/ use of an electronic computer. J. Appl. Meteor., 5, 296–300, https://
1520-0493(1980)108,0488:DRVINB.2.0.CO;2. doi.org/10.1175/1520-0450(1966)005,0296:UASABU.2.0.CO;2.
Liu, C., 2007: Description of products in University of Utah Rao, V. B., and K. Hada, 1990: Characteristics of rainfall over
Database. University of Utah, accessed 24 January 2018, Brazil: Annual variations and connections with the Southern
http://www.atmos.illinois.edu/;snesbitt/webdocs/pf6.1.pdf. Oscillation. Theor. Appl. Climatol., 42, 81–91, https://doi.org/
——, E. Zipser, D. Cecil, S. Nesbitt, and S. Sherwood, 2008: A 10.1007/BF00868215.
cloud and precipitation feature data base from nine years ——, S. H. Franchito, C. M. E. Santo, and M. A. Gan, 2015: An
of TRMM observations. J. Appl. Meteor. Climatol., 47, update on the rainfall characteristics of Brazil: Seasonal var-
2712–2728, https://doi.org/10.1175/2008JAMC1890.1. iations and trends in 1979–2011. Int. J. Climatol., 36, 291–302,
Loureiro, R., J. M. Saraiva, I. Saraiva, R. C. Senna, and A. S. Fredó, https://doi.org/10.1002/joc.4345.
2014: Estudo dos eventos extremos de precipitação ocorridos Rehbein, A., T. Ambrizzi, and C. R. Mechoso, 2017: Mesoscale con-
em 2009 no Estado do Pará (Study of extreme precipitation vective systems over the Amazon basin. Part I: Climatological
events that occurred in 2009 in the State of Pará). Rev. Bras. aspects. Int. J. Climatol., 38, 215–229, https://doi.org/10.1002/
Meteor., 29, 83–94, https://doi.org/10.1590/0102-778620130054. joc.5171.
Machado, L. A. T., and Coauthors, 2014: The CHUVA project: How Ropelewski, C. F., and M. S. Halpert, 1987: Global and re-
does convection vary across Brazil? Bull. Amer. Meteor. Soc., 95, gional scale precipitation patterns associated with the
1365–1380, https://doi.org/10.1175/BAMS-D-13-00084.1. El Niño/Southern Oscillation. Mon. Wea. Rev., 115, 1606–1626,
Maddox, R. A., 1980: Mesoscale convective complexes. Bull. https://doi.org/10.1175/1520-0493(1987)115,1606:GARSPP.
Amer. Meteor. Soc., 61, 1374–1387, https://doi.org/10.1175/ 2.0.CO;2.
1520-0477(1980)061,1374:MCC.2.0.CO;2. ——, and ——, 1989: Precipitation patterns associated with the
Matos, A. P., and J. P. Cohen, 2016: Circulação de brisa e a banda high index phase of Southern Oscillation. J. Climate, 2,
de precipitação na margem leste da baía do Marajó (Breeze 268–284, https://doi.org/10.1175/1520-0442(1989)002,0268:
circulation and precipitation band on the east bank of PPAWTH.2.0.CO;2.
Marajó Bay). Ci^ enc. Nat., 38, 21–27, https://doi.org/10.5902/ Sadowski, A. F., and R. E. Rieck, 1977: Stability indices. NOAA
2179460X19814. NWS Program Requirements & Planning Division Tech.
McGaughey, G., E. J. Zipser, R. W. Spencer, and R. E. Hood, Procedures Bulletin TPB-207, 8 pp.
1996: High-resolution passive microwave observations Satyamurty, P., and M. B. Rosa, 2020: Synoptic climatology of
of convective systems over the tropical Pacific Ocean. tropical and subtropical South America and adjoining seas
J. Appl. Meteor., 35, 1921–1947, https://doi.org/10.1175/ as inferred from Geostationary Operational Environmental
1520-0450(1996)035,1921:HRPMOO.2.0.CO;2. Satellite imagery. Int. J. Climatol., 40, 378–399, https://doi.org/
Miller, R. C., 1967: Notes on analysis and severe storm forecasting 10.1002/joc.6217.
procedures of the military weather warning center. USAF AWS Schneider, T., T. Bischoff, and G. H. Haug, 2014: Migrations and
Headquarters Tech. Rep. 200, 184 pp., https://apps.dtic.mil/dtic/tr/ dynamics of the intertropical convergence zone. Nature, 513,
fulltext/u2/744042.pdf. 45–53, https://doi.org/10.1038/nature13636.
Mohr, K. I., and E. J. Zipser, 1996: Mesoscale convective sys- Smith, E. A., A. Mugnai, H. J. Cooper, G. J. Tripole, and X. Xiang,
tems defined by their 85-GHz ice scattering signature: Size 1992: Foundations for statistical-physical precipitation retrieval
3608 MONTHLY WEATHER REVIEW VOLUME 149

from passive microwave satellite measurements. Part I: Brightness- Viegas, J., and Coauthors, 2019: Caracterização dos diferentes ti-
temperature properties of a time-dependent cloud-radiation pos de El Niño e seus impactos na América do Sul a Partir de
model. J. Appl. Meteor., 31, 506–531, https://doi.org/10.1175/ dados observados e modelados. Rev. Bras. Meteor., 34, 43–67,
1520-0450(1992)031,0506:FFSPPR.2.0.CO;2. https://doi.org/10.1590/0102-7786334015.
Sodré, G. R., M. I. Votorino, J. C. P. Cohen, and B. C. Moraes, Vivekanandan, J., J. Turk, G. L. Stephens, and V. N. Bringi,
2015: Estudo observacional da convecção de mesoescala em 1990: Microwave radiative transfer studies using combined
diferentes superfícies no estado do Pará. (Study of mesoscale multiparameter radar and radiometer measurements during
convection in different areas in Pára state). Rev. Bras. Geogr. COHMEX. J. Appl. Meteor., 29, 561–585, https://doi.org/
Fís., 8, 1281–1293, https://doi.org/10.5935/1984-2295.20150068. 10.1175/1520-0450(1990)029,0561:MRTSUC.2.0.CO;2.
Souza, E. B., M. T. Kayano, J. Tota, L. Pezzi, G. Fisch, and Wanzeler da Costa, C. P. W., and P. Satyamurty, 2016: Inter-
C. Nobre, 2000: On the influences of the El Niño, La Niña and hemispheric and inter-zonal moisture transports and mon-
Atlantic Dipole pattern on the Amazonian rainfall during soon regimes. Int. J. Climatol., 36, 4705–4722, https://doi.org/
1960-1998. Acta Amazon., 30, 305–318, https://doi.org/10.1590/ 10.1002/joc.4662.
1809-43922000302318. Wright, J. S., R. Fu, J. R. Worden, S. Chakraborty, N. E. Clinton,
Spencer, R., H. Goodman, and R. Hood, 1989: Precipitation re- C. Risi, Y. Sun, and L. Yin, 2017: Rainforest-initiated wet sea-
trieval over land and ocean with the SSM/I: Identification and son onset over the southern Amazon. Proc. Natl. Acad. Sci.
characteristics of the scattering signal. J. Atmos. Oceanic USA, 114, 8481–8486, https://doi.org/10.1073/pnas.1621516114.
Technol., 6, 254–273, https://doi.org/10.1175/1520-0426(1989) Zhuang, Y., R. Fu, J. A. Marengo, and H. Wang, 2017: Seasonal
006,0254:PROLAO.2.0.CO;2. variation of shallow-to-deep convection transition and its link
Tanaka, L. M. S., P. Satyamurty, and L. A. T. Machado, 2014: to the environmental conditions over the Central Amazon.
Diurnal variation of precipitation in central Amazon Basin. Int. J. Geophys. Res. Atmos., 122, 2649–2666, https://doi.org/10.1002/
J. Climatol., 34, 3574–3584, https://doi.org/10.1002/joc.3929. 2016JD025993.
Copyright of Monthly Weather Review is the property of American Meteorological Society
and its content may not be copied or emailed to multiple sites or posted to a listserv without
the copyright holder's express written permission. However, users may print, download, or
email articles for individual use.