1. Introduction
A prolonged interval of drying desiccated the Sahel region of Africa during the final three decades of the twentieth century. This period, now referred to as the “great Sahel drought,” is arguably the most severe drought recorded during the observational period (Held et al. 2005; Philippon et al. 2010). However, at the turn of the twenty-first century, the severe drying that characterized the Sahel during the late twentieth century transitioned to a wetter period with more abundant rainfall. This transition suggested the possibility of a shift in the predominant climate regime over the coming decades (Paeth and Hense 2004).
The distribution of rainfall is heterogeneous across West Africa. The Sahel, a semiarid zone spanning the width of the continent between 10° and 20°N, lies at the northernmost extent of the West African monsoon (WAM) propagation. Consequently, precipitation in the Sahel is sensitive to changes in the monsoon position. During boreal summer, mesoscale convective systems (MCSs) and vigorous, short-lived squall lines—often organized within synoptic-scale African easterly waves—contribute the majority of rainfall in the Sahel (Leroux 2001; Lebel et al. 2003). Farther south along the Guinea Coast, the northward passing of the intertropical convergence zone (ITCZ) during the boreal spring season and its southward retreat during the boreal fall season provide the coast with two rainy seasons, separated by a brief, dry period during boreal summer.
The predominant view among researchers largely attributes the variability of West African rainfall to changes in global sea surface temperatures, with localized land surface fluxes modulating the sea surface temperature forcing (Zeng et al. 1999; Nicholson 2001; Giannini et al. 2003). Several studies have successfully reproduced the historical record of rainfall in the Sahel using climate models forced only with observed global SSTs (e.g., Giannini et al. 2003; Lu and Delworth 2005), and a number of modeling studies have linked the prolonged drought of the mid-1960s through the early 1990s to the cross-equatorial differential heating observed globally during this time period (Folland et al. 1986; Rotstayn and Lohmann 2002; Hoerling et al. 2006; Ackerley et al. 2011). These results suggest that, as the Southern Hemisphere oceans warmed more rapidly than those in the Northern Hemisphere, the cross-equatorial monsoon circulation weakened, moving moisture and the area of strongest airmass convergence farther south away from the Sahel. Likewise, the studies of Giannini et al. (2003) and Hagos and Cook (2008) found a relationship between warming of the Indian Ocean and the persistent dry conditions throughout the Sahel during the latter half of the twentieth century. These studies attribute large-scale subsidence and the resultant drought conditions over the Sahel to a Rossby wave response to heating of the Indian Ocean.
However, despite the widely accepted role of SSTs in shaping recent rainfall variability in West Africa, the dynamics governing the response of the West African climate to elevated global radiative forcing is less clear. For instance, there is little agreement in the sign of projected twenty-first-century precipitation change over the Sahel and Guinea Coast in the suite of atmosphere–ocean general circulation models (GCMs) archived as part of the Coupled Model Intercomparison Project phase 3 (CMIP3) (Meehl et al. 2007a) and used for the Intergovernmental Panel on Climate Change Fourth Assessment Report (Meehl et al. 2007b). Held et al. (2005) examined output from the CMIP3 models and found the most common response in the Sahel to be a slight increase in precipitation by the end of the twenty-first century. However, this increase in precipitation was not universal, and the Geophysical Fluid Dynamics Laboratory GCM, which matched the observed variability of the twentieth-century rainfall record in the Sahel particularly well, projected substantial drying by the end of the twenty-first century. Haarsma et al. (2005) suggested that continued global warming will cause an increase in precipitation in the Sahel as a result of the expected differential heating between the oceans and the land. They cited a decrease in the mean sea level pressure over land caused by warmer temperatures as the driver of increased monsoon flow—and hence moisture—onto the continent. However, Biasutti et al. (2009) analyzed the CMIP3 models and found an inconsistent relationship between projected strengthening of the land–ocean temperature contrast and twenty-first-century Sahel rainfall. Rather, Biasutti et al. found the correlation between projected changes to the strength and spatial distribution of the Saharan heat low and projected changes in Sahelian rainfall to be robust across the CMIP3 ensemble. Further, despite inconsistencies in projected seasonal mean rainfall totals among the current generation of GCMs, Biasutti and Sobel (2009) found the anomalies at the start and end of the rainy season to be robust across the CMIP3 ensemble, indicating a delay in the monsoon onset over the Sahel by the end of the twenty-first century.
Ackerley et al. (2011) disaggregated the effect of twenty-first-century projected radiative forcing into that due to increased greenhouse gas concentrations and that due to changes in sulfate aerosol loading. The authors found large increases in monsoon precipitation throughout West Africa as a result of enhanced levels of CO2 concentrations. Conversely, their simulations yielded little to no response in precipitation throughout the Sahel and drying in the Guinea Coast as a result of the projected aerosol forcing. South of the Sahel, the CMIP3 ensemble mean shows little change in precipitation throughout the Guinea Coast region, with the number of models split nearly evenly in the projected sign of precipitation anomalies at the end of the twenty-first century (Christensen et al. 2007). As suggested by Biasutti and Giannini (2006), the lack of agreement in the sign of the projected twenty-first-century change in West African precipitation between the CMIP3 atmosphere–ocean GCMs (AOGCMs) suggests that the processes governing the response to changes in radiative forcing must be studied in the individual models.
The fact that changes in atmospheric constituent concentrations can affect the atmospheric circulation both directly through changes in atmospheric heating and indirectly through changes in SSTs further complicates understanding of the proposed mechanisms that are likely to shape West African climate in the future (Lu 2009). In assessing the influences of these direct and indirect effects on the global circulation over the twentieth century, Deser and Phillips (2009) found both to be crucial for explaining particular atmospheric circulation trends, with the direct radiative effect driving changes in midlatitude winds, and the SST forcing modifying the Walker circulation. Likewise, Paeth and Hense (2004) found a recovery in Sahelian rainfall at the end of the twentieth century only in those simulations in which greenhouse forcing was included in addition to observed sea surface temperature forcing. Those authors suggested that both the direct atmospheric response of the tropospheric circulation to enhanced greenhouse gas levels and the indirect response through warming of ocean temperatures and the resultant enhancement of available monsoon moisture were responsible for the precipitation trend reversal. However, in assessing the potential response to elevated greenhouse forcing later in the twenty-first century, Paeth and Hense (2004) did not separate the roles of direct atmospheric radiative forcing and SST forcing, leaving open the question of which forcing will dominate the precipitation and circulation responses of the WAM should global radiative forcing become considerably higher than at present.
Given the significant impact of the great drought of the twentieth century, the response of West African climate to elevated radiative forcing is of great importance to the well-being of those living in the region. The observed shift toward wetter conditions at the turn of the twenty-first century suggests a possible influence from enhanced radiative forcing, but the drought reversal is limited to a short time period and a relatively small change in forcing. We therefore seek to quantify the relative contributions of elevated atmospheric radiative forcing and associated changes in SSTs to projected late twenty-first-century changes in summer precipitation in West Africa and to understand the atmospheric dynamics shaping those relative contributions.
2. Methods
a. Model description and experimental design
Our ensemble experiments were based on the Community Climate System Model version 3 (CCSM3) simulations of the Special Report on Emissions Scenarios (SRES) A1B scenario contributed by the National Center for Atmospheric Research (NCAR) to the CMIP3 project. These CCSM3 simulations are described in Meehl et al. (2006a) and are archived at NCAR.
We conducted a set of four 5-member ensemble simulations designed to isolate the individual and tandem effects of atmospheric and SST forcing in the late twenty-first century of the SRES A1B scenario (Table 1). These simulations used a “time slice” approach, with the atmospheric component of CCSM3 (CAM3) being forced by atmospheric constituent concentrations and SSTs from the original CCSM3 integrations. For each ensemble member, the years 1977–99 were used for the contemporary period and the years 2077–99 were used for the future period. The baseline experiment with twentieth-century atmospheric constituents and SSTs (20Atm_20SST) was forced with 1977–99 atmospheric constituents and the 1977–99 SSTs calculated in the original CCSM3 integrations. Similarly, 20Atm_21SST was forced with 1977–99 atmospheric constituents and the 2077–99 SSTs calculated in the original CCSM3 integrations. Experiment 21Atm_20SST was forced with 2077–99 atmospheric constituents and the 1977–99 SSTs calculated in the original CCSM3 integrations. Experiment 21Atm_21SST was forced with 2077–99 atmospheric constituents and the 2077–99 SSTs calculated in the original CCSM3 integrations.
CAM3 simulations. Each simulation consists of five ensemble members. The first three years are discarded to account for model spinup, yielding 100 years in each simulation. The atmospheric constituents and sea surface temperatures are the same as those in the original CCSM3 A1B simulations archived by NCAR.
We generated a five-member CAM3 ensemble for each of the four simulations. Each of the five members used a SST dataset from one of five coupled CCSM3 AOGCM ensemble members (designated by NCAR as c, e, b.ES, f.ES, and g.ES). For each ensemble member, we reran CAM3 with the same resolution as in the coupled CCSM3 experiments (T85 with 26 levels in the vertical), using the respective time series of monthly-mean CCSM3 SSTs as a prescribed boundary condition. For a given ensemble simulation, the atmospheric constituent concentrations were identical in the five members and represented both anthropogenic and natural forcings, including GHGs, black carbon, ozone, solar variability, volcanic aerosols, and sulfate aerosols, as described in Meehl et al. (2006a). We discarded the first three years of each time slice to account for model spinup, leaving four sets of five-member, 20-yr ensemble CAM3 simulations, with 100 simulated years in each 20-yr analysis period.
We use these four ensemble simulations to calculate five experimental differences that quantify the climate response to prescribed changes in atmospheric constituent concentrations and SSTs (Table 1). Simulations 20Atm_20SST and 21Atm_21SST are time-slice reproductions of the late-century periods of the original coupled CCSM3 integrations. Therefore, we can compare the climate response to the individual twenty-first-century atmospheric and SST forcings with the response to the combined forcing by subtracting 20Atm_20SST from 21Atm_21SST (hereafter referred to as the COMBINED experiment, which can also be compared with the original CCSM3 experiment). In addition, we can test the climate response to twenty-first-century atmospheric constituents and SSTs relative to both the twentieth-century and twenty-first-century baselines. We test the climate response to the twenty-first-century atmospheric constituents by subtracting 20Atm_21SST from 21Atm_21SST (hereafter ATM-21BASE), thereby isolating the influence of late twenty-first-century atmospheric forcing within the context of SSTs that are projected for the late twenty-first century. We compare this response with that calculated by subtracting 20Atm_20SST from 21Atm_20SST (hereafter ATM-20BASE), which isolates the influence of late twenty-first-century atmospheric forcing within the context of SSTs that were calculated for the late twentieth century. Likewise, we test the climate response to SST forcing induced by twenty-first-century atmospheric constituent concentrations by subtracting 21Atm_20SST from 21Atm_21SST (hereafter SST-21BASE), thereby isolating the influence of late twenty-first-century SSTs within the context of atmospheric forcing that is projected for the late twenty-first century. We compare this response with that calculated by subtracting 20Atm_20SST from 20Atm_21SST (hereafter SST-20BASE), which isolates the influence of late twenty-first-century SSTs within the context of atmospheric forcing for the late twentieth century. (See Figs. 2 and 3 for comparisons of the climate response in the COMBINED experiment with that in the coupled CCSM3 experiment and for comparisons of the climate response in the 21BASE experiments with that in the 20BASE experiments.)
b. Analysis
Here we focus on the West African monsoon season. Following Grist and Nicholson (2001), Cook and Vizy (2006), and others, we defined this period as June–September (JJAS). A Student’s t test is employed to calculate the statistical significance of simulated changes (95% confidence level) using a 100-yr sample obtained by concatenating the five 20-yr time slices from each ensemble simulation. We compare the CAM3 representation of precipitation and temperature against the coupled ocean–atmosphere model runs from the coupled CCSM3 ensemble. Note that, due to the unavailability of model output data from one CCSM3 ensemble member (g.ES) in the CMIP3 archives, the comparison is based on the mean of four ensemble members (c, e, b.ES, f.ES). Additionally, we compared the CAM3 representation of atmospheric circulation against data from the NCEP–NCAR reanalysis project 2 (R-2) (Kanamitsu et al. 2002). To facilitate these comparisons, the reanalysis data were regridded to the CAM3 T85 grid.
JJAS percent contribution to precipitation change [(2080–99)–(1980–99)] for the (a) SST-20BASE forcing, (b) ATM-20BASE forcing, and (c) associated synergistic interaction term and the (d) SST-21BASE forcing, (e) ATM-21BASE forcing, and (f) associated synergistic interaction term.
Citation: Journal of Climate 25, 2; 10.1175/2011JCLI4183.1
As noted elsewhere (Pitman and Zhao 2000; Diffenbaugh 2005), the use of absolute values in this calculation is designed to allow both positive and negative changes to be counted toward the total change. The synergistic interaction term is defined as | COMBINED – (SST + ATM) |, or the difference between the COMBINED experiment and the sum of the individual SST and ATM experiments. The percent contributions are calculated for both the ATM-20BASE and SST-20BASE and the ATM-21BASE and SST-21BASE forcings.
3. Precipitation and temperature anomalies
a. Relative contribution to precipitation change over West Africa
Figure 1 quantifies the percent contribution to total simulated JJAS precipitation change from changes in atmospheric constituent concentrations, changes in SSTs, and the synergistic interaction between the two (see methods). The percent contribution is very similar between the 20BASE and 21BASE experiments (cf. panels a,d; b,e; c,f). Both ATM-20BASE and ATM-21BASE forcings contribute between 40% and 80% of the total simulated precipitation change throughout the western and central Sahel but no greater than 30% in the Guinea coast region. In particular, the precipitation response over Mali, Niger, Burkina Faso, and Mauritania is dominated by changes in direct atmospheric radiative forcing. Further, areas of the western Sahel—such as over Mali, Mauritania, and Niger—experience a contribution from SST forcing of 30% or less in both SST-20BASE and SST-21BASE experiments. Conversely, the SST forcing contributes between 30% and 70% of the precipitation response in areas of the central Sahel (15°–30°E) in both SST-20BASE and SST-21BASE experiments. Over the Guinea coast, the SST forcing contributes between 50% and 80% of the total simulated precipitation change. Although the precipitation response to atmospheric and SST forcings is largely additive in both 20BASE and 21BASE experiments, the synergistic interaction term does help to determine the total simulated precipitation change over some areas, most notably in the Sahara.
The contrasting influences of atmospheric and SST forcing over the Sahel and Guinea coast regions is also seen in the pattern correlation of precipitation anomalies in the individual forcing experiments with precipitation anomalies in the COMBINED experiment (Table 2). Over the Sahel region, the pattern correlations between the precipitation anomalies in the COMBINED and ATM-21BASE (0.82) and ATM-20BASE (0.55) experiments are higher than those in the SST-21BASE (0.68) and SST-20BASE (0.52) experiments. Similarly, over the Guinea coast region, pattern correlations between the COMBINED and SST-21BASE (0.87) and COMBINED and SST-20BASE (0.90) are higher than those of the ATM-21BASE (0.03) and ATM-20BASE (−0.08) experiments. The role of atmospheric forcing is even greater over the western Sahel (15°W–15°E), yielding pattern correlations of 0.65 and 0.85 in the ATM-20BASE and ATM-21BASE experiments, respectively. These correlations further quantify the decisive influence of direct atmospheric radiative forcing over much of the Sahel and of SST forcing over the Guinea coast region. In the following sections, we assess the dynamics shaping this disparity in influence between the two effects.
Pattern correlation of precipitation anomalies in the ATM-21BASE, SST-21BASE, ATM-20BASE, and SST-20BASE experiments with precipitation anomalies in the COMBINED experiment. Sahel (12°–20°N, 15°W–30°E), Guinea coast (4°–12°N, 15°W–15°E).
b. Seasonal precipitation change
The coupled CCSM3 simulation of twentieth-century summer climate over West Africa has been examined in previous studies (Cook and Vizy (2006); Meehl et al. 2006b). The coupled model is able to capture the large-scale east–west oriented precipitation band associated with the region’s summer climate and the associated surface westerlies of the WAM. However, the onset of the monsoon over land is early by an average of one to two months, and precipitation migrates slightly too far north, into the southern extent of the Sahara.
Summer precipitation changes simulated in the COMBINED experiment are similar to those simulated in the original coupled CCSM3 experiment (Figs. 2a,b). In the COMBINED experiment, statistically significant drying of 0.25–0.75 mm day−1 is projected just inland along the Guinea coast regions of Ghana, the Ivory coast, Guinea, and Nigeria. A dry anomaly between 0.25 and 1.25 mm day−1 associated with a slight northward shift in the location of the monsoon circulation is also found along the equator in the Gulf of Guinea and the eastern Atlantic. Precipitation enhancement of 0.25–0.75 mm day−1 is projected along the immediate southern coastline of Guinea, with slightly greater increases of 1.00–2.00 mm day−1 along the western coastline. Enhanced rainfall values in excess of 2.00 mm day−1 are found in a zonal band between 7° and 11°N off the African mainland. The ensemble mean from the COMBINED simulation also exhibits increases in rainfall of 0.25–1.00 mm day−1 throughout much of the Sahel east of 5°W and over the southern Sahara.
JJAS change in total precipitation [(2080–99)–(1980–99)] for (a) the coupled CCSM3 simulation, (b) the COMBINED forcing, the (c) ATM-20BASE and (d) SST-20BASE forcings, and the (e) ATM-21BASE and (f) SST-21BASE forcings. Differences are significant at the 95% confidence level.
Citation: Journal of Climate 25, 2; 10.1175/2011JCLI4183.1
Analysis of the ATM-20BASE and ATM-21BASE experiments (Figs. 2c,e) shows a significant band of drying located along the equator to roughly 7°N in the eastern Atlantic and Gulf of Guinea but little impact on precipitation over land in the Guinea coast region. Conversely, the ATM-20BASE and ATM-21BASE forcings exhibit the enhanced precipitation response over the western Sahel seen in the COMBINED experiment, along with the wet anomaly projected over the central Sahel (15°E–30°W). The ATM-20BASE and ATM-21BASE experiments yield similar patterns of precipitation response, although the projected drying over the Gulf of Guinea and the wet anomaly in northwest Africa is stronger in the ATM-21BASE experiment.
The SST-20BASE and SST-21BASE experiments display the dry anomaly simulated throughout inland portions of the Guinea coast in the COMBINED experiment (Figs. 2d,f). Likewise, the zonal band of precipitation enhancement over the Atlantic between 7° and 11°N seen in the COMBINED experiment is also present in the SST experiments but with a slightly broader latitudinal extent between 5°N and 11°N. Conversely, the SST-20BASE and SST-21BASE experiments produce little change in precipitation throughout the western Sahel (west of 15°E), although they exhibit the wet anomaly observed in the central Sahel and southern Sahara in the COMBINED experiment. The SST-20BASE and SST-21BASE experiments yield similar patterns of precipitation response, although the values are slightly different along the southern Sahara border and in northwest Africa.
c. Seasonal temperature change
Summer 2-m air temperature changes simulated in the COMBINED experiment are very similar to those simulated in the original coupled CCSM3 experiment (Figs. 3a,b). The greatest projected warming is found in the Sahara north of 25°N, as well as in the far western Sahel, where temperatures increase from 3° to 4°C in the COMBINED experiment. More moderate increases from 2° to 3°C are projected for the Guinea coast region and southern portions of the Sahel from 10°W to 30°E in the COMBINED experiment. The smallest temperature increases are found in those locations where precipitation is projected to increase, particularly over the northern Sahel and far southern reaches of the Sahara between 15° and 22°N where temperatures increase between 1° and 2°C.
As in Fig. 2 but for the JJAS change in 2-m air temperature [(2080–99)–(1980–99)] for (a) the coupled CCSM3 simulation, (b) the COMBINED forcing, (c),(d) the ATM-20BASE and SST-20BASE forcings, and (e),(f) the ATM-21BASE and SST-21BASE forcings. Differences are significant at the 95% confidence level.
Citation: Journal of Climate 25, 2; 10.1175/2011JCLI4183.1
The difference between the ATM-20BASE and ATM-21BASE experiments is minimal, although the simulated cooling throughout the Sahel is greater in the ATM-21BASE experiment (Figs. 3c,e). Likewise, the SST-20BASE and SST-21BASE experiments resemble one another, with only slight differences between the two in portions of the northern Sahel. While the atmospheric and SST forcing both result in a pattern of temperature change over land similar to that of the COMBINED experiment, direct atmospheric forcing produces considerably weaker surface temperature increases over most of the continent. For instance, surface temperature warms by 2.5°–3.5°C over the Sahara and the far western Sahel in the SST-21BASE experiment but by only 0.5°–1.5°C over these areas in the ATM-21BASE experiment. In addition, the portion of the Sahel that shows minimum warming in the SST-21BASE experiment (1°–2°C) exhibits cooling in the ATM-21BASE experiment of (−0.5° to −1.5°C). These areas that exhibit JJAS cooling or reduced warming also exhibit increases in JJAS rainfall (Fig. 2), suggesting that cloud cover and surface moisture feedbacks (via increases in evapotranspiration, and latent heat flux) mute the surface temperature warming.
The late twenty-first-century surface temperature response in our COMBINED experiment is markedly similar to that seen in the CMIP3 multimodel mean (Christensen et al. 2007). This response is robust across the CMIP3 ensemble (Christensen et al. 2007), suggesting that the CMIP3 surface temperature response over West Africa is similarly dominated by SST forcing. A notable exception is the relative minimum in surface temperature increase along 15°–20°N seen in our COMBINED experiment (Fig. 3b) but not present in the CMIP3 ensemble mean. Given the sensitivity of this region to land/sea surface–atmosphere coupling, the disagreement in the sign of future precipitation anomalies throughout the Sahel (Held et al. 2005; Cook and Vizy 2006; Biasutti and Giannini 2006; Christensen et al. 2007; Biasutti et al. 2009) likely inhibits the formation of this localized minimum in surface temperature response when averaging across the multimodel ensemble.
d. Impacts of experimental design on simulated precipitation and temperature response
Our experimental design is potentially limited by the lack of dynamic atmosphere–ocean coupling in our atmosphere-only model simulations. As noted in Deser and Phillips (2009), atmosphere-only simulations forced with changing greenhouse gas concentrations are subject to artificially large downward surface energy fluxes. Similarly, the use of prescribed SST distributions in an atmosphere-only model simulation may be problematic over those ocean basins where overlying atmospheric circulation is the proximate cause of those SSTs (Kumar and Hoerling 1998). The lack of coupling can therefore potentially alter the simulated seasonal precipitation over some regions. In CCSM3, these regions are mostly confined to oceanic basins (Ashfaq et al. 2010). Further, the simulated precipitation and temperature changes are very similar in the COMBINED and coupled CCSM3 experiments over West Africa and the eastern Atlantic (Figs. 2 and 3), indicating that the lack of ocean–atmosphere coupling does not substantially alter the response in our atmosphere-only CAM3 simulations.
Further, our experimental design is potentially limited by the prescription of individual atmospheric and SST forcings in the atmosphere-only CAM3 simulations. This limitation is of least concern for the 20Atm_20SST and 21Atm_21SST simulations, which are time-slice reproductions of the late century periods of the original coupled CCSM3 integrations, and therefore use SSTs that were calculated in response to the respective late twentieth- and late twenty-first-century atmospheric constituent concentrations. Alternatively, this limitation is potentially of greater concern for the 20Atm_21SST and 21Atm_20SST simulations, which are forced with SSTs that were calculated in response to a considerably higher or lower level of atmospheric forcing, respectively. This forcing imbalance is an inherent feature of our atmosphere-only experiments, which are intended to test the atmospheric and SST forcings individually and in tandem. If this imbalance had a substantial impact on the calculated response to the individual forcings, we would expect to see a different calculated response relative to a baseline of twentieth-century atmospheric constituents and SSTs (20Atm_20SST) than relative to a baseline of twenty-first-century atmospheric constituents and SSTs (21Atm_21SST). However, our comparisons of the ATM-20BASE, ATM-21BASE, SST-20BASE, and SST-21BASE experiments show that the response to the individual forcings in our simulations is mostly insensitive to the baseline state (Figs. 2 and 3), suggesting that the impact of this forcing imbalance is minimal and that the simulated responses of precipitation and temperature are robust (at least within the CCSM3 model). Given this robustness, we proceed by focusing on the atmospheric processes shaping the precipitation and temperature responses in the ATM-21BASE and SST-21BASE experiments.
4. Circulation anomalies
a. SLP and surface wind patterns over West Africa
Figure 4 shows JJAS sea level pressure (SLP) and surface wind vectors from the NCEP reanalysis product and the CAM3 simulations. During the boreal summer months, a positive meridional surface temperature gradient exists between the wet, relatively cool Guinea coast and the dry, hotter Sahara to the north. The Saharan heat low (SHL) forms over the Sahara, resulting in westerly flow onto the continent at the equatorward side of the cyclonic feature. In addition, anticyclonic circulation around the St. Helena high near 30°S results in southerly flow toward the equator and onto West Africa. The low-level westerlies that result from these two pressure systems (roughly between the equator and 20°N) mark the West African monsoon circulation. CAM3 captures the basic low-level circulation patterns seen in the reanalysis data (Figs. 4a,b). Notable exceptions include stronger-than-observed sea level pressure in the Azores high in the North Atlantic and in the St. Helena high in the South Atlantic, along with stronger-than-observed westerly flow onto West Africa between 5° and 15°N emanating from the cyclonic circulation around the SHL.
JJAS (1980–99) mean sea level pressure and surface winds from (a) the NCEP reanalysis, (b) the 20Atm_20SST CAM3 simulation, and (c) the 21Atm_21SST CAM3 simulation. Change in JJAS mean sea level pressure and surface winds [(2080–99)–(1980–99)] for (d) the COMBINED forcing, (e) the SST-21BASE forcing, and (f) the ATM-21BASE forcing.
Citation: Journal of Climate 25, 2; 10.1175/2011JCLI4183.1
Figures 4d,e,f show the changes in SLP and surface wind vectors simulated by the CAM3 ensemble between the future (2080–99) and contemporary (1980–99) periods. In the COMBINED experiment, SLP decreases by 1–2 mb over all of North Africa and the Mediterranean, resulting in a cyclonic circulation anomaly centered over the Mediterranean. Associated with this cyclonic circulation anomaly is an increase in the monsoon westerlies between 10° and 20°N. Southerly flow strengthens directly along the shore of the Guinea coast. A slight weakening of the St. Helena high in the South Atlantic causes a reduced anticyclonic circulation, diminishing the equatorward strength of the southeasterly trade winds.
Both SST-21BASE and ATM-21BASE experiments exhibit the anomalous pressure gradient force directed from the Gulf of Guinea and equatorial Atlantic toward northern Africa that is observed in the COMBINED experiment (Figs. 4e,f). The resultant winds shift the rainbelt associated with the monsoon further northward in each of the ATM-21BASE, SST-21BASE and COMBINED experiments. However, unlike the COMBINED experiment, the anomalous low observed in the SST-21BASE experiment is confined to the Sahara (north of 20°N) and Mediterranean. As a result, the resultant increase in westerly flow from the Atlantic is limited to the latitudes north of 15°N (Fig. 4e). Conversely, SLP decreases over all of northern Africa and the Sahel in the ATM-21BASE experiment (and very little south of the equator), yielding enhanced westerly flow onto the West African coast north of 5°N (Fig. 4f). Farther south, the weakening of the St. Helena high and the resulting cyclonic circulation anomaly over the South Atlantic are caused primarily by the simulated SST forcing (Fig. 4e). The SST forcing also results in an anomalous southeasterly flow over the Gulf of Guinea and the Guinea coast, as well as enhanced convergence of low-level winds over the Atlantic between the equator and 10°N.
Analyzing the CMIP3 suite, Biasutti et al. (2009) found a significant correlation between a deepening of the SHL and an increase in precipitation throughout the Sahel by the end of the twenty-first century. In particular, while all 21 GCMs simulate an enhanced land–ocean temperature contrast and a stronger SHL, the anomalous low is spatially confined to the northwest Sahara and Arabian Peninsula in those models that project a drier future for the Sahel by the end of the twenty-first century. Conversely, an increase in precipitation is found throughout the Sahel in those models that simulate a low that spreads across the entire Sahara into the Mediterranean, Europe, and the Middle East, and as far south as the Sahel. The results of our atmospheric and SST forcing experiments are consistent with the AOGCM analyses of Biasutti et al. (2009). Although the surface temperature response in the COMBINED experiment is largely driven by the SST-21BASE forcing, the greatest change to the SHL is seen in response to the direct atmospheric forcing (ATM-21BASE) (Fig. 4). The anomalous low resulting from the ATM-21BASE forcing extends across all of the Sahara into the Mediterranean, Europe, and the Middle East and as far south as the Sahel, with precipitation increasing across both the western and central Sahel. The simulated changes to the SHL in those CMIP3 GCMs that project a wetter twenty-first century for the Sahel may at least partly reflect the response to direct atmospheric radiative forcing isolated in our experiment.
b. Dynamics shaping Sahel precipitation
The sharp contrast in the simulated change in surface air temperature between the Guinea coast and the Sahel, and the Sahel and the Sahara, resulting from both the atmospheric and SST forcings (Figs. 3e,f) causes a modification of the regional meridional surface temperature gradient (Fig. 5). The strongest contemporary (positive) meridional gradient lies between 20° and 22°N (west of 5°E) and between 15° and 20°N (east of 5°E). The location of the maximum positive meridional surface temperature gradient shifts northward in the COMBINED experiment (Fig. 5), as well as in both SST-21BASE and ATM-21BASE experiments, with the effects of atmospheric and SST forcing being nearly additive (not shown). Such a shift in the surface temperature gradient is likely to affect the zonal atmospheric circulation pattern through the thermal wind relationship (Cook 1999; Thorncroft and Blackburn 1999; Maynard et al. 2002). The positive meridional surface temperature gradient induces easterly shear and eventually easterly flow above the surface westerlies. This easterly flow constitutes the African easterly jet (AEJ) at 600 mb. The position of the AEJ is determined by the location of the strongest meridional surface temperature gradient (Burpee 1972; Thorncroft and Blackburn 1999; Cook 1999) and can therefore be expected to be sensitive to the simulated northward shift in this gradient.
Change in JJAS meridional surface temperature gradient [(2080–99)–(1980–99)] for the COMBINED forcing. Contour interval is 0.001°C km−1. Dashed contours denote negative values. Background shading is the mean (1980–99) meridional surface temperature gradient from the 20Atm_20SST CAM3 simulation.
Citation: Journal of Climate 25, 2; 10.1175/2011JCLI4183.1
The core of the AEJ does indeed shift slightly northward in the COMBINED experiment (Figs. 6b,c; see boxes). The jet is weaker in the late twenty-first-century simulation, and the center of its core shifts from 16° to 17°N. There is also a strengthening of westerly flow throughout the lowest levels of the troposphere between 10° and 20°N in the COMBINED experiment (Fig. 6f). Above 850 mb there is a weakening of the easterly winds, particularly in the southern flank of the AEJ. Further, the combined forcing reduces the easterly flow of the TEJ between 250 and 200 mb, and strengthens the jet aloft (above 150 mb). [It should be noted that CAM3 captures the major zonal circulation features characteristic of the twentieth-century boreal summer season over West Africa, although the AEJ is positioned 4° farther north and is weaker (3 m s−1 in the jet core) than in the reanalysis dataset, and the tropical easterly jet (TEJ), located at 200 mb in the reanalysis dataset, is stronger and positioned northward and higher in the atmosphere in the CAM3 simulation (Fig. 6).]
JJAS (1980–99) mean zonal wind profiles averaged from 15°W to 30°E for (a) the NCEP reanalysis, (b) the 20Atm_20SST CAM3 simulation, and (c) the 21Atm_21SST CAM3 simulation. Change in JJAS mean zonal wind [(2080–99)–(1980–99)] for (d) the ATM-21BASE forcing, (e) the SST-21BASE forcing, and (f) the COMBINED forcing. Square boxes centered between 10° and 20°N in (b),(c) highlight the weakening and slight northward shift in the AEJ by the end of the twenty-first century.
Citation: Journal of Climate 25, 2; 10.1175/2011JCLI4183.1
The zonal mean circulation shows a strengthening of monsoon westerlies in the lower troposphere in the ATM-21BASE experiment, along with a weakening of the AEJ and an enhancement of the TEJ. Both strength and position of the AEJ have been shown to influence rainfall variability throughout the Sahel (Cook 1999; Grist and Nicholson 2001; Mohr and Thorncroft 2006). A weaker and anomalously northward AEJ is common in wet Sahelian summers, while a stronger and southward-displaced AEJ is found during Sahelian drought conditions (Newell and Kidson 1984). As noted in Cook (1999) and Abiodun et al. (2008), the AEJ influences midtropospheric moisture divergence over West Africa, and hence precipitation over the western Sahel by transporting moisture away from the continent (toward the Atlantic) below the level of condensation. The increase in precipitation throughout the western and central Sahel in response to the atmospheric forcing (Fig. 2e) is therefore consistent with a weakened AEJ exhibited by the ATM-21BASE experiment.
Further, as Grist and Nicholson (2001) emphasize, not only does the AEJ influence rainfall variability throughout West Africa, but it also responds to that rainfall variability. The AEJ is highly sensitive to surface and lower-tropospheric temperatures. Therefore, increased precipitation and the resultant modification of low-level temperatures through changes in latent and sensible heat flux directly influences the intensity and positioning of the AEJ through the thermal–wind relationship, which in turn modulates the precipitation distribution through offshore advection of midtropospheric moisture and through the development of transient African easterly wave disturbances.
The increase in precipitation along the border of the Sahel/Sahara in the ATM-21BASE experiment can be traced to the increase in moist, westerly flow from the Atlantic and the northward displacement of the monsoon circulation (Fig. 4f). This yields stronger convergence along the intertropical front (ITF), where the dry Harmattan northeasterlies meet the moist southwesterlies of the monsoon circulation. Indeed, the atmospheric forcing results in increased vertical motion throughout the troposphere over much of West Africa north of 15°N (Fig. 7d). Changes in vertical velocity favoring rising motion are strongest in the low to mid levels of the atmosphere between 19° and 24°N in the ATM-21BASE experiment and are linked to greater moisture convergence at low levels along the ITF.
JJAS (1980–99) mean vertical velocity (omega) profiles averaged from 15°W to 30°E for (a) the NCEP reanalysis, (b) the 20Atm_20SST CAM3 simulation, and (c) the 21Atm_21SST CAM3 simulation. Change in JJAS mean vertical velocity [(2080–99)–(1980–99)] for (d) the ATM-21BASE forcing, (e) the SST-21BASE forcing, and (f) the COMBINED forcing.
Citation: Journal of Climate 25, 2; 10.1175/2011JCLI4183.1
The SST forcing also triggers enhanced westerly flow onto northern West Africa and a northward displacement of the monsoon circulation (Fig. 4e), which in turn produces a precipitation pattern along the Sahel/Sahara border and the ITF that is very similar to that seen in the ATM-21BASE experiment. Between 20° and 24°N the combination of increased rising motion in response to both atmospheric and SST forcings results in enhanced upward vertical velocities of 0.6–1.0 Pa s−1 in the COMBINED experiment, and yields increased precipitation over most of northern Africa within this latitude band (Fig. 2b). Like the atmospheric forcing, the SST forcing induces weaker easterly flow at the level of the AEJ. However, unlike the atmospheric forcing, the SST forcing reduces the strength of the TEJ at 200 mb (Fig. 6e). A stronger TEJ is associated with wet Sahelian summer conditions, while a weaker TEJ is commonly observed during dry Sahel summers (Newell and Kidson 1984). Grist and Nicholson (2001) also note that there is greater upper-level divergence during wet years over the Sahel when the TEJ is stronger, suggesting that the stronger TEJ may provide an environment more conducive to ascent at lower levels of the troposphere and hence result in a greater number of convective disturbances. Suppression of rising motion is clearly visible throughout the midlevels of the troposphere in response to the SST forcing in our experiments (Fig. 7e). In particular, between 8° and 16°N, rising motion is suppressed above the AEJ and below the TEJ. The competing effects from the change in the AEJ and TEJ due to SST forcing help to explain the insignificant change in precipitation observed in the western Sahel in response to the SST forcing (Fig. 2d).
[It is important to note that the contemporary CAM3 ensemble mean simulates rising motion between 12° and 18°N above 700 mb, which is not present in the NCEP reanalysis dataset (Figs. 7a,b). This is likely a result of the excessive easterly flow associated with the TEJ and resultant upper-level divergence, as well as stronger-than-observed cyclonic circulation near the surface in the CAM3 contemporary simulation.]
Although the AEJ and TEJ are instrumental in determining the precipitation over West Africa, low-level moisture convergence also plays a significant role (Cook 1999; Hagos and Cook 2008; Lu 2009). Figures 8a,b show the JJAS change in moisture divergence/convergence 
Change in JJAS mean 925-mb moisture divergence [(2080–99)–(1980–99)] for (a) the ATM-21BASE forcing and (b) the SST-21BASE forcing. Anomalous moisture divergence due to the change in winds for (c) the ATM-21BASE forcing and (d) the SST-21BASE forcing. Anomalous moisture divergence due to the change in specific humidity for (e) the ATM-21BASE forcing and (f) the SST-21BASE forcing. Line contours of the respective precipitation anomalies are overlain on (a) and (b) (interval 0.5 mm day−1). Line contours of change in moisture divergence from (a) and (b) are overlain on (c)–(f) (interval 1.5 × 10−8 kg kg−1 s−1). Contours with magnitude greater than 0.5 × 10−8 shown.
Citation: Journal of Climate 25, 2; 10.1175/2011JCLI4183.1
To further understand how the atmospheric and SST forcings influence the observed changes in precipitation and circulation, we decompose the change in low-level moisture divergence into that due to anomalous humidity and mean flow 
c. Dynamics shaping Guinea coast precipitation
The Guinea coast derives much of its precipitation from the north–south migration of the ITCZ. The processes that control moisture convergence in this region are therefore critical in determining the magnitude and spatial distribution of rainfall. The simulated increase in surface temperature over the ocean in the SST-21BASE experiment (Fig. 3f) results in higher specific humidity levels in the lower troposphere through the Clausius–Clapeyron relation. This increase in atmospheric moisture combined with the increase in southeasterly flow from the equator to the Guinea coast (Fig. 4e) results in enhanced moisture convergence (Fig. 8b), and hence the increased precipitation projected directly along the coastline in the COMBINED and SST experiments (Fig. 2). Both SST-21BASE and ATM-21BASE experiments exhibit moisture divergence just south of the Guinea coast over the Gulf of Guinea and equatorial Atlantic (Figs. 8a,b), consistent with the northward migration of the monsoon expected from the changes in SLP seen in the COMBINED experiment. Over the interior of the Guinea coast, the anomalous moisture divergence due to changes in low-level circulation in the SST-21BASE experiment is opposite to that due to changes in low-level humidity, suggesting that the SST-induced dry anomaly projected in the interior Guinea coast is driven by the divergence of the low-level winds (Figs. 8d,f). In addition to the northward shift in the monsoon circulation, the SST-driven simulated change in the low-level winds (Fig. 4e) results in greater moisture convergence over the Atlantic west of the Guinean coast between 5°N and 10°N (Fig. 8b). This convergence results in greater precipitation over the ocean and limits moisture availability over the Guinea coast within this latitude band.
5. Conclusions
We test the relative roles of atmospheric radiative forcing and associated SST forcing in shaping potential future changes in the boreal summer climate over West Africa. We find that the precipitation and temperature response over West Africa to the two forcings are robust within our experimental design.
We find the direct atmospheric radiative forcing dominates changes in precipitation over the western Sahel, yielding a wet anomaly over the region by the end of the twenty-first century. The direct radiative forcing increases precipitation over the Sahel by 1) strengthening the Saharan heat low (SHL), which brings increased westerly monsoon flow onto West Africa; 2) weakening the African easterly jet (AEJ), which halts the advection of moisture offshore at midlevels of the atmosphere; and 3) increasing the strength of the tropical easterly jet (TEJ), which helps to promote ascent at midlevels of the atmosphere. The combination of a projected northward shift in the monsoon circulation and increased westerly and southwesterly flow from the Atlantic enhances moisture convergence along the intertropical front (ITF) and results in increased rainfall across the Sahel/Sahara border as a result of both the atmospheric and SST forcings. In the western Sahel, the SST forcing has a minimal effect on late twenty-first-century precipitation changes. Although the changes in SSTs weaken the strength of the AEJ, the concomitant weakening of the TEJ provides an environment that is less conducive to vertical ascent at midlevels, providing an overall suppression of rainfall.
The effect of direct atmospheric radiative forcing over the Guinea coast is far less pronounced than over the Sahel, with no significant changes simulated at the 95% confidence level. Rather, the response of precipitation over the Guinea coast is dominated by the effects of changes in SSTs. In particular, precipitation is reduced over the interior Guinea coast through a weakening of southwesterly flow onto West Africa that results in low-level moisture divergence. This weakening of southwesterly flow is likely the result of the decrease in sea level pressure associated with the weakening of the St. Helena high in the South Atlantic in response to SST forcing. The most marked increases in precipitation in response to SST forcing appear offshore to the west of the Guinea coast where moisture convergence increases. This increase in precipitation over the ocean reduces moisture availability for the interior of the Guinea coast.
Taken together, our results suggest that the direct radiative effects of late twenty-first-century changes in atmospheric constituent concentrations could cause a change in the climate dynamics of West Africa that partly ameliorates the potential of extended drought events like those experienced in the Sahel during the latter half of the twentieth century. Conversely, changes in SSTs induced by enhanced atmospheric radiative forcing could cause decreases in precipitation over much of the Guinea coast region. This dipole is opposite to what was experienced during the final decades of the twentieth century and suggests a potential shift in the regional moisture balance of West Africa should global radiative forcing continue to increase. While the lack of agreement in the simulated climate response over West Africa between different coupled GCMs suggests that specific regional details of the individual and combined effects of atmospheric and SST forcing are also likely to vary between models, our results highlight the important role that climate system feedbacks play in shaping the response of regional-scale climate to global-scale changes in radiative forcing.
Acknowledgments
This work was supported in part by NSF Award 0450221 and by the World Bank’s Trust Fund for Environmentally and Socially Sustainable Development. Computational resources were provided by Information Technology at Purdue (the Rosen Center for Advanced Computing, West Lafayette, Indiana). We thank the CCSM Climate Change Working group for making the CCSM3 simulations available at NCAR. NCEP reanalysis data were provided by the NOAA/ OAR/ESRL PSD, Boulder, Colorado, from their Web site at http://www.cdc.noaa.gov/.
REFERENCES
Abiodun, B. J., J. S. Pal, E. A. Afiesimama, W. J. Gutowski, and A. Adedoyin, 2008: Simulation of West African monsoon using RegCM3. Part II: Impacts of deforestation and desertification. Theor. Appl. Climatol., 93, 245–261.
Ackerley, D., B. B. B. Booth, S. H. E. Knight, E. J. Highwood, D. J. Frame, M. R. Allen, and D. P. Rowell, 2011: Sensitivity of twentieth-century rainfall to sulfate aerosol and CO2 forcing. J. Climate, 24, 4999–5014.
Ashfaq, M., C. B. Skinner, and N. S. Diffenbaugh, 2010: Influence of SST biases on future climate change projections. Climate Dyn., 36, 1303–1319, doi:10.1007/s00382-010-0875-2.
Biasutti, M., and A. Giannini, 2006: Robust Sahel drying in response to late 20th century forcings. Geophys. Res. Lett., 33, L11706, doi:10.1029/2006GL026067.
Biasutti, M., and A. H. Sobel, 2009: Delayed Sahel rainfall and global seasonal cycle in a warmer climate. Geophys. Res. Lett., 36, L23707, doi:10.1029/2009GL041303.
Biasutti, M., A. H. Sobel, and S. J. Camargo, 2009: The role of the Sahara low in summertime Sahel rainfall variability and change in the CMIP3 models. J. Climate, 22, 5755–5771.
Burpee, R. W., 1972: The origin and structure of easterly waves in the lower troposphere of North Africa. J. Atmos. Sci., 29, 77–90.
Christensen, J. H., and Coauthors, 2007: Regional climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 847–940.
Cook, K. H., 1999: Generation of the African easterly jet and its role in determining West African precipitation. J. Climate, 12, 1165–1184.
Cook, K. H., and E. Vizy, 2006: Coupled model simulations of the West African monsoon system: Twentieth- and twenty-first-century simulations. J. Climate, 19, 3681–3703.
Deser, C., and A. S. Phillips, 2009: Atmospheric circulation trends, 1950–2000: The relative roles of sea surface temperature forcing and direct atmospheric radiative forcing. J. Climate, 22, 396–413.
Diffenbaugh, N. S., 2005: Sensitivity of extreme climate events to CO2-induced biophysical atmosphere–vegetation feedbacks in the western United Sates. Geophys. Res. Lett., 32, L07702, doi:10.1029/2004GL022184.
Folland, C. K., T. N. Palmer, and D. E. Parker, 1986: Sahel rainfall and worldwide sea temperatures, 1901-85. Nature, 320, 602–607.
Giannini, A., R. Saravanan, and P. Chang, 2003: Oceanic forcing of Sahel rainfall on interannual to interdecadal time scales. Science, 302, 1027–1030.
Grist, J. P., and S. E. Nicholson, 2001: A study of the dynamic factors influencing the rainfall variability in the West African Sahel. J. Climate, 14, 1337–1359.
Haarsma, R. J., F. M. Selten, S. L. Weber, and M. Kliphuis, 2005: Sahel rainfall variability and response to greenhouse warming. Geophys. Res. Lett., 32, L17702, doi:10.1029/2005GL023232.
Hagos, S. M., and K. H. Cook, 2008: Ocean warming and late-twentieth-century Sahel drought and recovery. J. Climate, 21, 3797–3814.
Held, I. M., T. L. Delworth, J. Lu, K. L. Findell, and T. R. Knutson, 2005: Simulation of Sahel drought in the 20th and 21st centuries. Proc. Natl. Acad. Sci. USA, 102, 17 891–17 896.
Hoerling, M., J. Hurrell, J. Eischeid, and A. Phillips, 2006: Detection and attribution of twentieth-century Northern and Southern African rainfall change. J. Climate, 19, 3989–4008.
Kanamitsu, M., W. Ebisuzaki, J. Woollen, S.-K. Yang, J. J. Hnilo, M. Fiornio, and G. L. Potter, 2002: NCEP-DOE AMIP-2 Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 1631–1644.
Kumar, A., and M. P. Hoerling, 1998: On the specification of regional SSTs in AGCM simulations. J. Geophys. Res., 103, 8901–8907.
Lebel, T., A. Diedhiou, and H. Laurent, 2003: Seasonal cycle and interannual variability of the Sahelian rainfall at hydrological scales. J. Geophys. Res., 108, 8389, doi:10.1029/2001JD001580.
Leroux, M., 2001: The Meteorology and Climate of Tropical Africa. Springer, 548 pp.
Lu, J., 2009: The dynamics of the Indian Ocean sea surface temperature forcing of Sahel drought. Climate Dyn., 33, 445–460.
Lu, J., and T. L. Delworth, 2005: Oceanic forcing of the late 20th century Sahel drought. Geophys. Res. Lett., 32, L22706, doi:10.1029/2005GL023316.
Maynard, K., J.-F. Royer, and F. Chauvin, 2002: Impact of greenhouse warming on the West African summer monsoon. Climate Dyn., 19, 499–514.
Meehl, G. A., and Coauthors, 2006a: Climate change projections for the twenty-first century and climate change commitment in the CCSM3. J. Climate, 19, 2597–2616.
Meehl, G. A., J. M. Arblaster, D. M. Lawrene, A. Seth, E. K. Schneider, B. P. Kirtman, and D. Min, 2006b: Monsoon regimes in the CCSM3. J. Climate, 19, 2482–2495.
Meehl, G. A., C. Covey, T. Delworth, M. Latif, B. McAvaney, J. F. B. Mitchell, R. J. Stougger, and K. E. Taylor, 2007a: The WCPP CMIP3 multimodel dataset: A new era in climate change research. Bull. Amer. Meteor. Soc., 88, 1383–1394.
Meehl, G. A., C. Covey, and Coauthors, 2007b: Global climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 747–845.
Mohr, K. I., and C. D. Thorncroft, 2006: Intense convective systems in West Africa and their relationship to the African easterly jet. Quart. J. Roy. Meteor. Soc., 132, 163–176.
Newell, R. E., and J. W. Kidson, 1984: African mean wind changes between Sahelian wet and dry periods. Int. J. Climatol., 4, 27–33.
Nicholson, S. E., 2001: Climatic and environmental change in Africa during the last two centuries. Climate Res., 17, 123–144.
Paeth, H., and A. Hense, 2004: SST versus climate change signals in West African rainfall: 20th-century variations and future projections. Climatic Change, 65, 179–208.
Philippon, N., F. J. Doblas-Reyes, and P. M. Ruti, 2010: Skill, reproducibility and potential predictability of the West African monsoon in coupled GCMs. Climate Dyn., 35, 53–74.
Pitman, A. J., and M. Zhao, 2000: The relative impact of observed change in land cover and carbon dioxide as simulated by a climate model. Geophys. Res. Lett., 27, 1267–1270.
Rotstayn, L., and U. Lohmann, 2002: Tropical rainfall trends and the indirect aerosol effect. J. Climate, 15, 2103–2116.
Thorncroft, C. D., and M. Blackburn, 1999: Maintenance of the African easterly jet. Quart. J. Roy. Meteor. Soc., 125, 763–786.
Zeng, N., J. D. Neelin, K.-M. Lau, and C. J. Tucker, 1999: Enhancement of interdecadal climate variability in the Sahel by vegetation interaction. Science, 286, 1537–1540.
