Synoptic Climatology of Subtropical Cyclone Impacts on Near-Surface Winds Over the South Atlantic Basin

Abstract

Winds and precipitation are two meteorological variables of great impact on the economy and society, especially when extreme events occur. The objective of this study is to investigate the subtropical cyclone (SC) impacts on near-surface wind extremes (defined by 95th percentile) and precipitation in the South Atlantic basin. For the period 1979–2015, the analysis used two reanalysis data (ERA5 and Climate Forecast System Reanalysis [CFSR]) and estimated precipitation from the Tropical Rainfall Measuring Mission (TRMM) considering the extended warm (from September to February) and cold (from March to August) seasons. A total of 169 SCs occurred in the warm season against 87 in the cold one. Two approaches are used to quantify SC impacts: Lagrangian, where the composites follow the cyclone center, and Eulerian, which considers the averaged environment during the cyclone lifecycle. In general, both approaches indicate SC contribution to the positive anomalies of mean and extreme near-surface wind speeds to the east and northeast of a cyclogenetic region in southeastern Brazil (RG1). For all analyses conducted, wind speeds from CFSR are higher than in ERA5. The SC activity contributes to the enhancement of precipitation to the east and inside RG1 in an organized northwest-southeast positive anomaly band from the continent to the ocean, while favoring drier conditions to the south. These positive precipitation anomalies are associated with an anomalous low pressure, low-level moisture flux convergence and warm advection occurring at the same time with an upper-level divergent flow near RG1. All these features are intensified in the cold season.

Introduction

Winds result from the differential warming in diverse time and spatial scales in the atmosphere and are fundamental for the maintenance of the climate system. Furthermore, their application as a form of renewable energy has increased over the time. The impact of the wind speed on the economic sector has brought great interest in the scientific world; for instance, the occurrence of intense wind, gusts, and extremes associated with synoptic systems.

Several authors (Bitencourt et al., 2011; da Rocha et al., 2004; de Jesus et al., 2021; Gramcianinov et al., 2021) related near-surface intense winds with cyclonic systems development over the southwestern South Atlantic Ocean, and in particular the southeastern coast of Brazil (Campos et al., 2018; da Rocha et al., 2004). Among the types of cyclones developing in this region are the subtropical ones, which are hybrid systems showing characteristics from the tropical (warm core at low levels) and extratropical (cold at upper levels) ones (da Rocha et al., 2019; Hart, 2003). As presented by Yanase et al. (2014) and synthesized by da Rocha et al. (2019), subtropical cyclones (SCs) develop in different oceanic basins of the world, as eastern Australia where they are named “east coast cyclones” (Cavicchia et al., 2018; Hopkins & Holland, 1997; Pepler et al., 2016), North Atlantic (Evans & Guishard, 2009; González-Alemán et al., 2015; Guishard, 2006), South Atlantic (de Jesus et al., 2021; Evans & Braun, 2012; Gozzo et al., 2014), Pacific (Caruso & Businger, 2006; Silva et al., 2022), and Mediterranean (Cavicchia et al., 2014; Fita et al., 2007).

The increasing interest in SCs in the South Atlantic basin is related to their potential to transition to tropical, as occurred in 2004 with the first hurricane in the basin named Catarina (McTaggart-Cowan et al., 2006; Pezza & Simmonds, 2005; Veiga et al., 2008). After hurricane Catarina, the first SC studied in the South Atlantic basin was Anita (de Abreu & da Rocha, 2015; Dias Pinto et al., 2013; Dutra et al., 2017; Reboita et al., 2017a), while the first two SC climatologies were elaborated by Evans and Braun (2012) and Gozzo et al. (2014). Using ERA-40 reanalysis for the period 1957–2007, Evans and Braun (2012) found a total of 63 SCs, which results in ∼1.2 events per year. On the other hand, Gozzo et al. (2014) using NCEP1 and ERA-Interim reanalysis obtained ∼7 (with standard deviation of ±3) SCs per year. In terms of seasonality, Evans and Braun (2012) obtained more SCs occurring in autumn–winter, while for Gozzo et al. (2014) they are more frequent in summer–autumn. This difference is explained by some difference in the methodology, since Gozzo et al. (2014) did not impose the criteria of maximum wind at 925 hPa and cutoff low in mid-upper levels as did Evans and Braun (2012). A long-term climatology of SCs, including present and future climate scenarios, was recently obtained by de Jesus et al. (2022), who showed a negative trend in the frequency and an intensification of SCs in the future (2050–2080).

The mechanisms favoring SC development includes the presence of a cutoff low or a small-amplitude trough as forcings at mid-upper levels (da Rocha et al., 2019; Gozzo et al., 2014; Holland et al., 1987; McTaggart-Cowan et al., 2006; Reboita et al., 2017b), which also occurs over the South Atlantic basin (da Rocha et al., 2019; Gozzo et al., 2014). In addition, their maintenance occurs by thermodynamic and dynamic processes; while the thermodynamic component is related to the evaporation from the ocean, which contributes to the warm core at low levels and near-surface perturbation, the dynamic component assists with a cyclone or trough to the north (low levels) of an anticyclone at upper levels resembling an atmospheric blocking dipole at mid-upper levels (type Rex; Rex, 1950) and mass divergence to the east of the trough, contributing to upward motion (da Rocha et al., 2019; Gozzo et al., 2014; Reboita et al., 2017b).

In oceanic basins of the world, SCs impact coastal areas by inducing stronger winds and rainfall (Quinting et al., 2019; Reboita, da Rocha, et al., 2019). For instance, over the South Atlantic basin, Reboita, da Rocha, et al. (2019) analyzed the lifecycle of six SCs and showed that they produced intense near-surface wind speeds, exceeding in various time steps 17 m s⁻¹ at 925 hPa.

Case studies highlighted the relevance of SC impacts (e.g., Brasiliense et al., 2017; Dalagnol et al., 2021; de Souza & da Silva, 2021; Silva et al., 2022); however, there is not a long period analysis on this subject over the South Atlantic basin. Therefore, this study aims to obtain a long-term climatology of the impacts of SCs on near-surface wind speed extremes and precipitation in this basin. The analysis considers two different reanalyses (ERA5 and Climate Forecast System Reanalysis [CFSR]) and approaches (Eulerian and Lagrangian). The paper is organized as follows: Sections 2 and 3 describe the data and methodology, Section 4 the results, and Section 5 conclusions.

Data Reanalysis

In this study, we use the variables (wind components at 10 m height, geopotential height at 500 hPa, sea level pressure (SLP), mass divergence and wind components at 200 hPa, temperature advection, and moisture flux divergence at 925 hPa) at each 6 hr (00, 06, 12 and 18 UTC) from the reanalysis: (a) CFSR (Saha et al., 2010) and Climate Forecast System Reanalysis Version 2 (CFSR v2, Saha et al., 2014) with horizontal resolution of 0.5 latitude by longitude; (b) ERA5 from the European Center for Medium-Range Weather Forecast-Copernicus Climate Change Service (ECMWF-C3S; Hersbach et al., 2018) with grid spacing of 0.25° (latitude and longitude). The precipitation is given by the Tropical Rainfall Measuring Mission (TRMM, Huffman et al., 2007), which is a gridded (0.5 of horizontal resolution) satellite product available at each 3 hr, and it is used to obtain the accumulated precipitation at each 6 hr. In function of data availability, the analysis covers the period 1979–2015 for reanalysis and only 1998–2015 for TRMM.

Cyclone Tracking and Classification

SC classification is provided by Gozzo et al. (2014, 2017) and it consists of two steps: (a) all cyclones are tracked based on the relative vorticity (Reboita, da Rocha, Ambrizzi, & Sugahara, 2009; Sugahara, 2000) using ERA-Interim reanalysis (Dee et al., 2011); (b) cyclones are classified as SC through the Cyclone Phase Space (CPS; Hart, 2003) and must sustain the subtropical features for at least 36 hr of their life cycles (Gozzo et al., 2014). In this case, the CPS parameters were calculated using the ERA-Interim reanalysis at each 6 hr by using geopotential, SLP, and u and v wind components in nine vertical levels (1,000, 925, 900, 850, 700, 600, 500, 400, and 300 hPa).

The algorithm for tracking all cyclones considers the cyclonic relative vorticity (negative values in the Southern Hemisphere) of the wind at 925 hPa and involves three main stages: (a) identification of minimum relative vorticity, which must be less or equal to −1.5 × 10⁻⁵ s⁻¹; (b) location after the first step, known as the first guess; (c) searching for the next positions by extrapolating the mean velocity between two previous successive locations. Furthermore, the cyclones must exist for at least 24 hr and a maximum of 10 days. The tracking area and most frequent SC genesis regions (named as RG1 by Gozzo et al., 2014) are shown in Figure 1.

Figure 1. Cyclone tracking domain (black rectangle) and RG1 (red rectangle) following Gozzo et al. (2014). Some Brazilian states mentioned in the text are also indicated in the map.

To classify the SCs, the dynamic and thermodynamic structures of the cyclones are analyzed based in three parameters: (a) thermal symmetry (B), (b) low-level (900–600 hPa layer) thermal wind ($$-{V}_{T}^{L}$$), and (c) upper-level (600–300 hPa layer) thermal wind ($$-{V}_{T}^{U}$$). The thermal symmetry measures the difference of the thickness of the 900–600 hPa layer between two semicircles with a 500 km radius from the cyclone center. For the South Atlantic basin, Gozzo et al. (2014, 2017) and de Jesus et al. (2022) classified a cyclone as subtropical when −25 m < B < 25 m, $$-{V}_{T}^{L}$$ ≥ −50, and (c) $$-{V}_{T}^{U}$$ ≤ −10. These values guarantee the presence of thermal symmetry and hybrid structure, which has to be satisfied for at least 36 consecutive hours (de Jesus et al., 2022; Gozzo et al., 2014). Other two criteria for SC classification (Evans & Braun, 2012; Guishard, 2006), such as the gale-force wind threshold and the presence of a cutoff low at 500 hPa, were not considered by Gozzo et al. (2014).

Methods Climatology of Extremes

Over the South Atlantic basin, some case studies have shown that one major environmental impact of SCs is the intense near-surface winds and precipitation (Dalagnol et al., 2021; Reboita, da Rocha, et al., 2019; Silva et al., 2022). However, there is no climatology indicating what sectors of the basin are more affected by these features. Here, we evaluate the near-surface winds associated with SCs by selecting the extreme events considering the 95th percentile, since this threshold is frequently used to evaluate extreme winds (Crespo et al., 2022; da Silva, 2013; da Silva et al., 2022). For this, two extended seasons are considered, the warm (from September to February—SONDJF) and cold (from March to August—MAMJJA) ones. For each of these seasons, an extreme occurs when the wind speed at 10 m height exceeds the 95th percentile at each synoptic time (0000, 0600, 1200 and 1800 UTC) and grid point.

Impacts of Subtropical Cyclones

The association between SCs and their features (winds and precipitation) considers two frames of reference: Eulerian and Lagrangian.

For the Lagrangian approach, the method accumulates the variable in a box following the SC center location at each 6 hr (as in Figures 2a and 2b), and then the wind speed/precipitation is averaged by the number of times that they passed the same grid point (as in Figure 2c). After that, the seasonal average is computed. The box covers 20° by 20° around the cyclone, that is, a radius of approximately 10°, as illustrated in Figure 2a. We opt to use a wider box than previous studies (Gozzo et al., 2014; Reboita, da Rocha, et al., 2019) since they found that SCs usually have extreme winds approximately 500 km away from the cyclone center over the South Atlantic basin. Hence, by using a bigger box, we can also capture impacts of larger SCs. Using the Lagrangian approach, we analyze the mean accumulated precipitation and wind speed, extreme wind speeds (exceeding the 95th percentile), and the maximum wind speed inside the box at each 6 hr. For illustration, Figure 2a shows the extreme winds during the genesis of the SC Anita, which occurred in March 2010 (Dias Pinto et al., 2013; Dutra et al., 2017), while Figures 2b and 2c present, respectively, accumulated and mean wind speeds for Anita’s lifecycle. For this event, the maximum accumulated velocity reaches ∼350 m s⁻¹, while the mean overcomes 17 m s⁻¹.

Figure 2. (a) Wind speed extremes at 10-m height (shaded, in m s−1) and the 20° × 20° box (red rectangle) around the cyclone’s center (orange dot) for the 0600 UTC 3 April 2010. (b) Accumulated and (c) averaged wind speed at 10-m height (m s−1) for the lifecycle of the subtropical cyclone Anita. Black dots in panels (b) and (c) indicate the locations of the cyclone center at each 6 hr.

For the Eulerian analysis, each date of the SC lifecycle is used to obtain the mean associated features for both warm and cold seasons. It is important to note that as the Eulerian approach considers a large domain around SCs for all time steps, other systems such as extratropical cyclones, fronts, mesoscale systems might be occurring concomitantly in the neighborhood. In addition, for each event, the cyclone center is in a different location. These conditions might influence the composite fields, masking hence the SC features. On the other hand, the Lagrangian frame preserves the SC features and indicates better the region affected by SCs, since the composites consider a smaller area around cyclone center and follow it at each time step. A weakness of the Lagrangian approach is the smaller number of events in a given area to obtain the composites, which may decrease the intensity of the anomalies.

In both Lagrangian and Eulerian approaches, composite fields are calculated (i.e., the average considering the dates of SC occurrence) and their differences to the seasonal climatology are referred to as anomalies.

Results Climatology of the Extreme Wind Speed at 10-m Height

Using the 95th percentile as a threshold for each reanalysis and season, the seasonal climatology (1979–2015) of the extreme wind speed at 10 m height and its difference in relation to the mean speed climatology are shown in Figure 3. The spatial patterns in both reanalyses are quite similar. Nevertheless, in most parts of the southwestern South Atlantic, CFSR has larger values of wind extremes than ERA5 (Figure 3).

Figure 3. (a, b) Climate Forecast System Reanalysis (CFSR; top) and (c, d) ERA5 (bottom) wind speed (m s−1) of the extreme events (i.e., wind speeds above the 95th percentile) for the (a, c) warm and the (b, d) cold seasons, and (e–h) the differences between the extremes and the mean wind speeds for the period 1979–2015. RG1 is indicated by the black rectangle.

As expected, wind speeds are more intense in high latitudes due to the transient systems occurring all over the year, with velocities ∼20 m s⁻¹ in CFSR and ∼18 m s⁻¹ in ERA5 for the cold months (Figures 3b and 3d). During the warm season, wind extremes weaken compared to the cold ones. Specifically in RG1, the extreme speed is in the range ∼12–17 m s⁻¹ considering CFSR and it decreases to ∼11–13 m s⁻¹ in ERA5 (Figures 3a and 3c). During the cold season, maximum wind speeds are stronger over southern RG1 decreasing toward north. Over southern Brazil and Uruguay, the extremes are intense and associated with the development of extratropical cyclones and fronts, and mesoscale systems (Durkee & Mote, 2010; Teixeira & Satyamurty, 2007). The difference between the extremes and climatological wind speed evidences stronger winds in CFSR compared with ERA5, mainly to the south 35°S and near the south-southeast coasts of Brazil (Figures 3e–3h). Moreover, Figure 3 presents the higher differences over the ocean (reaching 11 m s⁻¹) than over the continent (∼4 m s⁻¹).

From the annual cycle point of view, the averaged wind speed in RG1 (Figure 4a) shows the minimum wind speed in March (∼5.5 m s⁻¹) and the maximum in September (∼7 m s⁻¹). Again, it can be clearly noticed in Figure 4a that CFSR presents stronger winds compared to ERA5. The intensification of winds in September occurs since the South Atlantic subtropical anticyclone is located further south (Crespo et al., 2022) at the same time that there is a thermal low pressure system over the continent (Paraguay region), resulting in the intensification of the horizontal pressure gradients (Oliveira & Quaresma, 2018; Reboita, Ambrizzi, et al., 2019; Reboita, da Rocha, Ambrizzi, & Caetano, 2009). In addition, Satyamurty and De Mattos (1989) and Reboita, da Rocha, Ambrizzi, and Sugahara (2009) also showed that during the summer and spring (warm season), the southeast coast of Brazil is cyclonic and frontogenetically active favoring stronger winds, especially in the spring (Andrade, 2005).

Figure 4. Boxplots of the monthly (a) mean and (b) extreme wind speeds (m s−1) averaged in RG1 for the period of 1979–2015 in Climate Forecast System Reanalysis (CFSR; red) and ERA5 (gray).

While the mean wind speed in RG1 has a similar pattern and intensity in both reanalyses, the annual cycle differs for the wind extremes: in CFSR, the annual cycle has weak amplitude, with extremes median spanning from ∼9.3 to 10.2 m s⁻¹, with one peak in April and the other in October, while the minima occur in June and December; in ERA5, the median of extremes range from ∼8.0 to 9.2 m s⁻¹ with maximum in April and minima in August and February. Moreover, the boxplots in ERA5 in August and September are vertically larger, showing the higher interannual variability of extremes in these months. Overall, the cold and warm seasons have the median of wind speed extremes in RG1 differing in 1 m s⁻¹ between reanalyses: 9.6 m s⁻¹ in CFSR and 8.6 m s⁻¹ in ERA5. The difference in the median and extreme wind speeds in reanalyses in South Atlantic basin was already discussed by Cardoso (2019) and Crespo et al. (2022), who showed that compared with coastal buoys ERA5 has smaller biases in representation of the mean wind speed, while CFSR is better for the extreme winds. Moreover, other studies for different regions of the globe also reported this aspect (e.g., Çalışır et al., 2021; Thomas et al., 2021). Many aspects distinguish CFSR from ERA5, such as horizontal resolution (∼38 vs. 25 km), assimilation techniques and data, dynamic cores and physical parameterizations of the models. As discussed by Belmonte Rivas and Stoffelen (2019), besides the increase of 20% in the ERA5 performance compared to ERA-Interim, the systematic error patterns persist in the new reanalysis: excess in zonal winds (too westerly) and defective meridional winds (less poleward) in midlatitudes, which were attributed by the insufficient signal of the poleward transport and diffusion of anticyclonic momentum by mesoscale turbulence in the ECMWF model (Belmonte Rivas & Stoffelen, 2019).

Seasonal Climatology of Subtropical Cyclones

Only the annual climatology of SC cyclogenesis over South Atlantic basins has been already documented in previous studies (e.g., da Rocha et al., 2019; de Jesus et al., 2022; Gozzo et al., 2014). Here, we present briefly in Figure 5, the SC track density and the time series of the frequency for warm and cold seasons.

Figure 5. Seasonal track density of the subtropical cyclones in the (a) warm and (b) cold seasons for the period 1979–2015. The density considers the number of cyclones per area multiplied by 106. In the right corner of (a) and (b) are shown the annual mean and standard deviation of subtropical cyclone (SC) genesis and RG1 is indicated by the black rectangle. (c, d) The annual absolute number of subtropical cyclones in warm (c) and cold (d) seasons.

SCs are more frequent in summer and autumn accounting for, respectively, 46% and 28% of total annual (Figures 5a and 5b), while few events occur in winter (6%) and spring (20%). When SCs are grouped into extended warm and cold seasons they represent, respectively, 66% and 34% of total annual.

The spatial distribution of SCs (Figures 5a and 5b) shows a concentration of trackings for warm and cold season within RG1, which is the main region of SC development and characterizes the semistationary character of most SCs as already noted in the annual mean (de Jesus et al., 2022; Gozzo et al., 2014). In both extended seasons, the most frequent cyclone density is in a region inside RG1 near the coast.

The interannual variability of SCs does not present a clear pattern or trend for the warm or cold seasons (Figures 5c and 5d), although there are alternate periods with high and low frequency of SCs. For instance, years with more active cyclogenesis such as 1983 and 2008 with 8 SCs by year, and 10 SCs by year in 2001 are observed in the warm season (Figure 5c), but several years without SC activity occur in the cold season (Figure 5d), accompanied by fewer cyclones. In both seasons, it is noted an alternation in time with active and inactive periods of SCs in the 37 years of data. In addition, it can be a manifestation of some low-frequency variability modes (Southern Annular Mode, El-Niño Southern Oscillation, South Multidecadal Atlantic Dipole, among others) that needs to be investigated in a future work.

Impacts of Subtropical Cyclones Wind Speed, Precipitation, and Other Meteorological Variables

In this section, we present the composites and anomalies of wind speed, geopotential height, SLP, mass divergence, temperature advection, and moisture flux divergence considering the SC lifecycle at each 6 hr. The total of 169 and 87 SCs are used for the composites in warm and cold seasons, respectively. Since the anomalies are statistically significant at the 95th confidence level, they are not highlighted in the maps. Figure 6 presents for both ERA5 and CFSR the wind speed anomalies associated with SCs considering Eulerian and Lagrangian frames. Positive anomalies indicate stronger wind intensity in the SC environment than climatology.

Figure 6. Anomaly of the wind speed at 10-m height (shaded, m s−1) for days with subtropical cyclones (SCs) considering the (a) Eulerian and (b) Lagrangian references for Climate Forecast System Reanalysis (CFSR; top) and ERA5 (bottom) for extended warm and cold seasons (indicated in the panels). RG1 is indicated by the black rectangle.

In the Eulerian analysis (Figure 6a), the warm season is characterized by a dipole over the ocean, with stronger winds to the north of 33°S and weaker ones to the south of 35°S. The positive wind anomalies reach the coast of Espírito Santo, south Bahia states, advancing to the south offshore. In the cold season, the pattern of anomalies is characterized by a tripole, that is, weaker winds to the north of 19°S and to the south of 32°S and stronger ones in 25°–30°S. Moreover, in the cold season, the positive anomalies occur away from the coast and affect mainly the open ocean, while in the warm season, the wind intensification associated with SCs is close to the coast. Taking RG1 as reference, it is noticed that in both seasons, wind intensification occupies its eastern side, which is explained by the intensification of northeasterly/easterly winds in the center-east of SCs (Figure S2 in Supporting Information  S1). At same time, the anomalous anticyclonic circulation southward of SCs (centered in ∼45°S; Figure 7) induces an easterly circulation (∼35°S), which counterbalances with the background flow (region with zero anomalies; Figure 6a). On the other hand, the negative anomalies centered in ∼40°S (in the south of the domain; Figure 6) indicate that the mean wind speed of the SCs (Figure S2 in Supporting Information S1) is weaker than climatology (Figure S1 in Supporting Information S1). In general, the patterns of wind speed anomalies are similar in both reanalyses.

Figure 7. Anomalies of geopotential height (contours, intervals at each 10 m) at 500 hPa, horizontal winds at 10-m height (vectors, m s−1), and sea level pressure (shaded, hPa) for subtropical cyclones (SCs) considering the Eulerian reference in (a) Climate Forecast System Reanalysis (CFSR) and (b) ERA5 during the warm (left) and cold (right) seasons.

From the Lagrangian perspective (Figure 6b), the patterns of wind speed anomalies are somehow similar to the Eulerian ones (Figure 6a); however, the positive anomalies following SC tracking are enhanced and the negative ones are weaker. During the warm season, the wind speed positive anomalies are northwest-southeast oriented reflecting the preferential track density to the southeast (Figure 5a), while in cold season, they are more zonally elongated as also occur in the Eulerian approach (Figures 5b and 6a). In the Lagrangian approach, the higher negative anomalies to the south of 30°S are displaced to the west compared with that in the Eulerian analysis. In addition, for both Eulerian (Figure 6a) and Lagrangian (Figure 6b) approaches, the ERA5 presents weaker wind speed anomalies than CFSR in the warm season, while the opposite occurs in the cold season.

The positive anomalies of the wind speed to the east of RG1 shown in the Eulerian and Lagrangian approaches are in line with Gozzo et al. (2014), who found the maximum winds between 350 and 450 km away from the cyclone center. As most of the cyclones form inside RG1, this distance is coherent with the anomalies in Figure 6. In addition, the six SCs analyzed by Reboita, da Rocha, et al. (2019) also presented more intense winds occupying their east and southeast quadrant. Analyzing all types of cyclogenesis inside RG1, de Jesus et al. (2022) found stronger mean winds at 1,000 hPa predominately from northeast in a similar region of positive anomalies in Figure 6.

During the whole SC lifecycle, negative anomalies of SLP and geopotential height at 500 hPa are observed over the southeast coast of Brazil in both seasons, accompanied by a cyclonic circulation near the surface (Figure 7). This center remains anchored near the southeastern coast of Brazil in the next few days (figure not shown), confirming the semistationary feature of SCs in the region, as already noted by Gozzo et al. (2014). To the south of the low pressure area, there is a positive geopotential anomaly center associated with a strong high pressure system and the easterly/northeasterly anomalous winds are stronger to the southeast of the anomalous low pressure center (Figure 7). This dipole pattern in the pressure field is also apparent in other SC composites using coarse resolution reanalysis (Evans & Braun, 2012; Gozzo et al., 2014). It is interesting to note that while the positive geopotential anomaly at the 500 hPa tilt to west with height in relation to the anomalous high pressure at the surface, the anomalous low pressure has a barotropic vertical structure. In the cold season, the anomalous low pressure at the surface within RG1 is more east-west elongated and stronger than in the warm season (Figure 7). Furthermore, the barotropic vertical structure (negative anomalies of geopotential at 500 hPa and SLP) is more evident and the anomalies are stronger than in the warm months.

The anomalies for precipitation and further meteorological variables (upper-level mass divergence and winds, and low-level temperature advection and moisture flux divergence) are also calculated at every 6 hr during SC lifecycle in the Eulerian reference (Figures 8a and 9). The composites are shown only for ERA5 since CFSR has similar patterns.

Figure 8. Warm (left) and cold (right) (a) anomaly of precipitation in Eulerian frame (shaded, mm/day) and (b) accumulated precipitation (shaded, mm) Lagrangian reference from the Tropical Rainfall Measuring Mission data set for the period 1998–2015. RG1 is indicated by the black rectangle.

Figure 9. Anomalies of (a) temperature advection (shaded, K day−1) at 925 hPa, (b) moisture flux (vectors, g kg−1 m s−1) and its divergence (shaded, g kg−1 s−1) at 925 hPa, and (c) mass divergence (shaded, 10−5 s−1) and winds (vector, m s−1) at 200 hPa for subtropical cyclones (SCs) considering the Eulerian reference in ERA5 during the warm (left) and cold (right) seasons.

For both seasons, the SC activity results in enhanced precipitation in an elongated band oriented in the northwest-southeast direction, from the continent toward the South Atlantic, crossing the center-east sector of RG1. The positive precipitation anomalies are associated with the anomalous low pressure, low-level moisture flux convergence (Figure 9b) and warm advection (Figure 9a) occurring at the same time with an upper-level divergent flow (Figure 9c), contributing to upward motion. All these features are intensified in the cold season, which may explain the more intense precipitation anomaly in Eulerian (Figure 8a) or the accumulated precipitation in the Lagrangian reference (Figure 8b). The northerly/northeasterly flow from eastern side of the subtropical anticyclone, together with northwesterly flow from the continent, acts transporting moisture to RG1 and neighborhoods (low-level convergence, upper-level divergence, and warm advection acting to decrease the vertical static stability) favor the upward motions, contributing to organize the convection mainly in eastern-southern side of cyclones. At the same time, the suppression of precipitation to the south (Figure 8) is explained by the presence of anomalous high pressure inducing moisture flux divergence and cold advection at low levels, while at upper-level predominates air mass convergence (Figure 9). Over the south Brazil, this negative anomaly has similar intensity in warm and cold seasons; however, during the cold season, the cold advection and upper-level convergence reach eastern portions, which would explain the negative precipitation over the ocean in the Eulerian reference (Figure 8).

In the Lagrangian analysis, the composites of accumulated precipitation following the cyclones center at each 6 hr are shown in Figure 8b. The composite indicates more intense precipitation in a northwest-southeast band, occupying the center-east of RG1 that more or less coincides with the areas with positive precipitation anomalies in the Eulerian approach (Figure 8a). In the northeast and center-south of RG1, rainfall is more intense during the cold than warm season. The extended and intense northwest-southeast band of precipitation over the continent, especially during the cold season, matches the more intense cyclones that occur in autumn (analyzed separately) as also with the six SCs (three occurring in autumn) analyzed by Reboita, da Rocha, et al. (2019).

Extreme Winds

Figure 10 presents the anomalies of extreme winds for SC occurrence in the Eulerian reference. First, the wind extremes were defined as the velocities exceeding 95th percentile using winds at 6 hr in warm and cold seasons from 1979 to 2015. Second, the composites were obtained as the average of the extreme wind field for all SC lifecycle, that is, average of the velocities above the 95th percentile. After that, the composites for each season were subtracted from the 95th percentile fields. This provides a wider perspective of the extreme winds while SCs are occurring and what the regions are more affected by them. The Lagrangian analysis is not shown since the patterns are not conclusive.

Figure 10. Anomalies of extreme wind speed at 10-m height (shaded; m s−1) considering subtropical cyclone (SC) lifecycle in (a) Climate Forecast System Reanalysis (CFSR) and (b) ERA5 for warm (left) and cold (right) seasons. RG1 is indicated by the black rectangle. The stippled areas are significant at the 95th level.

In the warm season, positive anomalies of wind extremes occupy the northeast of RG1 and the coast near 25°S (Figure 10), which coincides with the region with the higher density of cyclone trackings (Figure 5). It is interesting to note that the positive anomaly of wind extreme within RG1 is closer to the coast than positive ones from the mean speed (Figure 6). The extreme speeds during SCs are weaker than climatology in some sectors of the South Atlantic basin as to the south of 30°S, in agreement with the weakening of the mean winds (Figure 6). For both reanalyses, the regions of negative and positive anomalies are similar, and statistically significant at the 95th confidence level.

In the cold season, positive speed extreme anomalies are located outside of RG1 (but they are not significant according to the t Student’s test; Figure 10), in accordance with the location further east of the systems during this season, as shown in track densities (Figure 5). Still eastern RG1, there is a region of positive anomalies, which coincides with the area of higher track density (∼20°–25°S and 30°–20°W) in both reanalyses. The region with the more intense extreme winds is on the coast, near 28°S. In general, in the cold season, the positive anomalies of extreme winds are more intense and frequent over the South Atlantic basin, reflecting the more intense and organized SCs in the austral autumn.

Figure 11 shows the annual mean of wind extremes considering the average in RG1 for both climatology and only SCs. In general, CFSR presents higher velocities than ERA5 (as noted previously). Moreover, in both reanalyses, the wind speed extremes are higher during the occurrence of SCs than climatology, confirming the spatial anomalies seen in Figure 10. The time series of extremes present similarities with the annual frequency of SCs, that is, years with less SCs result in weaker extreme winds (1986, 1988, and 2003 for example). However, there are exceptions when fewer systems correspond to high velocities, indicating the contribution of SCs for extreme winds over the RG1.

Figure 11. Climatology of extreme winds (m s−1, speeds above the 95% percentile, red lines) and extreme winds related to subtropical cyclones (SCs; m s−1, gray lines) averaged in RG1 in ERA5 (dashed lines) and Climate Forecast System Reanalysis (CFSR; solid lines).

When taking the maximum velocity in a 20° × 20° box following the cyclone center tracks at each 6 hr, the values are also more intense in CFSR than in ERA5. Figure 12 shows annual boxplots with these maxima in ERA5 and CFSR for all SCs. The maximum velocities following the SCs (Figure 12) are always stronger than in RG1 (Figure 11). However, as in RG1, most years with the less frequency of SCs coincide with the lower annual maximum winds, as for example, 1988 and 2003, but there are some exceptions, that is, a smaller frequency of systems resulting in higher speeds. The Lagrangian approach shows many outliers, where winds exceeding 20 m s⁻¹ are registered by both reanalyses. Moreover, in very rare events, the wind speeds are up to 30 m s⁻¹ in the South Atlantic basin. This occurred for instance in 1979, with only four SCs resulting in stronger speeds and one and two outliers (up to 30 m s⁻¹) in ERA5 and CFSR, respectively.

Figure 12. Boxplots of the annual maximum wind speed (m s−1) at 10 m height extracted from a 20° × 20° box following subtropical cyclone (SC) centers at every 6 hr for ERA5 (gray) and Climate Forecast System Reanalysis (CFSR; red).

The interannual variability of the maximum speed is also seen in the average, median, and lower and upper quartiles from CFSR and ERA5, with CFSR presenting higher values in most of these parameters than ERA5. Differently from the climatology in RG1 (Figure 4b), the maximum velocities following cyclones center are more spread in most of the years in CFSR, indicating higher interannual variability of the maxima. In addition, the interannual variability of maximum speed following cyclone centers is quite similar in both reanalyses (Figure 12), which does not occur for the annual cycle of extreme winds in RG1 (Figure 4b).

Conclusions

Given the importance and impacts of SCs near the southeastern coast of Brazil (RG1), this study investigated the relationship between SCs and near-surface winds and precipitation using two approaches (Lagrangian and Eulerian) over the South Atlantic basin. The analysis used the fine resolution CFSR and ERA5 reanalyses for the period 1979–2015 and two extended seasons are considered, the warm (from September to February) and cold (from March to August) ones. The tracking of all cyclones and classification of SCs are provided by Gozzo et al. (2017).

ERA5 and CFSR climatologies of wind extremes (velocities above the 95th percentile) show the speeds increasing from north to the south of the South Atlantic basin in both warm and cold seasons. Stronger winds in the south of the basin are associated with transient systems, such as cyclones and fronts occurring all the year. Specifically for RG1, the extreme winds are similar in both seasons, but in CFSR they are 1 m s⁻¹ greater than ERA5.

Between 1979 and 2015, a total of 169 and 87 SCs occurred, respectively, in warm and cold seasons, providing ∼7 systems by year, with high concentration of track density near the south-southeast coast of Brazil (inside RG1). The time series indicates that although without a clear pattern there is a great interannual variability in the frequency of SC, with a maximum of 16 events/year in 2001 and no-events in 2007.

The analysis of the meteorological variables associated with SCs considers the Lagrangian (following the cyclone center) and Eulerian (mean SC environment) reference frames. The Eulerian reference considers all time steps where SC exists—which means that the composite includes configurations from other meteorological systems—while the Lagrangian approach is more focused on the processes related to SCs or local, since it follows the center of the cyclone. In the Eulerian approach, in the warm season, the SC activity results in wind speed positive anomalies predominating eastward of RG1 and negative anomalies at higher latitudes (35°–40°S), resembling a dipole pattern. On the other hand, the cold season velocity anomalies resemble a tripole pattern, with negative anomalies to the north and south of the positive ones located eastward of RG1. The positive speed anomalies in the warm season reflect SC trend to stay semistationary near the southeast Brazil coast, while in the cold season, the positive wind anomalies point out for more oceanic systems. The Lagrangian approach provides the intensification of winds in a northwest-southeast band in the warm season, while it has a similar pattern of the Eulerian analysis during cold season. The similar wind speed anomalies close to the southeastern Brazilian coast in both references confirm that the wind pattern is predominated by SC processes. The intensification of the winds associated with SCs to the east of RG1 occurs in a region that is in line with previous studies (de Jesus et al., 2022; Gozzo et al., 2014; Reboita, da Rocha, et al., 2019).

In ERA5 and CFSR reanalyses, the composites considering the whole SC lifecycles (Eulerian approach) show a positive wind speed anomaly eastern RG1, which is explained by the intensification of northeasterly/easterly winds in the center-east of SCs. At same time, an anomalous anticyclonic circulation southward of SCs induces an easterly circulation (∼35°S), which counterbalances the background flow. On the other hand, the negative anomalies centered in ∼40°S (in the south of the domain) indicate that the mean wind speed of the SCs is weaker than climatology. For both seasons, the SC activity results in enhanced precipitation in an elongated band oriented in the northwest-southeast direction, from the continent toward the South Atlantic, crossing the center-east sector of RG1. The positive precipitation anomalies are associated with the anomalous low pressure, low-level moisture flux convergence and warm advection occurring at the same time with an upper-level divergent flow. All these features are intensified in the cold season, which may explain the more intense precipitation anomaly in Eulerian and in the Lagrangian references. Considering the Eulerian approach, the extreme winds associated with SCs present positive anomalies to the east-northeast of RG1 in both seasons and reanalyses. In this case, CFSR and ERA5 have differences in the wind extremes, which is clear in both the annual cycle and time series in RG1, where speeds from CFSR are always higher than ERA5.

Our study highlights the impacts of SCs on wind intensification and reinforcement of precipitation near the southeastern Brazilian coast, at the same time that both variables weaken southward. This occurs in both seasons (warm and cold) and reanalyses (CFSR and ERA5), but the wind anomalies are generally stronger in CFSR than ERA5. In addition, according to the Lagrangian approach, many outliers, with winds exceeding 20 m s⁻¹, are registered by CFSR reanalysis along SC lifecycle in the South Atlantic basin.

Therefore, the climatology presented clarifies the SC contribution to adverse weather conditions near and far from their center, affecting from near the coast to open sea. In that sense, it is also important to investigate these aspects related to SCs in future climate projections, since there is an indication of a negative trend in the frequency of SCs over the South Atlantic basin, accompanied by an intensification of these systems in future warming scenarios (de Jesus et al., 2022).

Acknowledgments

The authors would like to thank the Copernicus Climate Change Service, ECMWF, and NOAA that provided data for this study and also Luiz Felippe Gozzo for the subtropical cyclone track. This work was partially supported by Petrobras (2017/00671-3), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq Grants 430314/2018-3 and 304949/2018-3).

Data Availability Statement

Data from CFSR (Saha et al., 2010, 2014), ERA5 (Hersbach et al., 2018), and TRMM (Huffman et al., 2007) were used in the creation of this manuscript. Figures were made with R (RasterVis, https://github.com/oscarperpinan/rastervis; Perpiñán & Hijmans, 2018) and Python (Matplotlib, https://matplotlib.org/; Caswell et al., 2021; Hunter, 2007).