1. Introduction
The southern African easterly jet (AEJ-S) is the Southern Hemisphere counterpart of the better-known northern African easterly jet (AEJ-N), and it is found over the central African subcontinent and the southeast Atlantic region (Fig. 1). Similar to the AEJ-N, the AEJ-S is an important part of the large-scale dynamical system over central Africa and the southeast Atlantic region, with impacts on the regional climate through its influence on the convective system and aerosol distribution (e.g., Nicholson and Grist 2003; Adebiyi and Zuidema 2016; Creese and Washington 2018; Tamoffo et al. 2019). Specifically, the central African region produces some of the most intense precipitation in the tropics, resulting in a strong convective latent heating and the highest frequency of lightning strikes in the world (Zipser et al. 2006; Toracinta and Zipser 2001; Washington et al. 2013). While central Africa is characterized by two main rainfall seasons, austral autumn (March–May) and spring (September–November), the stronger and wetter rainfall season (austral spring) is substantially influenced by the presence of the AEJ-S (Jackson et al. 2009; Kuete et al. 2020). In addition, the marine environment over the southeast Atlantic region has the most extensive biomass-burning aerosols overlying one of the largest low-level clouds in the world (Wood 2012; Klein and Hartmann 1993; Zhang et al. 2016). This combination makes the southeast Atlantic an important region for studies involving aerosol–cloud interactions (Stier et al. 2013), whereby the AEJ-S is a central dynamical feature for effective offshore aerosol transport (Adebiyi and Zuidema 2016). Despite the importance of the AEJ-S to the regional climate, it remains unclear whether climate models accurately represent the strength, location, and associated properties of the AEJ-S over the region.
Representation of the southern African easterly jet (AEJ-S) as the central dynamical feature over central Africa and the southeast Atlantic Ocean. In addition, the image shows the northern African easterly jet (AEJ-N), which is the Northern Hemisphere counterpart of AEJ-S. The image also shows the precipitation (white contours) taken from TRMM, aerosol optical depth (red to yellow shades, indicating the smoke aerosols), and the offshore low-level cloud (gray to white shades) fraction from MODIS. See section 2a for description of TRMM and MODIS datasets.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Previous studies have documented the AEJ-S as a midtropospheric easterly wind with speed greater than 6 m s−1 and located around approximately 600 hPa between September and November (e.g., Adebiyi and Zuidema 2016; Nicholson and Grist 2003; Kuete et al. 2020). These studies also indicated that the AEJ-S is a thermal wind channel between 5° and 15°S because of a strong meridional temperature gradient between two convective regimes of differing temperature-moisture characteristics. To the south of the AEJ-S is hot dry convection associated with the Kalahari/Angola heat low over the Namib–Kalahari desert (Fig. 1), with similar characteristics to the Sahara heat low (Lavaysse et al. 2009; Howard and Washington 2018). To the north of the AEJ-S is a cool moist region associated with the mesoscale convective system over the Congo-Zaire basin (Jackson et al. 2009). While both of these convective regions are responsible for the generation of the AEJ-S (Adebiyi and Zuidema 2016), a recent study suggested that the midlevel anticyclonic high pressure system associated with the Kalahari/Angola heat low is primarily responsible for the maintenance of the AEJ-S (Kuete et al. 2020). Unlike the Northern Hemisphere counterpart, the cool moist region over the central African continent that contributes to the generation of the AEJ-S is small (typically spanning about 15–20° wide), with the AEJ-S fading off as it exits the subcontinent due to unfavorable dynamical conditions over the southeast Atlantic Ocean (Nicholson and Grist 2003; Adebiyi and Zuidema 2016). As a result, the climatological distribution of the AEJ-S over the land and ocean collectively spans, on average, only about 40° in longitude, that is, between ∼5°W and 35°E during the September–November period (Adebiyi and Zuidema 2016). In addition, Kuete et al. (2020) also highlighted a link between the AEJ-S and the Southern Ocean westerly wave disturbance, whereby the wave could modulate the strength of the jet through its influence on the cross-latitudinal temperature gradient. Therefore, to accurately simulate the AEJ-S in the climate models, it is necessary to understand the nearby subtropical and midlatitude meteorological forcings, such as the surface meridional temperature gradient, the midlevel anticyclonic high pressure system, and the midlatitude westerly wave disturbances.
Over central Africa, where the AEJ-S occurs, one major challenge to understanding the large-scale dynamical features and the associated influence on the regional climate is the dearth of available ground-based observations (e.g., Washington et al. 2013). Fewer than three meteorological stations are operational over central Africa, and only about 10 rain gauges are available over the region (Nicholson et al. 1988; Washington et al. 2013). Consequently, observational constraints and validations of simulated large-scale dynamical processes over the region have been limited (Washington et al. 2013), leading to significant intermodel spread in simulated dynamical features (Creese and Washington 2018; Munday and Washington 2018; Creese and Washington 2016; Aloysius et al. 2016). For example, the global models in phase 5 of the Coupled Model Intercomparison Project (CMIP5) capture the annual rainfall cycle over the Congo–Zaire convective region (Creese and Washington 2018), but the spatial distribution varies significantly to the extent that Creese and Washington (2016) noted that the ensemble mean of the models is unlikely to represent the climatological state of rainfall over the region. To fill this gap, previous studies have used reanalysis datasets as proxy datasets (Parker 2016), in addition to satellite-based observations, to evaluate the large-scale dynamical systems over the region (e.g., Farnsworth et al. 2011). While satellite-based observation can be used to assess uncertainties in rainfall variabilities (e.g., Tompkins and Adebiyi 2012), reanalysis datasets are better suited to understand the processes underlying the large-scale circulation (Nicholson and Grist 2003; Adebiyi and Zuidema 2016). Despite their wide use, satellite-based products and reanalysis datasets also have their associated uncertainties (Beighley et al. 2011; Hua et al. 2019). Nevertheless, they remain useful to better understand the AEJ-S and its associated large-scale features over central Africa.
In this study, we document the characteristics of the AEJ-S and assess its representation in the CMIP6 models. While most previous studies over the central African region often focus on assessing the simulated rainfall variability, they largely ignore the associated AEJ-S (e.g., Balas et al. 2007; Farnsworth et al. 2011). Similarly, most previous modeling studies focus on the radiative impacts of smoke aerosol over the southeast Atlantic and often ignore the dynamical mechanism responsible for the aerosol transport over the region (e.g., Sakaeda et al. 2011; Mallet et al. 2019; Chang et al. 2023). Here, we use observation and reanalysis datasets to assess whether CMIP6 models capture the climatological features of the AEJ-S and its implication for aerosol, clouds, and precipitation distributions over the region. We find that the misrepresentation of the AEJ-S in CMIP6 models results in substantial biases in the magnitude and distribution of simulated aerosol, clouds, and precipitation over central Africa and the southeast Atlantic Ocean. We conclude that accurate representation of the AEJ-S in climate models is important for processes that influence regional climate over central Africa and the southeast Atlantic Ocean (Fig. 1).
2. Data and methods
We use monthly-mean products between 1980 and 2014 (except when otherwise stated) to better understand the representation of AEJ-S and its associated characteristics over central Africa. We focus on the September–October period because it represents the period when the midtropospheric winds are maximum, biomass-burning (smoke) aerosols are readily emitted and transported offshore, and the precipitation over the central African region is maximum (e.g., Nicholson and Grist 2003; Adebiyi and Zuidema 2016, 2018; Creese and Washington 2018; Ryoo et al. 2021). Below, we describe the datasets used, including the observed meteorological dataset, the reanalysis dataset (section 2a), and the CMIP6 model outputs (section 2b). We also describe the procedure used to classify the CMIP6 models based on their representation of the AEJ-S and its associated features (section 2c).
a. Observation and reanalysis datasets
We used the Modern-Era Retrospective Analysis for Research and Applications version 2 (MERRA-2) datasets (Gelaro et al. 2017), which provides a long-term aerosol and atmospheric reanalyzed dataset to assess whether CMIP6 models capture the characteristics of the AEJ-S and its associated features. MERRA-2 uses the Goddard Earth Observing System version 5 (GEOS-5) atmospheric model and data assimilation system (Rienecker et al. 2008; Molod et al. 2015) and provides atmospheric aerosol and meteorological information on 0.625° × 0.5° latitude–longitude grids with 42 pressure levels. The MERRA-2 modeling system also uses the 3DVAR Gridpoint Statistical Interpolation meteorological analysis scheme (Wu et al. 2002). In addition, MERRA-2 assimilates meteorological and aerosol observations, including bias-corrected aerosol optical depth from the Moderate Resolution Imaging Spectroradiometer (MODIS), Advanced Very High Resolution Radiometer (AVHRR), Multiangle Imaging Spectroradiometer (MISR), and ground-based AERONET (Randles et al. 2016, 2017) and simulates the aerosol fields with a radiatively coupled version of Goddard Global Ozone Chemistry Aerosol Radiation and Transport (GOCART) model (Colarco et al. 2010).
Previous studies that evaluated MERRA-2 products find that it compares well against independent observations (e.g., Buchard et al. 2017). Specifically, over central Africa and southeast Atlantic, several studies have used MERRA-2 for analysis involving meteorology and aerosols and have found it to compare well with observations (e.g., Adebiyi et al. 2015; Adebiyi and Zuidema 2018; Ryoo et al. 2021; Pistone et al. 2021). In addition, it is useful to note that because MERRA-2 assimilates aerosol-informed radiances from several satellite platforms, the resulting dynamical features, such as the representation of the AEJ-S, produced by MERRA-2 are expected to incorporate the radiative effects of aerosols over the region (e.g., Chaboureau et al. 2022). This contrasts with the NCEP reanalysis, which does not incorporate aerosol information. Because of the above reasons, we use only MERRA-2, although we expect other reanalysis datasets, such as ERA5, to result in similar conclusions (e.g., Adebiyi et al. 2015).
In addition to meteorology and aerosol products, we also use MERRA-2 precipitation estimates. MERRA-2 provides two estimates of precipitation—one that is based on the GEOS-5 atmospheric data assimilation system and the other where the bias in the model-generated precipitation estimates is corrected based on both gauge and satellite remote sensing data (Reichle and Liu 2014). Specifically, observational-based products used in correcting MERRA-2 precipitation estimates include the CPC Unified Gauge-Based Analysis of Global Daily Precipitation (CPCU) product and the CPC Merged Analysis of Precipitation (CMAP) based precipitation product (Adler et al. 2003; Xie et al. 2007).
We collected the monthly-averaged aerosol, precipitation, and meteorological variables from MERRA-2 between 1980 and 2014 (GMAO 2015), including the three-dimensional zonal and meridional winds, vertical velocity, temperature, specific humidity, two-dimensional aerosol optical depth (AOD), cloud fraction, and precipitation rates.
Furthermore, we used radiosonde measurements of atmospheric profiles over central Africa to benchmark MERRA-2 and evaluate AEJ-S representation in CMIP6 models. We obtained the meteorological variables between 2003 and 2014 using the daily vertical radiosonde measurements from the Integrated Global Radiosonde Archive (IGRA) and retrieved from the University of Wyoming Upper Air Sounding Archive (http://weather.uwyo.edu/upperair/sounding.html; last access June 2020). The daily observation at a station could include up to four radiosonde measurements and usually consist of pressure, height, temperature, dewpoint temperature, relative humidity, mixing ratio, wind speed, and direction as a function of standard pressure levels. Due to limited meteorological stations over central Africa, only one station (station #64400; FCPP) at Pointe-Noire in the Republic of Congo falls close to the center of the AEJ-S (see Fig. 2). While some other meteorological stations exist in the region, none falls close to the main region of the AEJ-S, although they generally capture the broad midtropospheric easterly winds of the region (see Fig. S1 in the online supplemental material). Since the focus of this paper is on the AEJ-S, we used the measurements at Pointe-Noire, Congo, #64400 FCPP (4.82°S, 11.90°E) to benchmark the wind estimates in the reanalysis dataset and assess the representation of the AEJ-S in the CMIP6 models (Fig. 2). We averaged the daily measurements to obtain monthly values of wind speed, direction, and temperature. Furthermore, we applied a quality control measure requiring the monthly values to have measurements of temperature and wind vertical profiles for a minimum of 5 days. Within the range allowed by the observation, the annual cycle of winds is largely insensitive to the threshold for the minimum number of days. In addition, because of inconsistencies in the reported vertical levels, we required the difference in height between two successive pressure levels in each daily measurement to be less than 5 km. For all cases that pass the above criteria, we performed logarithm interpolation to obtain uniform vertical profiles.
Distribution of the southern African easterly jet (AEJ-S). (a) The monthly variations of the zonal winds obtained for MERRA-2 (red line and pink shades) and collocated with observation (black line and gray shades) at the nearby meteorological station (#64400; FCPP) at Pointe-Noire (4.82°S, 11.90°E) in the Republic of Congo at the altitude of 600 ± 50 hPa (to accommodate for the poor vertical resolution of the soundings). MERRA-2 is compared only for months and years that satisfy the quality control criteria requiring that each measurement has the pressure level, temperature, and wind vertical information between 600 ± 50 hPa for a minimum of 5 days in each month (see text in section 2 and Table S1). The shades are the mean standard errors. (b) September–October (1980–2014) mean 600-hPa zonal winds (m s−1; shading) highlighting the AEJ-S for MERRA-2 (zonal wind speed stronger than −6 m s−1; red contours). Blue boxes delineate 0°–12°E (at left) and 12°–24°E (at right) between 2° and 14°S.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
In addition to using the radiosonde measurements to benchmark the reanalysis estimate of atmospheric winds, we used satellite-based observation of AOD, clouds, and precipitation to validate the representation of aerosol and precipitation in the reanalysis dataset. Specifically, we obtained the 550-nm-wavelength aerosol optical depth and total cloud cover from MODIS onboard the Terra satellite (Remer et al. 2005) and the precipitation rate product from Tropical Rainfall Measuring Mission (TRMM) Multisatellite Precipitation Analysis 3B42 (Huffman et al. 2007). For MODIS, we use the gridded monthly AOD and total cloud cover that are part of collection 6, MOD08, Level-3 products, with a spatial resolution of 1° × 1° and available since 2000 (Levy et al. 2013; Platnick et al. 2015). For TRMM, we use the monthly-averaged precipitation rate that is estimated from the 3-hourly 0.25° × 0.25°. Because the MODIS and TRMM datasets do not extend back to 1980, we used the September and October datasets between 2000 and 2014 since we are only interested in the climatological distribution of AOD, clouds, and precipitation rates. Preliminary analysis that used MERRA-2 suggests that the spatial distributions of AOD and precipitation rates are largely similar between the climatological periods of 2000–14 and 1980–2014.
b. CMIP6 model
Using the radiosonde, MODIS, TRMM, and MERRA-2, we evaluated the representation of AEJ-S and its associated features in 14 CMIP6 models. The names, institutions, and spatial resolution of these Earth system models are shown in Table 1. The criteria for selecting these models are based on the complete availability of needed variables needed for the analysis. Specifically, we selected models that deliver complete outputs for all meteorological variables (e.g., temperature, winds, humidity) as well as outputs for aerosol, clouds, and precipitation fields. In addition, we also require that these models have their first realization (r1i1p1f1) for historical forcing experiments, which were generated based on historical concentrations of CO2 and other long-lived greenhouse gases, land use, natural and anthropogenic aerosols, volcanoes, and solar forcing (Eyring et al. 2016). Because not all CMIP6 models give outputs that satisfy these criteria, it limits the available number of models that can be part of the analysis to only 14.
List of CMIP6 models used in this study.
We obtained CMIP6 monthly meteorological variables, including the three-dimensional (3D) zonal, meridional, and vertical winds, temperature, and specific humidity, as well as monthly total cloud cover, precipitation rate, and aerosol optical depth. While meteorological fields in CMIP6 models largely depend on individual model physics, aerosol fields depend on the treatment of aerosol emission processes, chemical production, transport, and deposition processes (e.g., Collins et al. 2017; Hoesly et al. 2018). For example, the spread in aerosol fields would be dominated by the differences in emission treatment over the central African region (see section S1 in the online supplemental material), whereas the distribution of aerosol fields farther downstream may be influenced by meteorological parameters, including the midtropospheric winds over the southeast Atlantic Ocean. As a measure of the total aerosol concentration, we focus on aerosol optical depth as the bulk aerosol parameter of interest because its distribution, particularly farther downstream, can be directly linked to the transport processes influenced by the winds and AEJ-S. In contrast, the relationships between the AEJ-S and other aerosol properties, such as single-scattering albedo, may be complicated since their distribution can be dominated by other processes, including aging in the atmosphere, which is beyond the scope of this study. In addition, we do not consider the analysis of aerosol vertical distribution because not all CMIP6 models deliver 3D aerosol variables, such as the mass mixing ratios. For example, among the selected CMIP6 models (see Table 1), only three (BCC-ESM1, GFDL-ESM4, and MRI-ESM2) have an available mass mixing ratio for black carbon, which is a good proxy for the total biomass-burning aerosols over the region.
Because of the differences in horizontal resolution, we regrid all model outputs to the common resolution of 2.8° × 2.8° with a distance-weighted technique that used the four nearest neighbor values of fields between grids (e.g., Jones 1998). Consequently, we calculated the model ensemble mean from the regridded models as the average of all 14 models.
c. Classification of CMIP6 models
Because of the inherent diversity among CMIP6 models, we expect the resulting model output to simulate varying meteorology over central Africa. Therefore, we had to define what determines a good representation of the AEJ-S and the associated features in the CMIP6 models. A good representation of the AEJ-S in CMIP6 models should capture the magnitude, altitude, and spatial distribution of the midtropospheric winds that characterize the observed AEJ-S over central Africa and the southeast Atlantic Ocean. Consequently, we classified models with a good representation of the AEJ-S as models with low bias relative to observation and reanalysis estimates of the AEJ-S over central Africa. Specifically, we defined low-bias CMIP6 models as models with mean absolute biases of midtropospheric easterly wind speed less than 50% of the maximum bias among other models (see section 5). In contrast, high-bias CMIP6 models were models with mean absolute biases of midtropospheric easterly wind speed higher than 50% of the maximum bias among other models. In addition, to capture the diversity among CMIP6 models, we assessed this classification separately at 600 hPa and at the AEJ-S centroid, which may differ from 600 hPa in some models. Although this classification is relative to the maximum bias among the CMIP6 models, the results and conclusion are the same compared to measurements at the meteorological station or MERRA-2 (see section 5). Therefore, of the 14 models used in this analysis, three models—TaiESM, ACCESS-ESM, and IPSL-CM6A—are classified as low-bias CMIP6 models, with strong representation of AEJ-S, and three models—BCC-ESM, CanESM, and GFDL-ESM—are classified as high-bias CMIP6 models with weak representation of AEJ-S (see section 5).
3. The AEJ-S in MERRA-2
Our first objective is to validate the representation of the AEJ-S in MERRA-2 dataset. We do so by comparing the MERRA-2 monthly-mean midtropospheric zonal winds with spatiotemporally collocated radiosonde measurements from a nearby meteorological station (see section 2) close to the center of the AEJ-S (Fig. 2). Despite limited available measurements that pass our quality control criteria (see section 2a and Table S1), the seasonal variation of the midtropospheric zonal winds in MERRA-2 mostly agrees with the observation (Fig. 2a). Specifically, MERRA-2 agrees that the maximum easterly wind speed, representing the AEJ-S, occurs between September and October over the meteorological station. However, MERRA-2 underestimates the September mean zonal wind speed by 9% and overestimates the October wind speed by about 10%. Despite these discrepancies in the monthly means, the September–October average wind speed compares reasonably well with the observation, resulting in a bias of less than 1%. Because collocated MERRA-2 estimates of midtropospheric zonal winds compare well with observations over the region and because of other reasons highlighted in section 2a, therefore, we used the MERRA-2 dataset to assess the properties of the AEJ-S.
MERRA-2 indicates that the AEJ-S is centered on the western part of the central African subcontinent and extends westward over the southeast Atlantic Ocean. Therefore, as shown below, the representation of the AEJ-S in MERRA-2 follows the climatological representation shown in previous studies (e.g., Adebiyi et al. 2015; Adebiyi and Zuidema 2016; Kuete et al. 2020, 2023). For September–October easterly wind speed stronger than 6 m s−1, the AEJ-S is between ∼30°E and ∼0°, and from ∼14° to ∼2°S (Fig. 2b). This is also the case individually for September and October mean distribution (Fig. S2).
As a function of altitude, the AEJ-S is approximately between 550 and 720 hPa, with the maximum easterly winds located at ∼635 hPa over land and ∼660 hPa over the ocean (black contours in Figs. 3a,b). Although the AEJ-S is centered at a higher altitude, strong easterly winds generally occur above ∼750-hPa altitude, north of ∼12°S over central Africa and the southeast Atlantic region (blue shades in Figs. 3a,b). In contrast to the midtropospheric easterly wind, the westerly wind undercut the AEJ-S in the boundary layer (below ∼750 hPa) both over the ocean and land (red shades in Figs. 3a,b). This westerly wind brings deep moisture from the Atlantic Ocean over the Congo-Zaire basin, mostly north of 12°S (green shades in Figs. 3c,d). However, south of 12°S over land, the boundary layer is characterized by strong easterly winds that are largely disconnected from the westerly wind north of it, and the easterly winds transport dry continental air over the ocean (cf. Figs. 3a,b and Figs. 3c,d). Similarly, the circulations associated with either the boundary layer easterlies or westerlies are largely disconnected from the circulation in the midtroposphere (e.g., Garstang et al. 1996). In the midtroposphere, the AEJ-S bounds the northern part of an anticyclonic circulation with returning westerly winds south of 20°S (Figs. 2b and 3a,b). Furthermore, this anticyclonic circulation is a result of a midtropospheric high pressure system that, in turn, modulates the broader large-scale dynamical system (e.g., Driver and Reason 2017; Kuete et al. 2020).
Vertical cross section of the AEJ-S and associated characteristics. September–October (1980–2014) mean pressure–latitude cross section of (left) zonal winds (U; m s−1), (center) specific humidity (Qυ; g kg−1) and potential temperature (θ; K; red contours), and (right) the meridional virtual potential temperature gradient (
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Consequently, the strength of the AEJ-S is maintained by this midtropospheric high pressure system and the meridional temperature–moisture gradients over central Africa (Figs. 3c–f). The high pressure system creates a divergent center in the midtroposphere that facilitates the Kalahari/Angola heat low, which is characterized by enhanced surface heating and strong dry convection (Adebiyi and Zuidema 2016; Kuete et al. 2020; Howard and Washington 2018). This hot dry convection is evident above ∼900 hPa and south of ∼10°S over land in Fig. 3c (red contours). With the influx of moisture from the Atlantic Ocean and the convection over the Congo–Zaire basin north of 10°S, a strong meridional temperature–moisture gradient is developed over land. As a result, the AEJ-S maximizes at the top of the negative meridional virtual potential temperature (VPT) gradient (e.g., Nicholson and Grist 2003), which extends from the lower troposphere to the midtroposphere (blue shades in Fig. 3e). While temperature gradients have a substantial influence on the magnitude of the AEJ-S compared to the moisture gradient (e.g., Cook 1999), the moisture gradient is important to localize the meridional VPT gradient, particularly over land (e.g., Adebiyi and Zuidema 2016). Furthermore, Fig. 3f shows that the meridional VPT gradient still exists between 0° and 12°E, sustaining the September–October AEJ-S beyond the continent where it is developed. However, while the meridional temperature–moisture gradient facilitates the development of the AEJ-S, it cannot be maintained climatologically over the ocean beyond ∼0°W when averaged between September and October (Fig. 2b). Although similar to the AEJ-N, an individual AEJ-S could propagate beyond the 0°W boundary. Overall, MERRA-2 captures well the magnitude, structure, and large-scale mechanism associated with the AEJ-S over central Africa and the southeast Atlantic.
4. The AEJ-S in CMIP6 models
We assess the September–October averages of midtropospheric winds in CMIP6 models and find a significant model spread in the strength and position of AEJ-S over central Africa and the southeast Atlantic region. Of the 14 CMIP6 models considered (see Table 1), six models show substantially weak 600-hPa easterly winds and AEJ-S (Figs. 4a–f), and the remaining eight models have moderate to strong 600-hPa easterly winds and AEJ-S (Figs. 4g–n). The models with weak AEJ-S have spatial extents that are largely limited to east of 8°E, strengthening only when the wind is close to the coastline (Figs. 4a–f). In contrast, models with moderate to strong AEJ-S capture the direction and broad region of the easterly winds and the spatial extent of the AEJ-S when compared to MERRA-2 (cf. Figs. 2b and 4g–n). Specifically, for models with strong AEJ-S, the spatial distribution of the AEJ-S extends farther west of 0°E. Together, the ensemble mean of CMIP6 models shows comparable spatial distribution with MERRA-2 but with bias in the strength and position of the AEJ-S (Fig. 4o)
The AEJ-S in CMIP6 models. September–October (1980–2014) mean CMIP6 easterly winds at 600 hPa, showing the AEJ-S as winds stronger than 6 m s−1 (red contours).
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
In addition to the spread in the 600-hPa midtropospheric winds, there is a general model spread in the vertical distribution of the easterly and westerly winds associated with the AEJ-S. To show this spread, we average separately over the land between 12° and 24°E (Fig. 5) and the ocean between 0° and 12°E (Fig. S3). Consistent among the models is the existence of easterly winds in the midtroposphere extending into the boundary layer (BL), generally south of ∼13°S, and the BL westerly wind north of it, undercutting the midtropospheric AEJ-S (Fig. 5 and Fig. S3). However, major differences among the models are reflected in the altitude and magnitude of the AEJ-S, its associated midtropospheric easterly winds, and the altitude and magnitude of the BL westerly winds. Specifically, the altitude of the BL westerly winds varies approximately between 854 and ∼773 hPa (mean of 839 hPa) over land (Fig. 5) and between ∼914 and ∼812 hPa (mean of 903 hPa) over the ocean (Fig. S4). Similarly, across the CMIP6 models, the center of the AEJ-S varies between ∼663 and ∼562 hPa and from 671 to 603 hPa over land and ocean, respectively, descending by an average of ∼15 hPa from land (643 hPa) to ocean (658 hPa; Fig. S4). As with the spread in altitude and longitudinal extent of the AEJ-S, there are also variabilities in the latitudinal extent of the easterly and westerly winds. Specifically, the southern edge of the boundary layer westerly winds varies between ∼10° and 17°S over land. For the midtropospheric easterly winds, it varies between ∼20° and ∼17°S over land and between ∼17°S and ∼15°S over the ocean (Fig. S4).
The AEJ-S vertical distribution in CMIP6 models. September–October (1980–2014) mean meridional cross section of CMIP6 zonal winds (m s−1) averaged over the land component of the AEJ-S (between 12° and 24°S) for individual models (Figs. 5a–n) and the model ensemble (Fig. 5o). The black contours show the AEJ-S as the easterly winds greater than 6 m s−1. Blue shades are the easterly winds, and red shades are the westerly winds. The corresponding figure averaged over the ocean part of the AEJ-S is shown in Fig. S3.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
The vertical distribution of easterly and westerly winds is largely associated with the vertical distribution of temperature and moisture over the region. We compare the vertical distribution of specific humidity (green and purple shades) and the meridional virtual potential temperature (VPT) gradient among the CMIP6 models, averaged over the land segment of the AEJ-S (Figs. 6 and 7; the ocean components are shown in Figs. S5 and S6). We find that most CMIP6 models simulate the overall moisture–temperature general structure, but there are substantial spreads in the magnitude of the average temperature and moisture over the region. For example, regions of high and low moisture or temperature values are broadly consistent among the CMIP6 models. That is, the deep-layer moisture regions with specific humidity larger than 8 g kg−1 are largely consistently north of ∼15°S and are associated with the region of westerly winds. Similarly, the drier regions in the boundary layer are consistently south of ∼15°S and are associated with the easterly winds that bring continental air offshore (cf. Figs. 5 and 6). In addition, the southern region shows consistently high potential temperature values associated with the Kalahari/Angola heat low (Munday and Washington 2017). This is evident, for example, in the 310-K potential temperature value that extends to the surface over the Angola and Namibia region (red contours in Fig. 6 and Fig. S5). The distribution of these temperature and moisture profiles also produces negative meridional VPT gradients that occur between ∼20°S and the equator over land (see blue shades in Fig. 7).
Associated characteristics of AEJ-S in the CMIP6 models. September–October (1980–2014) mean meridional cross section of CMIP6 specific humidity (g kg−1; green and purple shades) and potential temperature (K; red contours) averaged over the land component of the AEJ-S (between 12° and 24°S) for individual models (Figs. 6a–n) and the model ensemble (Fig. 6o). The black contours show the AEJ-S as the easterly winds greater than 6 m s−1 and are highlighted in Fig. 5. The corresponding figure averaged over the ocean part of the AEJ-S is shown in Fig. S5.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Associated characteristics of AEJ-S in the CMIP6 models. September–October (1980–2014) mean meridional cross section of CMIP6 meridional virtual potential temperature gradient (K deg−1; red and blue shades) averaged over the land component of the AEJ-S (between 12° and 24°S) for individual models (Figs. 7a–n) and the model ensemble (Fig. 7o). The black contours show the AEJ-S as the easterly winds greater than 6 m s−1 and are highlighted in Fig. 5. The corresponding figure averaged over the ocean part of the AEJ-S is shown in Fig. S6.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Despite the broad similarity in the moisture and temperature structures among the CMIP6 models, there are significant spreads in the magnitude and vertical extent of moisture and temperature features connected to the generation and maintenance of the AEJ-S. For example, at 850 hPa, the magnitude of specific humidity averaged over the Congo-Zaire region (2°–14°S, 12°–24°E) varies between 10 and 12 g kg−1, and the magnitude of the potential temperature averaged over the Kalahari/Angola heat low region (12°–20°S, 12°–24°E) varies between ∼310 and 312 K among the CMIP6 models. These spread in moisture and temperature correlate with the spread in meridional VPT gradients, where it ranges between −0.19 and 0.3 K deg−1. In addition, we find that, in CMIP6 models, the altitude where negative meridional VPT gradients equal zero largely correlates with the altitude of AEJ-S centroid, consistent with previous studies (Fig. 8) (Adebiyi and Zuidema 2016; Kuete et al. 2023). However, the magnitude of the negative meridional VPT gradients below the AEJ-S in CMIP6 models does not correlate with the simulated magnitude at the AEJ-S centroid (Fig. 8). This suggests that other dynamical forcings likely influenced the strength of the AEJ-S in CMIP6 models (e.g., Kuete et al. 2023).
Associated characteristics of AEJ-S in the CMIP6 models. (top) The relationship between the altitude or pressure level (hPa) of the AEJ-S centroid and the altitude or pressure level (hPa) where the average meridional virtual potential temperature gradient goes to zero directly below the AEJ-S for individual CMIP6 model (colored marker), model ensemble (black solid marker), and MERRA-2 (red solid marker) for land (12°–24°E), ocean (0°–12°E), and both land and ocean (0°–24°E) component of AEJ-S. (bottom) The relationship between the average meridional virtual potential temperature gradient below the AEJ-S and the zonal wind speed at the AEJ-S centroid. Note that the linear regression (black line) and Pearson’s correlation value (R) were calculated only for the individual models and not with the model ensemble and the MERRA-2 values.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
5. Evaluating the AEJ-S in CMIP6 models
In addition to the spread among the CMIP6 models, the AEJ-S and its associated characteristics also differ from those obtained from station-based observation or MERRA-2. Specifically, the ensemble of CMIP6 models generally underestimates the AEJ-S strength (Fig. 9). We compare the 600-hPa zonal wind collected at the meteorological station in Congo with collocated zonal winds from MERRA-2 and individual CMIP6 models (Fig. 9a). Whereas MERRA-2 reproduces the observation reasonably well (differing by only 0.06 m s−1), the ensemble of CMIP6 models substantially underestimates the observed easterly winds by 2.1 m s−1 or 25% over the meteorological station. Among the individual CMIP6 models, the bias ranges from 0.6 to 3.7 m s−1. Since the MERRA-2 estimate reproduces the station observation well, we use it to obtain a broader assessment of AEJ-S in the CMIP6 models by averaging over the land and ocean region of AEJ-S (see Fig. 2b for box region). Over both land and ocean, the ensemble of CMIP6 models underestimates the wind speed of AEJ-S when compared against MERRA-2. Specifically, the CMIP6 models underestimate the MERRA-2 easterly winds at 600 hPa by about 0.84 m s−1 over land (cyan bars in Fig. 9b), by about 0.99 m s−1 over the ocean (pink bars), and by about 0.95 m s−1 collectively over both (orange bars).
The bias in the CMIP6-simulated AEJ-S. (a) Comparison of zonal winds between observations (black bar), MERRA-2 (red bar), and the CMIP6 models (gray bars) at a meteorological station (#64400; FCPP) in the Republic of Congo at the altitude of 600 hPa. The comparison is made for the months and years that the observations are available (see section 2a and Table S1). Relative to MERRA-2 averaged for September and October (1980–2014), we calculated (b) the bias in averaged CMIP6-simulated easterly winds at a uniform altitude of 600 hPa, (c) the altitude bias in the estimated AEJ-S centroid, which varies as a function of model (see Fig. S4), and the bias in averaged CMIP6-simulated easterly winds at the AEJ-S centroid. These are evaluated over land (cyan; averaged between 12° and 24°E), ocean (pink; averaged between 0° and 12°E), and both over land and ocean (orange; averaged between 0° and 24°E). It is worth noting that positive (negative) values in (c) mean that the CMIP6 AEJ-S centroid is lower (higher) than MERRA-2. Similarly, the positive (negative) values in (b) and (d) mean CMIP6 models mean that easterly winds underestimate (overestimate) the estimate from MERRA-2.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Although assessing the easterly wind at 600 hPa is a good indicator that models underestimate AEJ-S, it does not account for altitude variation in CMIP6-simulated easterly winds. To account for the variability of AEJ-S altitude, we estimate the altitude of maximum easterly winds in the midtroposphere (AEJ-S centroid) and estimate the bias in CMIP6-simulated AEJ-S wind speed at this altitude (see Figs. 9c,d). We find that the altitude of the AEJ-S centroid in the ensemble mean of CMIP6 models is slightly lower (by 8 hPa) over land and slightly higher (2.8 hPa) over the ocean when compared against MERRA-2 (Fig. 9c). Although the altitude of the AEJ-S centroid is slightly higher over the ocean, more than half of the CMIP6 models individually simulate them to be lower in altitude than in MERRA-2 (Fig. 8). The overall height of the AEJ-S over both land and ocean for the ensemble of CMIP6 models is lower than MERRA-2, despite the zero altitude of the meridional VPT gradient being about the same (see red and black solid markers in Fig. 8). Further analysis indicates that, at the core of the AEJ-S, the CMIP6 models underestimate MERRA-2 by an average of 0.3 m s−1, which range from −1.3 to +1.2 m s−1 (Fig. 9d). This range in the easterly wind bias is less than when the comparison is considered at 600 hPa without accounting for the differences in the vertical location of the AEJ-S centroid in each model (cf. Figs. 9b,d).
Although the bias in the representation of AEJ-S and its associated characteristics exist across CMIP6 models, some models perform better than others, therefore representing AEJ-S better than other models. We classify the models into those with a low bias and others with a high bias to further understand the impacts of AEJ-S representation on the simulated aerosol, clouds, and precipitation distribution (see section 2c for details). Specifically, three models – TaiESM, ACCESS-ESM, and IPSL-CM6A have strong AEJ-S and low absolute bias when compared against measurements at the meteorological station or MERRA-2 (Fig. 9a). These models have mean absolute biases that are less than 50% of the maximum bias among other models when assessed at 600 hPa or the AEJ-S centroid (see Figs. 9b,d). In contrast, three models—BCC-ESM, CanESM, and GFDL-ESM—have weak AEJ-S and high absolute bias, with mean absolute biases of more than 50% of the maximum bias among other models. Although the TaiESM model has slightly higher absolute bias than ACCESS-ESM and IPSL-CM6A models, we count it as a low-bias CMIP6 model because its absolute bias for midtropospheric easterly winds is still lower than other models, especially when considered at the AEJ-S centroid (Fig. 9d). In contrast, we exclude MRI-ESM2-0, E3SM, and CESM models from the low-bias models because they have absolute biases that are more than 50% of the mean when assessed at 600 hPa or the AEJ-S centroid. Last, the three high-bias models consistently have the largest biases among the CMIP6 models when assessed at 600 hPa or the AEJ-S centroid.
Overall, the composite average of the low-bias models better represents the AEJ-S and its associated characteristics relative to the composite average of the high-bias models (cf. Fig. 10 and Fig. S7 to Figs. 3a,c,e). Specifically, at the meteorological station, the average bias in the 600-hPa easterly winds is about 3 times more in high-bias models (underestimated by about ∼3.3 m s−1) than in low-bias models (underestimated by only ∼1.0 m s−1; see Fig. 9a). A similar difference in bias is also estimated at the AEJ-S centroid between the high-bias and low-bias models (cf. Figs. 10a,b with Fig. 5o). These discrepancies result in a stronger underestimation of wind speed in the midtroposphere for high-bias models than in low-bias models (Figs. 10a,b and 11a,b). In addition, we also find stronger hot dry convection over the Namib–Kalahari region for the low-bias models than for high-bias models (Figs. 10c,d and 11c,d), which results in a negative meridional VPT gradient with altitude and magnitude higher than the high-bias models (Figs. 10e,f and 11e,f). Specifically, the anomalous increase in the potential temperature over the Namib–Kalahari region is more localized in the southern flank of the AEJ-S in the composite of low-bias models with strong AEJ-S than high-bias models with weak AEJ-S (red contour in Figs. 11c,d). This localized anomalous heating, in addition to the strong meridional gradient induced by the companion anomalous cooling north of the AEJ-S (see blue contour in Fig. 11c), likely contributes to the stronger AEJ-S in the low-bias CMIP6 models. While these differences are significant over land, there are also similar results for the ocean component of the AEJ-S for the low-bias and the high-bias models (Figs. S7 and S8). Overall, the low-bias CMIP6 models better represent the AEJ-S and its associated characteristics than the high-bias models compared to MERRA-2.
Composite averages for CMIP6 models with low and high bias in AEJ-S. September–October (1980–2014) pressure–latitude cross section of (left) zonal winds (m s−1), (center) specific humidity (g kg−1) and potential temperature (K; red contours), and (right) the meridional virtual potential temperature gradient (K deg−1) for (top) low-bias models with strong AEJ-S (TaiESM, ACCESS-ESM, IPSL-CM6A) and (bottom) high-bias models with weak AEJ-S (BCC-ESM, CanESM, and GFDL-ESM), averaged over land (12°–24°E). The black contour highlights the southern African easterly jet (AEJ-S) with a zonal wind speed stronger than −6 m s−1. The corresponding figure averaged over the ocean part of the AEJ-S is shown in Fig. S7.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Composite bias in CMIP6 models when compared to MERRA-2. September–October (1980–2014) bias of (left) wind speed (m s−1), (center) specific humidity (g kg−1) and potential temperature (K; contours; red positive and blue negative), and (right) the meridional virtual potential temperature gradient (K deg−1 multiplied by 10) for (top) low-bias models with strong AEJ-S (TaiESM, ACCESS-ESM, IPSL-CM6A) and (bottom) high-bias models with weak AEJ-S (BCC-ESM, CanESM, and GFDL-ESM), averaged over land (12°–24°E). The black contour highlights the southern African easterly jet (AEJ-S) with a zonal wind speed stronger than −6 m s−1. The corresponding figure averaged over the ocean part of the AEJ-S is shown in Fig. S8.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
6. Influence of the AEJ-S on CMIP6-simulated clouds, precipitation, and aerosol distribution
Since the differences between the low-bias and high-bias models for AEJ-S and its associated characteristics capture the range of biases identified in CMIP6 models, we thus use this classification of CMIP6 models to explore how the characteristics of simulated AEJ-S may influence clouds, precipitation, and aerosol distribution over the region.
a. CMIP6 AEJ-S and cloud distribution
We examine the total cloud cover simulated by CMIP6 models, and we find a substantial relationship between the simulated AEJ-S and cloud distribution over central Africa and the southeast Atlantic regions. Although our analysis assessed the total cloud cover, which is a combination of low-, mid-, and high-level clouds, the cloud distributions over central Africa are generally dominated by mid-to-high-level clouds mostly associated with the continental deep convective systems (see yellow contour in Fig. 12b based on MODIS observation), and the cloud distributions over the southeast Atlantic Ocean are dominated by low-level subtropical clouds (see purple contour region in Fig. 12b based on MODIS observation). Most CMIP6 models struggle to simulate the magnitude and spatial distribution of these cloud systems resulting in large model spread over central Africa and the southeast Atlantic Ocean (Fig. S9). Specifically, the total cloud cover ranges between 58% and 82% when averaged over central Africa below the AEJ-S region (0°–10°S, 12°–24°E; Fig. S10a) and 60%–85% when averaged over the southeast Atlantic region (10°–20°S, 2°W–10°E; Fig. S10b), following Klein and Hartmann (1993). Collectively, the ensemble of CMIP6 models estimates the total cloud cover of about 70% over the defined central African region and about 69% over the defined southeast Atlantic region (Fig. S10 and black boxes in Fig. 12). These estimates from the ensemble of CMIP6 models generally underestimate the observational estimates from MODIS, which show total cloud cover of about 76% and 93% over the two defined regions, respectively (Figs. S10a,b). Although the ensemble of CMIP6 models underestimates the magnitude of the total cloud cover, it generally captures the spatial distribution of the region with dominant low-level clouds and the region with mid-to-high-level clouds.
The relationship between AEJ-S and simulated total cloud cover. (top) September–October climatology of total cloud cover (%; shaded) for the (a) ensemble of CMIP6 models (averaged between 1980 and 2014) and (b) MODIS observation (2000–14). (middle) Composite mean for (c) low-bias models with strong AEJ-S and (d) high-bias models with weak AEJ-S. (bottom) Bias in (e) the ensemble of models, (f) low-bias models, and (g) high-bias models. Black contours show the 600-hPa easterly winds stronger than 6 m s−1 representing AEJ-S; MERRA-2 AEJ-S is shown on (b). Also, purple and yellow contours in (b) show regions where low-level clouds and mid-/high-level clouds dominate based on MODIS observation. Black boxes in (a)–(d) delineate regions further accessed over central Africa (0°–10°S, 12°–24°E) and the southeast Atlantic region (10°–20°S, 2°W–10°E) shown in Fig. S10.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Despite the wide spread in individual models, we find that the composite average of low-bias CMIP6 models with strong AEJ-S better represents the total cloud cover, both over central Africa and the southeast Atlantic Ocean, than the high-bias CMIP6 models with weak AEJ-S (Figs. 12c,d). Over the southeast Atlantic, although both low-bias and high-bias models underestimate the MODIS observation, the high-bias models with weak AEJ-S underestimate the observation far worse (about 24% more) than the low-bias models with strong AEJ-S (Figs. 12f,g). In addition, the general spatial cloud distribution over the southeast Atlantic Ocean is better represented in the composite of low-bias models with strong AEJ-S than the composite of high-bias CMIP6 models with weak AEJ-S (Figs. 12c,d). Over the central African region, the low-bias models with strong AEJ-S slightly underestimate the total cloud cover over the continental region with dominant high-level clouds (Fig. 12f). In contrast, the high-bias models with weak AEJ-S broadly overestimate the total cloud cover over the same region when compared to MODIS observations (Fig. 12g). While these differences exist over the western part of central Africa, both low-bias and high-bias models overestimate the total cloud cover more over the eastern part of central Africa (red shades in Figs. 12f,g).
Because previous studies have postulated that AEJ-S can induce changes in the large-scale vertical velocity (Adebiyi and Zuidema 2016; Kuete et al. 2020), we examine the distribution of the large-scale vertical velocity at 500 hPa to understand further the relationship between AEJ-S and cloud distribution in CMIP6 models. Consistent with the cloud distribution of predominantly low-level clouds, CMIP6 models simulate large-scale subsidence over the southeast Atlantic Ocean with the largest magnitude south of ∼20°S (brown shades in Figs. 13 and 11). In contrast, regions with predominantly mid-to-high level clouds are associated with large-scale updraft over central Africa (dark green in Fig. 13 and Fig. S11). While all CMIP6 models simulate these general regions of upward and downward vertical motion, they do so with different magnitudes, resulting in substantial spread among the models (Fig. S11). Overall, the ensemble of CMIP6 models generally overestimates the large-scale subsidence south of ∼20°S over the southeast Atlantic Ocean (Fig. 13e). However, over central Africa, the model bias is more complex, with regions where the ensemble of CMIP6 models overestimates and underestimates the region of mean upward and downward motion when compared to MERRA-2.
The relationship between the AEJ-S and simulated vertical velocity at 500 hPa. (top) September–October climatology of vertical velocity (Pa min−1; shaded) for the (a) ensemble of CMIP6 models and (b) MERRA-2 (averaged between 1980 and 2014). (middle) Composite mean for (c) low-bias models with strong AEJ-S and (d) high-bias models with weak AEJ-S. (bottom) Bias in (e) the ensemble of models, (f) low-bias models and (g) high-bias models. Black contours show the 600-hPa easterly winds stronger than 6 m s−1 representing AEJ-S; MERRA-2 AEJ-S is shown in (b).
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Furthermore, we find that low-bias CMIP6 models with strong AEJ-S have smaller discrepancies with MERRA-2 than the high-bias CMIP6 model with weak AEJ-S (Figs. 13f,g). Specifically, over the central African region with mid-to-high level clouds, we find that the composite of high-bias models with weak AEJ-S overestimates the 500-hPa upward motion about 2 times more than the composite of low-bias models with strong AEJ-S. This enhanced upward motion contributes to the overestimation of the total cloud fraction over the same region in the high-bias models (cf. Figs. 12g and 13g). Similarly, we find that composite of high-bias models with weak AEJ-S overestimates the subsidence south of ∼20°S over the southeast Atlantic Ocean than the composite of low-bias models with strong AEJ-S. This enhanced subsidence contributes to the reduced total cloud cover over the southeast Atlantic in the high-bias models (cf. Figs. 12g and 13g). While most of the changes in large-scale vertical motion may be away from the core of the AEJ-S region, they nonetheless influence the total cloud cover because the AEJ-S is connected to the large-scale dynamical system, such as the midtroposphere high pressure system and the midlatitude westerly disturbance over the region (e.g., Nicholson and Grist 2003; Jackson et al. 2009; Adebiyi and Zuidema 2016; Kuete et al. 2020).
b. CMIP6 AEJ-S and precipitation distribution
Similar to the cloud distribution, we find a strong relationship between the precipitation rates and the strength of the AEJ-S over equatorial and central Africa. There are substantial variabilities in the magnitude and distribution of the precipitation rates among the CMIP6 models, which is consistent with the variabilities in total cloud cover over equatorial and central Africa that are dominated by mid-to-high level clouds (see Figs. S9 and S12, and Figs. 12 and 14). Although higher precipitation rates generally occur over the equatorial region than the southern flank of central Africa, the region of maximum precipitation varies substantially among individual CMIP6 models and does not follow the observed climatology (cf. Fig. S12 and Figs. 14b,c). Similarly, the ensemble mean of the CMIP6 models shows disagreement with the September–October mean in MERRA-2 and from Tropical Rainfall Measuring Mission (TRMM) observations (top panel in Fig. 14a). Specifically, the ensemble mean of CMIP6 models overestimates the precipitation rates relative to MERRA-2 and TRMM (Fig. 14f). The result also shows that the bias over the equatorial region indicates about 3 times more than the bias directly under AEJ-S. In contrast, MERRA-2 compares well with the observation over the entire region, having the same precipitation rate of 3.2 mm day−1 as TRMM when averaged over the continental portion of AEJ-S (black box in Figs. 14b,c).
The relationship between AEJ-S and simulated precipitation. (top) September–October climatology of precipitation rate (mm day−1; shaded) for the (a) ensemble of CMIP6 models (averaged between 1980 and 2014) and (b) MERRA-2 (1980–2014) and Tropical Rainfall Measuring Mission (TRMM) observation (2000–14). (middle) Composite mean for (d) low-bias models with strong AEJ-S (TaiESM, ACCESS-ESM, IPSL-CM6A) and (e) high-bias models with weak AEJ-S (BCC-ESM, CanESM, and GFDL-ESM). (bottom) Bias in (f) the ensemble of models, (g) low-bias models, and (h) high-bias models. All plots show the 600-hPa easterly winds stronger than 6 m s−1 representing AEJ-S (black contour); MERRA-2 AEJ-S is shown in (c).
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
Our result also suggests that the magnitude of the simulated AEJ-S can be associated with variability in the spatial distribution of precipitation rates. Specifically, we find that low-bias models with strong AEJ-S have overall less precipitation than high-bias models with weak AEJ-S (Figs. 14c,d,g,h). The precipitation rate is about 1.2 mm day−1 lower in the low-bias models with strong AEJ-S than in the high-bias models with weak AEJ-S when averaged directly over the AEJ-S region. This difference is even larger over the equatorial region, especially west of ∼25°E (Figs. 14g,h). However, both low-bias and high-bias models simulate higher precipitation rates to the east of AEJ-S close to the equator, which is consistent with total cloud cover over the same region (cf. Figs. 14g,h to Figs. 12f,g). This bias to the east of AEJ-S is likely due to the effect of topography on the vertical velocity over the region (e.g., Ogwang et al. 2014).
To understand the processes influencing the relationship between AEJ-S and the precipitation distribution, we examine the moisture flux convergence/divergence in CMIP6 models and MERRA-2. Our results showed that moisture flux divergence mostly occurs broadly at the level (and southern flank) of strong midtropospheric easterly winds, including the AEJ-S region (purple shades in Figs. 15a,b), which is consistent across all the CMIP6 models and MERRA-2 (see also Fig. S13). In contrast, the region of moisture flux convergence occurs north of the AEJ-S region, extending farther north of the equator (green shaded region in Fig. 15 and Fig. S13) (e.g., Creese et al. 2019). In addition, we find that low-bias models with realistic AEJ-S and precipitation are associated with stronger midlevel moisture flux divergence over the continent than the high-bias models with weak AEJ-S and higher value of precipitation rates (see purple shades in Figs. 15c,d). Consequently, for a composite of low-bias models (i.e., best-performing models), the strong midlevel moisture flux divergence limits the moisture flux convergence over the midlevel equatorial region north of the AEJ-S, resulting in decreased precipitation rates over the equatorial and central African regions (cf. Figs. 15c and 14d). In contrast, for a composite of high-bias models, weak midlevel moisture flux divergence south of AEJ-S allows for increased moisture flux convergence over the midlevel equatorial region resulting in increased precipitation over the equatorial and central African region (cf. Figs. 15d and 14e). Over the ocean, the AEJ-S leads to a broad region of moisture convergence, which is stronger in low-bias models with strong AEJ-S than the high-bias models with weak AEJ-S (Fig. S12).
The relationship between AEJ-S and simulated moisture flux convergence and divergence. The September–October (1980–2014) mean moisture flux convergence (g kg−1 day−1; green) and divergence (g kg−1 day−1; purple) for (a) ensemble of CMIP6 models, (b) MERRA-2, (c) composite of low-bias models, and (d) composite of high-bias model, all averaged over land (12°–24°E). Also shown are the biases in composite means for (e) low-bias and (f) high-bias models relative to MERRA-2. The black contour highlights the southern African easterly jet (AEJ-S) with a zonal wind speed stronger than −6 m s−1.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
c. CMIP6 AEJ-S and aerosol distribution
Similar to the simulated cloud cover and precipitation in CMIP6, there are substantial bias and variability among the CMIP6 models in simulating the magnitude and climatological distribution of aerosols, which consist primarily of biomass-burning (smoke) aerosols, over central Africa and the southeast Atlantic Ocean during the September and October months. We find that all CMIP6 models underestimate the magnitude of aerosol optical depth (AOD) over the region when compared to observed estimates from MODIS and almost all (12 out of 14) when compared against the reanalysis estimate from MERRA-2 (Figs. 16a–c). Specifically, when averaged over the continental and oceanic AEJ-S region, our results indicate that the AOD from MODIS, MERRA-2, and the ensemble of CMIP6 models is 0.49, 0.33, and 0.27, respectively (see the black box in Figs. 16a–c and Fig. S14). This indicates that the ensemble of CMIP6 models underestimates the observed AOD by 45% (20% relative to MERRA-2). Furthermore, among individual CMIP6 models, there are large variabilities in the magnitude of simulated AOD, with the IPSL-CM6A model simulating the least AOD value, which underestimates MODIS by 74% (Figs. 16d–q; MERRA-2 by 61%).
Aerosol optical depth in CMIP6 models. September–October mean distribution of aerosol optical depth (AOD) for (a) the ensemble of CMIP6 models, (b) MERRA-2, and (c) MODIS, as well as (d)–(q) the individual CMIP6 models. All plots show the 600-hPa easterly winds stronger than 6 m s−1 representing AEJ-S (black contours).
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
In addition, there is substantial diversity in the CMIP6-simulated AOD spatial distribution over the region. For example, the distribution of AOD over land varies substantially, with some models only showing high values of AOD at the coastal regions (e.g., Figs. 16g,j) in contrast to MODIS or MERRA-2 distributions that show the highest AOD valuebeyond the coastal regions and farther inland (e.g., Figs. 16b,c). Similarly, while the transports of the aerosols are often aided by the AEJ-S and the midtropospheric easterly winds, the offshore spatial distribution of AOD also varies significantly, and the relationships between AEJ-S and AOD distribution among the CMIP6 models may sometimes be nonlinear. For example, the westward extents of the climatological aerosol transport vary among the individual models. While some models show offshore climatological distributions that are mostly east–west, most models show aerosol transport that tends in the northwest direction away from the typical zonal direction during September and October period. Consequently, the offshore climatological distribution significantly differs from that observed by MODIS (cf. Figs. 16a,c).
Furthermore, because the aerosol emission treatments vary among CMIP6 models (e.g., Collins et al. 2017; van Marle et al. 2017; see also section S1), resulting in differences in the magnitude of the simulated AOD (see Fig. 16), it is difficult to assess the influence of AEJ-S on simulated aerosol distribution. Therefore, to understand the relationship between AEJ-S and offshore aerosol transport, CMIP6 models must be compared relative to a common reference. To do so, we normalize the AOD spatial distribution by the averaged value over the AEJ-S region for each CMIP6 model (Fig. 17 and Fig. S15). We find that, for a composite of high-bias models with weak AEJ-S, the aerosol distribution is mostly confined over the continent, with climatological transport mostly in the northwest direction (Fig. 17b). In contrast, for a composite of low-bias models with strong AEJ-S, AEJ-S facilitates more lateral, typical zonal transport of aerosol from central Africa (Fig. 17a). This low-bias aerosol distribution with strong AEJ-S is comparable to the observed aerosol distribution from MODIS and MERRA-2 (cf. Figs. 17a,c,d) during the September–October period. Overall, our analysis indicates that an accurate representation of the AEJ-S is important for the magnitude of simulated AOD and the overall distribution of aerosol over continental central Africa and offshore over the southeast Atlantic region.
The relationship between AEJ-S and simulated aerosol distribution. The figure shows normalized aerosol spatial distribution (%) calculated as the percentage ratio of aerosol optical depth (AOD) over every location to the AOD averaged over the AEJ-S region (black box region; 2°–14°S, 0°–24°E) for (a) the composite of low-bias models with strong AEJ-S, (b) the composite of high-bias models with weak AEJ-S, (c) MERRA-2, and (d) MODIS observation. The white contours indicate regions with values equal to 100%. The black contours indicate the 600-hPa easterly winds stronger than 6 m s−1 representing the AEJ-S.
Citation: Journal of Climate 36, 22; 10.1175/JCLI-D-23-0083.1
7. Discussion and summary
The southern African easterly jet (AEJ-S) is an important midtropospheric dynamical feature over central Africa and the southeast Atlantic region with substantial impacts on the regional climate (Fig. 1). This study assesses how well climate models represent the AEJ-S, which is critical to accurately simulate the magnitude and distributions of aerosols, clouds, and precipitation over the region. Specifically, we used station-based measurements, satellite-based observations, and MERRA-2 to evaluate the characteristics and representation of AEJ-S in 14 climate models that are part of phase 6 of the Coupled Model Intercomparison Project (CMIP6). We selected these 14 CMIP6 models based on the models with all available outputs of aerosol, clouds, precipitation rates, and other meteorological variables, including temperature and specific humidity, in the historical forcing experiments (see section 2b). We used the monthly-mean CMIP6 products between 1980 and 2014 and focused on the September–October period when the midtropospheric winds are maximum, biomass-burning (smoke) aerosols are readily emitted and transported offshore, and the precipitation over the region is strongest (Nicholson and Grist 2003; Adebiyi and Zuidema 2016; Creese and Washington 2018; Ryoo et al. 2021).
We find a significant model spread in the strength, position, and spatial distribution of the AEJ-S over central Africa and the southeast Atlantic region among the CMIP6 models. The mean 600-hPa easterly winds show substantial differences in the spatial extent of AEJ-S in CMIP6 models that range from the AEJ-S confined over continental central Africa to the AEJ-S extending far over the southeast Atlantic Ocean (Fig. 4). These differences also reflect in the magnitude of the simulated AEJ-S, with all models underestimating the radiosonde measurements of 600-hPa easterly winds over the meteorological station by an amount that ranges from 0.6 to 3.7 m s−1. (Fig. 9). Similarly, when averaged over the AEJ-S region identified by MERRA-2, all but one model underestimates the MERRA-2 600-hPa easterly winds over land and ocean. Collectively, the ensemble mean of CMIP6 models underestimates the AEJ-S wind speed at 600 hPa by 2.1 ms−1 or by 25% when compared to the observation over the meteorological station and by 0.95 m s−1 (by 0.84 m s−1 over land and 0.99 m s−1 over the ocean) when compared to MERRA-2 averaged over the AEJ-S region (Fig. 9).
Similarly, there are disagreements in the vertical distribution of the AEJ-S among CMIP6 models, as indicated by the spread in the altitude level of the AEJ-S with maximum wind speed (AEJ-S centroid). Specifically, there is a spread among CMIP6 models of about 100 hPa over land and about 68 hPa over the ocean in the simulated altitude of the AEJ-S centroid (Fig. 5 and Fig. S3). This results in an average descent of the AEJ-S centroid by about 15 hPa when moving from land (643 hPa) to ocean (658 hPa) in the ensemble mean of CMIP6 models (Fig. S4). In addition, we find that the AEJ-S centroid averaged over land and ocean in the ensemble of CMIP6 models is lower in altitude than MERRA-2, although this is about 8 hPa lower over land and about 2.8 hPa higher over the ocean (Fig. 9). We also find that when the variabilities in the AEJ-S centroid are considered, CMIP6 models still underestimate the AEJ-S wind speed by an average of 0.3 m s−1, which could range from −1.3 to +1.2 m s−1 among the individual models (Fig. 9). Despite the disagreements among CMIP6 models, we find that the altitude of the near-surface meridional temperature gradients largely correlates with the altitude of the AEJ-S centroid (Fig. 8). However, there is a weak relationship between the magnitude of the negative meridional virtual potential temperature gradients below the AEJ-S and the magnitude at the AEJ-S centroid, suggesting that other dynamical forcings likely influenced the strength of AEJ-S in CMIP6 models.
To assess the impacts of the representation of AEJ-S in CMIP6 models, we classified the models into those with low and high biases based on the estimates of their midtropospheric easterly wind speed assessed at 600 hPa and the AEJ-S centroid. Specifically, we classify the CMIP6 models into low-bias and high-bias models based on the mean absolute biases of midtropospheric easterly wind speeds that are less and greater than 50% of the maximum bias among other models, respectively (Fig. 9). Consequently, we find that a composite of low-bias CMIP6 models results in higher values of near-surface negative meridional temperature gradient and stronger hot-dry convection, resulting in stronger AEJ-S than the high-bias models (Figs. 10 and 11). Overall, the low-bias CMIP6 models better represent the AEJ-S and its associated characteristics than the high-bias models compared to MERRA-2.
Using this classification of AEJ-S representation, we find strong relationships between the representation of the AEJ-S in CMIP6 models and the simulated clouds, precipitation, and aerosol distributions. First, we find that composite of low-bias models with strong AEJ-S better represents the total cloud cover than the composite of high-bias models with weak AEJ-S. Specifically, we find that composite of high-bias models with weak AEJ-S underestimates the total cloud cover over the southeast Atlantic Ocean, where low-level cloud cover dominates substantially more than the composites of low-bias models with strong AEJ-S (Fig. 12). In addition, the composite of high-bias models overestimates the total cloud cover over the central African region, where mid-to-high level clouds dominate more than composites of low-bias models (Fig. 12). Our analysis further reveals that these changes in total cloud cover and distribution are associated with changes in the simulated large-scale vertical velocity (Fig. 13). Specifically, the composites of high-bias models with weak AEJ-S overestimate the mean updraft over the central African region and overestimate the large-scale subsidence over the southeast Atlantic Ocean than the composite of low-bias models with strong AEJ-S.
Second, we find that low-bias models with strong AEJ-S have overall less rainfall over the equatorial and central Africa region than high-bias models with weak AEJ-S (Fig. 14). Further analysis indicates that low-bias models are associated with strong midlevel moisture flux divergence in the vicinity of the AEJ-S, limiting the moisture flux convergence north of the AEJ-S (Fig. 15). In contrast, the midlevel moisture flux divergence in high-bias models is weaker, allowing for stronger anomalous moisture flux convergence north of AEJ-S than the low-bias models. A similar relationship between AEJ-S, moisture flux divergence, and precipitation has also been documented in previous studies, wherein months with stronger AEJ-S have stronger moisture flux divergence and lesser precipitation than months with weaker AEJ-S (e.g., Washington et al. 2013; Adebiyi and Zuidema 2016; Kuete et al. 2020). Overall, the low-bias models better simulate the precipitation over central Africa than the high-bias models when compared against observation and MERRA-2.
Third, the low-bias CMIP6 models with stronger AEJ-S better simulate aerosol distribution over central Africa and the southeast Atlantic Ocean than high-bias models with weaker AEJ-S when compared against observation and MERRA-2. Collectively, the ensemble of CMIP6 models underestimates the observed AOD by about 45% and by about 20% relative to MERRA-2 (Fig. 16). To accommodate the differences in the treatments of aerosol emission among the CMIP6 models, we normalize the simulated AOD spatial distribution by the averaged value over the AEJ-S region for each CMIP6 model. We find that the climatological distribution in the low-bias models indicates more lateral, typical zonal transport of the aerosol over the southeast Atlantic than the high-bias models, which have most of the aerosols confined over the continent or transport northwest to the equatorial Atlantic Ocean (Fig. 17).
Although our results highlight the role of AEJ-S as the central dynamical feature, other large-scale and synoptic-scale dynamical systems over the broader southern African continent and the southeast Atlantic region may also impact aerosol transport, clouds, and precipitation distributions, as discussed in this study. While some dynamical systems are independent of the process associated with AEJ-S or out of phase with its variabilities, other dynamical systems that correlate with the simulated AEJ-S will have their influences on the aerosol, cloud, and precipitation distribution accounted for in our composite analysis. For example, previous studies have highlighted the midlatitude upper-level westerly-wave disturbance can influence the variability of AEJ-S through its impact on vertical velocity and northward advection of cold air over the Namib–Kalahari heat low (Kuete et al. 2020; Ryoo et al. 2021). Consistent with these previous studies, we find increased subsidence south of ∼20°S over the southeast Atlantic Ocean and reduced heat low over the Namib–Kalahari dryland (see Figs. 11–13). In contrast, it is unclear the influence of other dynamical features, such as those associated with the intertropical convergence zone and the northern African easterly wave, which may be independent of AEJ-S (e.g., Hsieh and Cook 2005; Ventrice and Thorncroft 2013), could have on aerosol transport, clouds, and precipitation distribution over central Africa and the southeast Atlantic region. Such analysis will require further studies.
Our results re-emphasize the importance of the AEJ-S as the central dynamical feature essential to accurately estimate climate processes and interactions that involve aerosols, clouds, and precipitation over central Africa and the southeast Atlantic. For example, the distributions of clouds and aerosol are important for aerosol–cloud interactions and the overall offshore radiative budget over the southeast Atlantic. A previous study shows that about 75% of CMIP6 models underestimate the shortwave absorption of aerosol over the southeast Atlantic, in part because of the difficulty in capturing the aerosol spatial distribution (Mallet et al. 2021). As a result, these models also fail to capture the positive (warming) direct radiative forcing over the region, which has substantial implications for the estimates of global energy budgets both in current and future climates (e.g., Stier et al. 2013; Zuidema et al. 2016; Mallet et al. 2021). Although aerosol–cloud interactions also depend on the aerosol vertical distribution that is not shown in this study because of a lack of data for the selected CMIP6 models (see section 2b), previous studies have also shown that models find it difficult to simulate the aerosol vertical structure when AEJ-S is not accurately represented (e.g., Adebiyi and Zuidema 2016; Mallet et al. 2019; Ryoo et al. 2021; Chaboureau et al. 2022). While the southeast Atlantic Ocean is dominated by low-level clouds, previous studies have also highlighted the presence of midlevel clouds that cannot be ignored in aerosol–cloud interaction studies over the region (e.g., Adebiyi et al. 2020). Overall, because the distribution of aerosol is important for aerosol–cloud interactions, accurate representation of dynamical systems in climate models, especially the AEJ-S and associated midtropospheric easterly winds that transport these aerosol particles over the southeast Atlantic is, therefore, important.
Furthermore, accurate representation of AEJ-S is also important for the interactions between meteorology, aerosol, clouds, and precipitation over central Africa. While the representation of the AEJ-S alone could influence the moisture flux convergence and divergence that directly impact precipitation amounts and distributions, it could also affect the distribution of aerosols that serve as cloud condensation nuclei or ice nuclei, therefore indirectly influencing the initiation, intensity, and distribution of precipitation over the region (e.g., Andreae and Rosenfeld 2008). In addition, the radiative effects of the aerosol also have the potential to inhibit convection by increasing the static stability over land (e.g., Tosca et al. 2015; Ajoku et al. 2020). Therefore, a better representation of the AEJ-S could improve the interactions between dynamics and aerosol in accurately simulating the precipitation characteristics over central Africa. Furthermore, inaccurate representation of offshore aerosol transport could bias the estimate of precipitation in other regions. For example, recent studies show that aerosol transports from central Africa have substantial impacts on the West African monsoon system (Solmon et al. 2021). Overall, we conclude that accurate representation of the AEJ-S in climate models is important for processes that influence regional climate over central Africa and the southeast Atlantic Ocean (Fig. 1).
Acknowledgments.
The first author was supported by funding from the University of California–Merced and the University of California Office of the President. The authors appreciate the World Climate Research Programmes (WCRP) Working Group on Coupled Modelling, which is responsible for CMIP6 models, and we acknowledge the climate modeling groups listed in Table 1 for producing and making their model outputs available.
Data availability statement.
MERRA-2 is obtained from the Goddard Earth Sciences Data and Information Services Center (GES DISC) (https://doi.org/10.5067/FH9A0MLJPC7N, last access: 24 August 2018; GMAO 2015). MODIS atmosphere products are available through the LAADS web (http://ladsweb.nascom.nasa.gov/; last access: 10 September 2019; Platnick et al. 2015). The CMIP6 models are openly available and can be accessed here: https://esgf-node.ipsl.upmc.fr/projects/cmip6-ipsl/ (last access: 6 June 2019).
REFERENCES
Adebiyi, A. A., and P. Zuidema, 2016: The role of the southern African easterly jet in modifying the southeast Atlantic aerosol and cloud environments. Quart. J. Roy. Meteor. Soc., 142, 1574–1589, https://doi.org/10.1002/qj.2765.
Adebiyi, A. A., and P. Zuidema, 2018: Low cloud cover sensitivity to biomass-burning aerosols and meteorology over the southeast Atlantic. J. Climate, 31, 4329–4346, https://doi.org/10.1175/JCLI-D-17-0406.1.
Adebiyi, A. A., P. Zuidema, and S. J. Abel, 2015: The convolution of dynamics and moisture with the presence of shortwave absorbing aerosols over the southeast Atlantic. J. Climate, 28, 1997–2024, https://doi.org/10.1175/JCLI-D-14-00352.1.
Adebiyi, A. A., P. Zuidema, I. Chang, S. P. P. Burton, and B. Cairns, 2020: Mid-level clouds are frequent above the southeast Atlantic stratocumulus clouds. Atmos. Chem. Phys., 20, 11 025–11 043, https://doi.org/10.5194/acp-20-11025-2020.
Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 1147–1167, https://doi.org/10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.
Ajoku, O., J. R. Norris, and A. J. Miller, 2020: Observed monsoon precipitation suppression caused by anomalous interhemispheric aerosol transport. Climate Dyn., 54, 1077–1091, https://doi.org/10.1007/s00382-019-05046-y.
Aloysius, N. R., J. Sheffield, J. E. Saiers, H. Li, and E. F. Wood, 2016: Evaluation of historical and future simulations of precipitation and temperature in central Africa from CMIP5 climate models. J. Geophys. Res. Atmos., 121, 130–152, https://doi.org/10.1002/2015JD023656.
Andreae, M. O. O., and D. Rosenfeld, 2008: Aerosol-cloud-precipitation interactions. Part 1. The nature and sources of cloud-active aerosols. Earth-Sci. Rev., 89, 13–41, https://doi.org/10.1016/j.earscirev.2008.03.001.
Bader, D. C., R. Leung, M. Taylor, and R. B. McCoy, 2019a: E3SM-Project E3SM1.1ECA model output prepared for CMIP6 CMIP. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.11444.
Bader, D. C., R. Leung, M. Taylor, and R. B. McCoy, 2019b: E3SM-Project E3SM1.1 model output prepared for CMIP6 CMIP. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.11442.
Bader, D. C., R. Leung, M. Taylor, and R. B. McCoy, 2019c: E3SM-Project E3SM1.0 model output prepared for CMIP6 CMIP. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.2294.
Balas, N., S. E. Nicholson, and D. Klotter, 2007: The relationship of rainfall variability in West Central Africa to sea-surface temperature fluctuations. Int. J. Climatol., 27, 1335–1349, https://doi.org/10.1002/joc.1456.
Beighley, R. E., and Coauthors, 2011: Comparing satellite derived precipitation datasets using the Hillslope River Routing (HRR) model in the Congo River basin. Hydrol. Processes, 25, 3216–3229, https://doi.org/10.1002/hyp.8045.
Boucher, O., and Coauthors, 2018: IPSL IPSL-CM6A-LR model output prepared for CMIP6 CMIP historical. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.5195.
Boucher, O., and Coauthors, 2020: Presentation and evaluation of the IPSL-CM6A-LR climate model. J. Adv. Model. Earth Syst., 12, e2019MS002010, https://doi.org/10.1029/2019MS002010.
Buchard, V., and Coauthors, 2017: The MERRA-2 aerosol reanalysis, 1980 onward. Part II: Evaluation and case studies. J. Climate, 30, 6851–6872, https://doi.org/10.1175/JCLI-D-16-0613.1.
Chaboureau, J.-P., L. Labbouz, C. Flamant, and A. Hodzic, 2022: Acceleration of the southern African easterly jet driven by the radiative effect of biomass burning aerosols and its impact on transport during AEROCLO-sA. Atmos. Chem. Phys., 22, 8639–8658, https://doi.org/10.5194/acp-22-8639-2022.
Chang, I., and Coauthors, 2023: On the differences in the vertical distribution of modeled aerosol optical depth over the southeastern Atlantic. Atmos. Chem. Phys., 23, 4283–4309, https://doi.org/10.5194/acp-23-4283-2023.
Colarco, P., A. Silva, M. Chin, and T. Diehl, 2010: Online simulations of global aerosol distributions in the NASA GEOS‐4 model and comparisons to satellite and ground‐based aerosol optical depth. J. Geophys. Res., 115, D14207, https://doi.org/10.1029/2009JD012820.
Collins, W. J., and Coauthors, 2017: AerChemMIP: Quantifying the effects of chemistry and aerosols in CMIP6. Geosci. Model Dev., 10, 585–607, https://doi.org/10.5194/gmd-10-585-2017.
Cook, K. H., 1999: Generation of the African easterly jet and its role in determining West African precipitation. J. Climate, 12, 1165–1184, https://doi.org/10.1175/1520-0442(1999)012<1165:GOTAEJ>2.0.CO;2.
Creese, A., and R. Washington, 2016: Using qflux to constrain modeled Congo basin rainfall in the CMIP5 ensemble. J. Geophys. Res. Atmos., 121, 13 415–13 442, https://doi.org/10.1002/2016JD025596.
Creese, A., and R. Washington, 2018: A process-based assessment of CMIP5 rainfall in the Congo basin: The September–November rainy season. J. Climate, 31, 7417–7439, https://doi.org/10.1175/JCLI-D-17-0818.1.
Creese, A., R. Washington, and C. Munday, 2019: The plausibility of September–November Congo basin rainfall change in coupled climate models. J. Geophys. Res. Atmos., 124, 5822–5846, https://doi.org/10.1029/2018JD029847.
Danabasoglu, G., D. Lawrence, K. Lindsay, W. H. Lipscomb, and W. G. Strand, 2019: NCAR CESM2 model output prepared for CMIP6 CMIP historical. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.7627.
Danabasoglu, G., and Coauthors, 2020: The Community Earth System Model version 2 (CESM2). J. Adv. Model. Earth Syst., 12, e2019MS001916, https://doi.org/10.1029/2019MS001916.
Driver, P., and C. J. C. Reason, 2017: Variability in the Botswana high and its relationships with rainfall and temperature characteristics over southern Africa. Int. J. Climatol., 37, 570–581, https://doi.org/10.1002/joc.5022.
Dunne, J. P., and Coauthors, 2020: The GFDL Earth System Model version 4.1 (GFDL-ESM 4.1): Overall coupled model description and simulation characteristics. J. Adv. Model. Earth Syst., 12, e2019MS002015, https://doi.org/10.1029/2019MS002015.
Eyring, V., S. Bony, G. A. Meehl, C. A. Senior, B. Stevens, R. J. Stouffer, and K. E. Taylor, 2016: Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev., 9, 1937–1958, https://doi.org/10.5194/gmd-9-1937-2016.
Farnsworth, A., E. White, C. J. R. Williams, E. Black, and D. R. Kniveton, 2011: Understanding the large scale driving mechanisms of rainfall variability over Central Africa. African Climate and Climate Change, Advances in Global Change Research, Vol. 43, Springer, 101–122.
Garstang, M., P. D. Tyson, R. J. Swap, M. Edwards, P. Kallberg, and J. Lindesay, 1996: Horizontal and vertical transport of air over southern Africa. J. Geophys. Res., 101, 23 721–23 736, https://doi.org/10.1029/95JD00844.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
GMAO, 2015: MERRA-2 tavgM_2d_aer_Nx: 2d, Monthly mean,Time-averaged,Single-Level,Assimilation,Aerosol Diagnostics V5.12.4. Goddard Earth Sciences Data and Information Services Center (GES DISC), accessed 24 August 2018, https://doi.org/10.5067/FH9A0MLJPC7N.
Hoesly, R. M., and Coauthors, 2018: Historical (1750–2014) anthropogenic emissions of reactive gases and aerosols from the Community Emissions Data System (CEDS). Geosci. Model Dev., 11, 369–408, https://doi.org/10.5194/gmd-11-369-2018.
Howard, E., and R. Washington, 2018: Characterizing the synoptic expression of the Angola low. J. Climate, 31, 7147–7165, https://doi.org/10.1175/JCLI-D-18-0017.1.
Hsieh, J.-S., and K. H. Cook, 2005: Generation of African easterly wave disturbances: Relationship to the African easterly jet. Mon. Wea. Rev., 133, 1311–1327, https://doi.org/10.1175/MWR2916.1.
Hua, W., L. Zhou, S. E. Nicholson, H. Chen, and M. Qin, 2019: Assessing reanalysis data for understanding rainfall climatology and variability over Central Equatorial Africa. Climate Dyn., 53, 651–669, https://doi.org/10.1007/s00382-018-04604-0.
Huffman, G. J., and Coauthors, 2007: The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor., 8, 38–55, https://doi.org/10.1175/JHM560.1.
Jackson, B., S. E. Nicholson, and D. Klotter, 2009: Mesoscale convective systems over western equatorial Africa and their relationship to large-scale circulation. Mon. Wea. Rev., 137, 1272–1294, https://doi.org/10.1175/2008MWR2525.1.
Jones, P. W., 1998: A User’s Guide for SCRIP: A Spherical Coordinate Remapping and Interpolation Package. Los Alamos National Laboratory, https://github.com/SCRIP-Project/SCRIP.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6, 1587–1606, https://doi.org/10.1175/1520-0442(1993)006<1587:TSCOLS>2.0.CO;2.
Krasting, J. P., and Coauthors, 2018: NOAA-GFDL GFDL-ESM4 model output prepared for CMIP6 CMIP historical. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.8597.
Kuete, G., W. Pokam Mba, and R. Washington, 2020: African Easterly Jet South: Control, maintenance mechanisms and link with southern subtropical waves. Climate Dyn., 54, 1539–1552, https://doi.org/10.1007/s00382-019-05072-w.
Kuete, G., W. Pokam Mba, R. James, E. Dyer, T. Annor, and R. Washington, 2023: How do coupled models represent the African Easterly Jets and their associated dynamics over Central Africa during the September–November rainy season? Climate Dyn., 60, 2907–2929, https://doi.org/10.1007/s00382-022-06467-y.
Lavaysse, C., C. Flamant, S. Janicot, D. J. Parker, J. P. Lafore, B. Sultan, and J. Pelon, 2009: Seasonal evolution of the West African heat low: A climatological perspective. Climate Dyn., 33, 313–330, https://doi.org/10.1007/s00382-009-0553-4.
Lee, W.-L., and H.-C. Liang, 2019: AS-RCEC TaiESM1.0 model output prepared for CMIP6 CMIP. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.9684.
Levy, R. C., S. Mattoo, L. A. Munchak, L. A. Remer, A. M. Sayer, F. Patadia, and N. C. Hsu, 2013: The Collection 6 MODIS aerosol products over land and ocean. Atmos. Meas. Tech., 6, 2989–3034, https://doi.org/10.5194/amt-6-2989-2013.
Mallet, M., and Coauthors, 2019: Simulation of the transport, vertical distribution, optical properties and radiative impact of smoke aerosols with the ALADIN regional climate model during the ORACLES-2016 and LASIC experiments. Atmos. Chem. Phys., 19, 4963–4990, https://doi.org/10.5194/acp-19-4963-2019.
Mallet, M., P. Nabat, B. Johnson, M. Michou, J. M. Haywood, C. Chen, and O. Dubovik, 2021: Climate models generally underrepresent the warming by Central Africa biomass-burning aerosols over the Southeast Atlantic. Sci. Adv., 7, eabg9998, https://doi.org/10.1126/sciadv.abg9998.
Molod, A., L. Takacs, M. Suarez, and J. Bacmeister, 2015: Development of the GEOS-5 atmospheric general circulation model: Evolution from MERRA to MERRA2. Geosci. Model Dev., 8, 1339–1356, https://doi.org/10.5194/gmd-8-1339-2015.
Munday, C., and R. Washington, 2017: Circulation controls on southern African precipitation in coupled models: The role of the Angola Low. J. Geophys. Res. Atmos., 122, 861–877, https://doi.org/10.1002/2016JD025736.
Munday, C., and R. Washington, 2018: Systematic climate model rainfall biases over southern Africa: Links to moisture circulation and topography. J. Climate, 31, 7533–7548, https://doi.org/10.1175/JCLI-D-18-0008.1.
Nicholson, S. E., and J. P. Grist, 2003: The seasonal evolution of the atmospheric circulation over West Africa and equatorial Africa. J. Climate, 16, 1013–1030, https://doi.org/10.1175/1520-0442(2003)016<1013:TSEOTA>2.0.CO;2.
Nicholson, S. E., J. Kim, and J. Hoopingarner, 1988: Atlas of African Rainfall and Its Interannual Variability. Department of Meteorology, Florida State University, 237 pp.
Ogwang, B. A., H. Chen, X. Li, and C. Gao, 2014: The influence of topography on East African October to December climate: Sensitivity experiments with RegCM4. Adv. Meteor., 2014, 143917, https://doi.org/10.1155/2014/143917.
Parker, W. S., 2016: Reanalyses and observations: What’s the difference? Bull. Amer. Meteor. Soc., 97, 1565–1572, https://doi.org/10.1175/BAMS-D-14-00226.1.
Pistone, K., and Coauthors, 2021: Exploring the elevated water vapor signal associated with the free tropospheric biomass burning plume over the southeast Atlantic Ocean. Atmos. Chem. Phys., 21, 9643–9668, https://doi.org/10.5194/acp-21-9643-2021.
Platnick, S., P. A. Hubanks, K. Meyer, and M. D. King, 2015: MODIS Atmosphere L3 Monthly Product (08_L3). MOD08_M3 MODIS/Terra Aerosol Cloud Water Vapor Ozone Monthly L3 Global 1Deg CMG, NASA MODIS Adaptive Processing System, Goddard Space Flight Center, accessed 10 September 2019, https://doi.org/10.5067/MODIS/MYD08_M3.006.
Randles, C. A., and Coauthors, 2016: The MERRA-2 aerosol assimilation. NASA Tech. Rep. Series 45, 156 pp., https://scholar.google.com/citations?view_op=view_citation&hl=en&user=3nfIo84AAAAJ&citation_for_view=3nfIo84AAAAJ:_Qo2XoVZTnwC.
Randles, C. A., and Coauthors, 2017: The MERRA-2 aerosol reanalysis, 1980 onward. Part I: System description and data assimilation evaluation. J. Climate, 30, 6823–6850, https://doi.org/10.1175/JCLI-D-16-0609.1.
Reichle, R. H., and Q. Liu, 2014: Observation-corrected precipitation estimates in GEOS-5. Accessed 30 June 2023, https://ntrs.nasa.gov/citations/20150000725.
Remer, L., and Coauthors, 2005: The MODIS aerosol algorithm, products, and validation. J. Atmos. Sci., 62, 947–973, https://doi.org/10.1175/JAS3385.1.
Rienecker, M. M., and Coauthors, 2008: The GEOS-5 data assimilation system-documentation of versions 5.0.1, 5.1.0, and 5.2.0. NASA Tech. Rep. Series 27, 118 pp., https://ntrs.nasa.gov/citations/20120011955.
Ryoo, J.-M., L. Pfister, R. Ueyama, P. Zuidema, R. Wood, I. Chang, and J. Redemann, 2021: A meteorological overview of the ORACLES (ObseRvations of Aerosols above CLouds and their intEractionS) campaign over the southeastern Atlantic during 2016–2018: Part 1—Climatology. Atmos. Chem. Phys., 21, 16 689–16 707, https://doi.org/10.5194/acp-21-16689-2021.
Sakaeda, N., R. Wood, and P. J. Rasch, 2011: Direct and semidirect aerosol effects of southern African biomass burning aerosol. J. Geophys. Res., 116, D12205, https://doi.org/10.1029/2010JD015540.
Solmon, F., N. Elguindi, M. Mallet, C. Flamant, and P. Formenti, 2021: West African monsoon precipitation impacted by the South Eastern Atlantic biomass burning aerosol outflow. npj Climate Atmos. Sci., 4, 54, https://doi.org/10.1038/s41612-021-00210-w.
Stier, P., and Coauthors, 2013: Host model uncertainties in aerosol radiative forcing estimates: Results from the AeroCom Prescribed intercomparison study. Atmos. Chem. Phys., 13, 3245–3270, https://doi.org/10.5194/acp-13-3245-2013.
Swart, N. C., and Coauthors, 2019: The Canadian Earth System Model version 5 (CanESM5.0.3). Geosci. Model Dev., 12, 4823–4873, https://doi.org/10.5194/gmd-12-4823-2019.
Tamoffo, A. T., and Coauthors, 2019: Process-oriented assessment of RCA4 regional climate model projections over the Congo basin under 1.5°C and 2°C global warming levels: Influence of regional moisture fluxes. Climate Dyn., 53, 1911–1935, https://doi.org/10.1007/s00382-019-04751-y.
Tompkins, A. M., and A. A. Adebiyi, 2012: Using CloudSat cloud retrievals to differentiate satellite-derived rainfall products over West Africa. J. Hydrometeor., 13, 1810–1816, https://doi.org/10.1175/JHM-D-12-039.1.
Toracinta, E. R., and E. J. Zipser, 2001: Lightning and SSM/I-ice-scattering mesoscale convective systems in the global tropics. J. Appl. Meteor., 40, 983–1002, https://doi.org/10.1175/1520-0450(2001)040<0983:LASIIS>2.0.CO;2.
Tosca, M. G., D. J. Diner, M. J. Garay, and O. V. Kalashnikova, 2015: Human-caused fires limit convection in tropical Africa: First temporal observations and attribution. Geophys. Res. Lett., 42, 6492–6501, https://doi.org/10.1002/2015GL065063.
van Marle, M. J. E., and Coauthors, 2017: Historic global biomass burning emissions for CMIP6 (BB4CMIP) based on merging satellite observations with proxies and fire models (1750–2015). Geosci. Model Dev., 10, 3329–3357, https://doi.org/10.5194/gmd-10-3329-2017.
Ventrice, M. J., and C. D. Thorncroft, 2013: The role of convectively coupled atmospheric Kelvin waves on African easterly wave activity. Mon. Wea. Rev., 141, 1910–1924, https://doi.org/10.1175/MWR-D-12-00147.1.
Washington, R., R. James, H. Pearce, W. M. Pokam, and W. Moufouma-Okia, 2013: Congo basin rainfall climatology: Can we believe the climate models? Philos. Trans. Roy. Soc., B368, 20120296, https://doi.org/10.1098/rstb.2012.0296.
Wood, R., 2012: Stratocumulus clouds. Mon. Wea. Rev., 140, 2373–2423, https://doi.org/10.1175/MWR-D-11-00121.1.
Wu, T., and Coauthors, 2020: Beijing Climate Center Earth System Model version 1 (BCC-ESM1): Model description and evaluation of aerosol simulations. Geosci. Model Dev., 13, 977–1005, https://doi.org/10.5194/gmd-13-977-2020.
Wu, W.-S., R. J. Purser, and D. F. Parrish, 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 2905–2916, https://doi.org/10.1175/1520-0493(2002)130<2905:TDVAWS>2.0.CO;2.
Xie, P., A. Yatagai, M. Y. Chen, T. Hayasaka, Y. Fukushima, Ch. M. Liu, and S. Yang, 2007: A gauge based analysis of daily precipitation over East Asia. J. Hydrometeor., 8, 607–626, https://doi.org/10.1175/JHM583.1.
Yukimoto, S., and Coauthors, 2019a: The Meteorological Research Institute Earth System Model version 2.0, MRI-ESM2.0: Description and basic evaluation of the physical component. J. Meteor. Soc. Japan, 97, 931–965, https://doi.org/10.2151/jmsj.2019-051.
Yukimoto, S., and Coauthors, 2019b: MRI MRI-ESM2.0 model output prepared for CMIP6 CMIP. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.621.
Zhang, J., and Coauthors, 2018: BCC BCC-ESM1 model output prepared for CMIP6 CMIP historical. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.2949.
Zhang, Z., K. Meyer, H. Yu, S. Platnick, P. Colarco, Z. Liu, and L. Oreopoulos, 2016: Shortwave direct radiative effects of above-cloud aerosols over global oceans derived from 8 years of CALIOP and MODIS observations. Atmos. Chem. Phys., 16, 2877–2900, https://doi.org/10.5194/acp-16-2877-2016.
Ziehn, T., and Coauthors, 2019: CSIRO ACCESS-ESM1.5 model output prepared for CMIP6 CMIP. Earth System Grid Federation, accessed 1 February 2023, https://doi.org/10.22033/ESGF/CMIP6.2288.
Zipser, E. J., C. Liu, D. J. Cecil, S. W. Nesbitt, and D. P. Yorty, 2006: Where are the most intense thunderstorms on Earth? Bull. Amer. Meteor. Soc., 87, 1057–1072, https://doi.org/10.1175/BAMS-87-8-1057.
Zuidema, P., J. Redemann, J. Haywood, R. Wood, S. Piketh, M. Hipondoka, and P. Formenti, 2016: Smoke and clouds above the southeast Atlantic: Upcoming field campaigns probe absorbing aerosol’s impact on climate. Bull. Amer. Meteor. Soc., 97, 1131–1135, https://doi.org/10.1175/BAMS-D-15-00082.1.