1. Introduction
The Sahel summer rainfall exhibits a sharp north–south gradient with an east–west belt-like distribution, corresponding to a radical shift of climate characteristics—from humid rain forest to desert in just a few degrees from south to north. Changes in the hydrological balance across the Sahel can have profound impacts on local livelihoods, economic development, and social stability, such as during the long-lasting drought in the 1970s and 1980s (Dai et al. 2004; Nicholson et al. 2012; Brönnimann et al. 2015). Oceanic forcings (Folland et al. 1986; Giannini et al. 2003; Rowell 2003; Lu and Delworth 2005; Chiang and Friedman 2012), anthropogenic aerosols (Kawase et al. 2010; Dong et al. 2014), land surface feedbacks (Zeng et al. 1999), and remote monsoon heating (He et al. 2017) are suggested to be the drivers of the long-lasting Sahel drought. The sensitive response of Sahel rainfall to climate change receives extensive attention (Rodríguez-Fonseca et al. 2015).
Following the drought conditions during the 1970s and 1980s, the summer Sahel rainfall has shown an increase since the 1990s (Li et al. 2012; Maidment et al. 2015). Several mechanisms have been presented for the recent Sahel rainfall recovery, including the radiative impact of greenhouse gases (Dong and Sutton 2015), anthropogenic sea surface temperature (SST) patterns (Maidment et al. 2015), anthropogenic warming of the Mediterranean Sea (Park et al. 2016), and the amplified warming over the Sahara due to greenhouse warming by water vapor (Evan et al. 2015; Zhou et al. 2016). All of these mechanisms imply that Sahel rainfall would continue to increase under a global warming scenario.
However, the future Sahel rainfall projections made by current general circulation models (GCMs) show a significant uncertainty (Cook and Vizy 2006; Caminade and Terray 2010; Biasutti 2013; Monerie et al. 2017). Caminade and Terray (2010) showed that the projections by phase 3 of the Coupled Model Intercomparison Project (CMIP3) models for Sahel rainfall change in response to global warming are highly uncertain. In comparison with CMIP3, phase 5 of CMIP (CMIP5) shows a tendency for the agreement among future projections, with slightly wetter conditions over the Sahel (Biasutti 2013; Rodríguez-Fonseca et al. 2015; Park et al. 2016). Nevertheless, the large spread of individual model projections still exists among CMIP5 models, and some models even project future drying in the Sahel (Monerie et al. 2017).
As mentioned above, Sahel rainfall variability during the twentieth century has been modulated by worldwide SST anomalies, including the tropical Atlantic, tropical Pacific, and Indian Oceans, and the Mediterranean Sea (Folland et al. 1986; Janicot et al. 1996; Rowell 2001, 2003). However, most of these observed SST–Sahel rainfall relationships do not work well in explaining the projected Sahel rainfall change in the twenty-first century (Biasutti et al. 2008; Caminade and Terray 2010), except that the Mediterranean SST warming relates to more rainfall in the Sahel in the future projection (Park et al. 2016). Furthermore, Held et al. (2005) proposed a drier Sahel in the future due primarily to the increase of greenhouse gases based on the results of one climate model that captures the main aspects of the twentieth-century rainfall record in the Sahel. On the other hand, some studies indicated that under a global warming scenario, Sahel rainfall will increase through the direct radiative forcing of CO2 increase, but decrease through the response of global SST warming (Biasutti 2013; Gaetani et al. 2017). All these results of previous studies suggest the challenge of skillful Sahel rainfall projections.
Another approach to address the model uncertainty in future Sahel rainfall projections is to assess the impact of limitations in present or historical climate performance. Some works have focused on the model’s historical simulation in attempt to find the relationship between historical simulation and future projection, implying that the model with a higher ability in simulating the present climate can project future changes more reliably (e.g., Knutti et al. 2010; Monerie et al. 2017). Specifically for the Sahel region, however, it was revealed that no clear relationship exists between local historical rainfall simulation errors and future rainfall changes (Monerie et al. 2017). Currently, there seems to be few studies on the possible connections between the uncertainty in Sahel rainfall changes and present-day model climatology errors over remote regions. Therefore, the purpose of this work is to present the difference in Sahel rainfall projections between the CMIP5 models, and investigate the possible connection between the uncertainty in Sahel rainfall projections and present rainfall simulation biases at the global scale. If there exists this kind of future–present relationship, it would potentially be useful to calibrate future projections of Sahel rainfall change. Li et al. (2016a,b) have provided an effective calibration method based on a relationship between future projections and present simulations, which applied an “observational constraint” of present equatorial western Pacific precipitation to calibrate the projections of tropical Pacific climate change. In virtue of this calibration method, a more reliable projection of the Sahel rainfall change would be anticipated, as the present model biases can be estimated much more reliably by observational or reanalysis data.
The rest part of this article is organized as follows. Section 2 briefly describes the models, data, and methods. In section 3, we perform a comparison between two sets of representative five-member ensemble results to present the difference between Sahel rainfall projections. Section 4 studies the relationship between Sahel rainfall changes and historical simulation biases among the models and the accordingly calibrated results based on this relationship. In addition, we also discuss the possible role of convective parameterization in inducing the present–future relationship. Section 5 provides concluding remarks.
2. Data
This study used 34 CMIP5 CGCMs (Taylor et al. 2012) for the historical and representative concentration pathway 8.5 (RCP8.5) experiments. The monthly mean variables include rainfall, evaporation (unavailable for the model CMCC-CMS), horizontal wind, and surface temperature, which were interpolated into a common 2.5° × 2.5° horizontal resolution by bilinear interpolation before our analysis. The historical and RCP8.5 simulation included 86 and 63 ensemble runs, respectively. Some models provide multiple ensemble members, and these members were averaged prior to intermodel comparisons. For calculation of standard deviations, correlation coefficients, or regressed coefficients, we first calculated them for each ensemble independently, and then averaged over multiple ensemble members. Similar results were achieved by using only one ensemble member for each model (figures not shown). The period from 1979 to 2005 was adopted as a baseline of present-day climatology, and the period from 2050 to 2099 was used as the future-projection climatology. We also used the period from 1900 to 2005 as present-day climatology, and the main results were similar. In this paper, we only showed the results by using the 1979–2005 period as present-day climatology, to match the period of observed data used. Table 1 lists the specific ensemble members for each model and experiment. More details on the models and experiments are presented on the website (http://cmip-pcmdi.llnl.gov/cmip5/availability.html). The projected future changes were evaluated by comparing the mean state during 2050–99 in the RCP8.5 experiment and the mean state during 1979–2005 in the historical experiment. As a measure of the model uncertainty, the intermodel spread of rainfall changes was calculated by the standard deviation across all the models.
Details of 34 CMIP5 models used in this study.
Furthermore, to assess historical simulation bias in mean climate, the observed monthly rainfall data from 1979 to 2005 were obtained from the Global Precipitation Climatology Project (GPCP v2.2; Adler et al. 2003) with the horizontal resolution 2.5° in longitude and latitude. Some other rainfall datasets with higher resolution, including GPCC (Schneider et al. 2011) and the Precipitation Reconstruction over Land (PREC/L; Chen et al. 2002) rainfall data with a 0.5° × 0.5° horizontal resolution, were also examined and showed similar results to GPCP. The mean state during 1979–2005 in the historical experiment was compared with that during 1979–2005 in observation to evaluate the historical simulation bias. Our analyses focused on boreal summer [July–September (JAS)], when the Sahel has abundant rainfall over the year.
3. Difference between wet and dry Sahel rainfall projections
Before looking into the difference between Sahel rainfall projections, we first present the multimodel ensemble (MME) mean of North African rainfall changes in Fig. 1a. The MME result mainly shows a wet projection in the central–eastern Sahel but slight dry in the western Sahel along the Atlantic coast, which is consistent with many previous studies (Fontaine et al. 2011; Diallo et al. 2012; Monerie et al. 2012, 2013). Additionally, the Red Sea, Arabian Plateau, and the offshore Somali coast also project rainfall increase.
Projected summer (JAS) mean rainfall change (mm day−1) under RCP8.5 (2050–99 mean minus 1979–2005 mean) over the Sahel and intermodel spread among CMIP5 models. (a) MME mean of projected rainfall changes over North Africa. (b) Intermodel standard deviation of JAS mean rainfall changes (color shading) across the 34 CMIP5 models. Contours are observed JAS mean climatological rainfall for the period 1979–2005. Contour interval is 2 mm day−1. (c) Area-averaged rainfall changes over the Sahel [gray rectangle with large intermodel diversity in (b); 10°–20°N, 15°W–40°E] for individual models. Green bars represent the five wetting models that project the wettest conditions (Wet_M), and sienna bars represent the five driest models (Dry_M).
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
Although it serves as a general measure to provide a robust climate change signal, the MME result has inherent limitations in consideration of substantial variations between individual models. Consequently, the intercomparison between different models and an understanding of model projection uncertainties are important and necessary. The intermodel spread of rainfall changes projected by individual CMIP5 models is calculated to illustrate the model uncertainty, as shown in Fig. 1b. The spread is large over the whole Sahel, especially in the western and central area, consistent with the regions with large climatological rainfall. The averaged intermodel standard deviation over the Sahel region, which is defined as the region of 10°–20°N, 15°W–40°E, is 0.83 mm day−1. This intermodel spread is larger than the MME mean of projected rainfall changes, which is only 0.33 mm day−1 (Fig. 1a). Different CMIP5 models disagree even on the sign of the future Sahel rainfall trend (Fig. 1c). Other areas like the Red Sea, Arabian Plateau, and the offshore Somali coast, where an increased rainfall is projected (Fig. 1a), also exhibit large spreads (Fig. 1b).
To highlight the differences in model projections, we chose five models that project the wettest conditions (hereafter called “wetting models”; MIROC-ESM-CHEM, MIROC-ESM, BNU-ESM, MIROC5, and FGOALS-g2), and five models that project the driest conditions (hereafter called “drying models”; CSIRO, GFDL-ESM2M, FIO-ESM, GISS-E2-R, and GISS-E2-H), and compare these two sets of 5-member ensemble results. As expected, the wetting models project substantial precipitation increase across the entire Sahel, while the maximum rainfall increase over the central Sahel is more than 2.5 mm day−1 (Fig. 2). In contrast, the drying models project a precipitation decrease, exhibiting an approximately opposite change pattern, with a relatively southward distribution. The precipitation increase in the Sahel averaged for the wetting models is 1.67 mm day−1, stronger than the precipitation decrease over the drying models (0.32 mm day−1), which is consistent with that shown in Fig. 1c. The models also project opposite-signed rainfall changes on the Atlantic coast of the Sahel; that is, the wetting models show a rainfall decrease over the Atlantic coast between 4° and 10°N, while the drying models show an increase in Atlantic coast rainfall. In addition, the wetting models project a strong increase in the Red Sea and offshore the Somali coast, but the drying models have very weak changes. We also examined another method of sampling models by using a weighted model mean (weighted by the rainfall change shown in Fig. 1c for each model) and achieved similar results (not shown).
Composite of rainfall change (mm day−1) for (a) wetting models and (b) drying models. Gray rectangle indicates the location of defined Sahel region as in Fig. 1.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
To further assess the meridional variability of rainfall change in the Sahel, we calculate the zonal mean rainfall between 15°W and 40°E for the wetting and drying models (Fig. 3). The summer rainfall over North Africa is primarily characterized by a zonal band centered at 5°–10°N (Fig. 1b). Both the wetting and drying models basically capture the meridional distribution of rainfall, and the wetting models overestimate the North African rainfall. However, it will be mentioned in the next section that the rainfall bias along the North African rainband is much weaker than the biases in some other remote regions. Figure 3 also shows that the projected rainfall changes look more like a northward expansion or southward contraction of the rainband at its north margin, in comparison with the historical simulation. In particular, the wetting models project more rainfall around the north margin of the rainband (Fig. 3a). This relationship between the Sahel rainfall and North African rainband shift is consistent with previous studies (e.g., Sheen et al. 2017). Furthermore, the projected rainfall changes to the south of the main rainband are weak for both the wetting and drying models. This weak change along the southern margin is because the rainfall changes at the Guinea Coast tend to cancel those out farther to the east (Fig. 2).
Zonal mean rainfall (mm day−1) between 15°W and 40°E for the GPCP, historical experiment, and RCP8.5 experiment for (a) wetting and (b) drying models. The latitudes for 2 mm day−1 of rainfall are denoted as R and H for RCP8.5 and historical, respectively. The latitude change of 2 mm day−1 is 2.65° and −0.54° in (a) and (b).
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
Such northward expansion or southward contraction of the North African rainband is of great importance for agriculture, migration, and social security in the Sahel. Simply considering 2 mm day−1 of rainfall as the boundary of the rainband, we locate the latitude of 2 mm day−1 rainfall from the zonal mean rainfall by a linear interpolation (Fig. 3) to quantify the extent of the rainband movement. Comparing the latitude difference between the historical and RCP8.5 experiments, this 2 mm day−1 line represents 2.65° of northward expansion of the future change for the wetting models and 0.54° of southward contraction for the drying models.
Understanding future water budget conditions in such a vulnerable region to drought is of great concern. Besides precipitation, evaporation is also one of the dominant factors in the water budget. For the future projections of evaporation, significant differences of future change over the Sahel also exist between the wetting and drying models (Figs. 4a,b). The wetting models project strong evaporation increase while the drying models mainly project a decrease in evaporation with the exception of a weak increase in the south margin of the Sahel. Since more evaporation is expected for the wetting models in the future, more precipitation alone does not guarantee increased availability of water resources for a local region, and vice versa for the drying models. Therefore, we further evaluate the projected changes in precipitation minus evaporation (P − E change; Figs. 4c,d), an element more closely related to water resources. The projected P − E changes behave similarly with precipitation changes in terms of spatial distribution (Fig. 2). The P − E is increased in the Sahel averaged for the wetting models (1.15 mm day−1) and decreased for the drying models (−0.25 mm day−1). They are both weaker than the precipitation changes because of the offset by evaporation.
(a),(b) Evaporation change (mm day−1), (c),(d) P − E change (mm day−1), and (e),(f) surface temperature change (°C) for wetting and drying models.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
To further illustrate the connection between precipitation and evaporation changes, we show the projection differences in surface temperature in Figs. 4e and 4f, which is crucial for water budget. Both sets of models project a temperature increase, but the wetting models project a much lower increase than the drying models. The central–eastern Sahel is the area with minimum increased temperature in the wetting models, where the maximum evaporation increase is projected. The different directions between evaporation and temperature change suggest that temperature is not the direct driver of evaporation changes. The temperature changes, on the other hand, may in turn affect the rainfall changes: the meridional temperature gradient is enhanced over the Sahel and Sahara for the wetting models and thus favorable to more rainfall at the Sahel (Burpee 1972; Thorncroft and Blackburn 1999; Cook 1999; Sylla et al. 2010), but the gradient change is weak for the drying models, consistent with the relatively weak change in rainfall.
4. Future change–present bias relationship
a. Historical simulation biases in South Asia and the western North Pacific associated with Sahel rainfall changes
In this section, we investigate whether the future model responses are linked to present-day climatological rainfall biases. The historical rainfall simulation biases are calculated by subtracting the mean state of observation. Figure 5a compares the difference of historical rainfall bias between the wetting and drying models and shows no significant anomaly over the Sahel. Thus, there is no clear relationship between the historical rainfall bias and future rainfall change in the Sahel region, consistent with the results in Monerie et al. (2017). In contrast, there is an apparent positive anomaly in South Asia (5°–30°N, 60°–85°E) and a negative anomaly in the western North Pacific (WNP; 10°–20°N, 130°E–160°W), suggesting that the models with such anomalies tend to project a strong enhancement of Sahel rainfall. The rainfall biases between these two regions are negatively correlated among the models, with the correlation coefficient r being −0.37. The correlation coefficient between the historical rainfall biases in South Asia (WNP) and the Sahel rainfall changes among all models is 0.68 (−0.66). Therefore, we define the model historical bias difference between South Asia and the WNP (bias in South Asia minus that in the WNP) to highlight the relationship between the historical model biases and Sahel rainfall changes. Figure 5b shows the scatterplot of historical bias differences versus Sahel rainfall changes for individual models. A close relationship between them is illustrated and the corresponding correlation coefficient is 0.81 among the models, which is statistically significant at the 99% confidence level by using a two-tailed t test. In addition, a bootstrapping resample analysis based on these models indicates the 90% confidence interval of the correlation coefficient is [0.59, 0.90]. The relationship is also significant after removing some outlier models: the intermodel correlation coefficient is 0.58 after removing the MIROC-ESM, MIROC5, and CSIRO. These values demonstrate the significance of the relationship shown in Fig. 5b. An intimate relationship is also revealed between the historical model bias and the northward expansion or southward contraction of the future rainband change. The correlation coefficient between historical bias differences and future projections of the northern 2 mm day−1 rainfall line shift is 0.78 among the models (Fig. 5c). This means that the models that have larger positive bias differences tend to project more northward expansion of the rainband in the future.
Relationship between the historical rainfall simulation bias and Sahel rainfall change. (a) Difference of simulated historical rainfall bias (mm day−1) between the wetting (Wet_M) and drying (Dry_M) models. Gray rectangles show the regions with significant anomalies, including South Asia (5°–30°N, 60°–85°E) and the WNP (10°–20°N, 130°E–160°W). Stippling indicates the difference exceeding the 95% confidence level using a two-tailed t test. (b) Scatterplot of simulated historical rainfall bias differences vs projected rainfall changes over the Sahel for individual models. (c) As in (b), but the ordinate represents the future changes of zonal mean latitude of 2 mm day−1 rainfall (defined as in Fig. 3, but for each model). In (b) and (c), black markers represent individual models, the red marker represents calibrated change, and the green marker represents MME mean change. The solid line represents the linear regression. The number at the upper-left corner is the correlation coefficient.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
There is also a close relationship between the projected P − E changes and the historical simulation biases (Fig. 6). The correlation coefficient between the P − E changes and historical bias differences is 0.80 among the models (the model CMCC-CMS is not used here because of the missing evaporation data). The 90% confidence interval of the correlation coefficient is [0.58, 0.89] according to the bootstrapping resample analysis. In addition, compared with the rainfall change (Fig. 5b), the projected P − E change shows a smaller spread among models (0.48 mm day−1), which is caused by the offset contribution from the projected evaporation (Figs. 4a,b). Furthermore, it was found that the historical biases are also closely related to the projections of evaporation and surface temperature in the Sahel, with significant correlation coefficients of 0.70 and −0.57 among the models, respectively (figure not shown). The models with more positive bias differences tend to project stronger evaporation and lower temperature increase in the future, and vice versa. These relationships imply that the historical simulation bias can influence future projection of the water budget in the Sahel, in addition to the precipitation change.
Relationship between the simulated historical rainfall bias differences and Sahel P − E changes among 33 CMIP5 GCMs. The black markers represent individual models, the red marker represents calibrated change, and the green marker represents MME mean change. The solid line represents the linear regression. The number at the upper-left corner is the correlation coefficient.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
b. Calibrated future change
Based on this present bias–future projection relationship, an “observational constraint” of precipitation in these two regions is applied to calibrate the projections of Sahel rainfall change, following Li et al. (2016a,b). The practical calibration procedure is carried out as follows: We first regress the future rainfall changes onto the historical bias differences among the models. The regression coefficients for each spatial grid are obtained from the above regression. We further calculate the rainfall change offset for each model by its historical bias difference multiplied by the regression coefficient. Future change of Sahel rainfall can then be obtained by the model ensemble after removing this change offset.
Figure 7a presents the spatial distribution of accordingly calibrated rainfall change, and suggests a rainfall increase across the entire Sahel. The increase over the central part is maximum across the Sahel, which exceeds 1 mm day−1. In addition, the Red Sea, Arabian Plateau, and the offshore Somali coast also have a strong rainfall increase in the future. Figure 7b shows the spatial distribution of regression coefficients between future rainfall changes and the historical bias differences. As expected, the significant regression coefficients exceed the 95% confidence level over the whole Sahel.
(a) Calibrated rainfall change (mm day−1) over the Sahel. (b) Spatial distribution of regression coefficient between future rainfall changes and the historical rainfall bias differences across the CMIP5 models for each grid point. Stippling indicates the relationship exceeding the 95% confidence level using a two-tailed t test.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
The calibrated rainfall change differs from the MME result in several aspects (Figs. 1a and 7a). First, the calibrated Sahel rainfall change is stronger than the MME one. The rainfall increase averaged over the Sahel is 0.69 mm day−1, much higher than the MME result (0.33 mm day−1; Fig. 5b). Similarly, the calibrated northern line for 2 mm day−1 of rainfall is shifted northward by 1.18°, much greater than the MME result (0.59°; Fig. 5c). The calibrated northward shift of this line is equivalent to an area increase of about 78 million hectares. Second, the MME result shows a slightly dry projection along the Atlantic coast, but there are no dry conditions in the calibrated one (Figs. 1a and 7a). Third, the calibrated rainfall increase is also greater in the Red Sea, Arabian Plateau, and the offshore Somali coast. The calibrated rainfall increase over the offshore Somali coast is even more than 1.5 mm day−1.
After calibrating the projected P − E according to this future projection–present bias relationship, we obtain an increase of 0.46 mm day−1 of the P − E averaged over the Sahel. The calibrated result is much larger than the MME result (0.21 mm day−1; Fig. 6). The P − E increases across the entire Sahel become stronger after calibration, and the maximum increase over the central Sahel exceeds 0.5 mm day−1 (Fig. 8). The significant regression coefficients behave similarly to those for precipitation (Fig. 7b) and give rise to the calibration reliability over these significant regions. The P − E increase suggests more water resources available in the twenty-first century, even under the circumstances of global warming. Besides the Sahel, the Red Sea, Arabian Plateau, and the area offshore the Somali coast also get a much stronger P − E increase after calibration. The P − E change will be crucial for the ecosystems in these regions. For instance, the P − E change may affect salinity and upwelling in the offshore Somali coast (Schott et al. 2009; Praveen et al. 2016), which is one of most productive ecosystem regions in the World Ocean.
P − E change (mm day−1) over the Sahel for (a) MME mean P − E change and (b) calibrated P − E change. (c) Spatial distribution of regression coefficient between future P − E changes and the historical rainfall bias differences across the CMIP5 models for each grid point. Stippling indicates the relationship exceeding the 95% confidence level using a two-tailed t test.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
c. Projection of rainfall and circulation changes associated with the historical biases
In this subsection, we further examine the future projection–present bias relationship, from the view of the models’ simulation bias of historical rainfall. Figure 9a shows the simulated historical rainfall bias differences (bias in South Asia minus that in the WNP) for individual models. Nearly two-thirds of the models have a negative bias difference, and the amplitude of the negative bias difference is larger than that of the positive bias difference. Figure 9b shows the projected rainfall changes regressed onto this historical rainfall bias difference. A significant wetting condition of the future rainfall projections is found over the Sahel. It suggests that the models with positive bias difference tend to project a wetting Sahel in the future, and vice versa. This result confirms the above present bias–future projection relationship.
(a) Simulated historical rainfall bias differences (bias in South Asia minus that in the WNP; mm day−1) for individual models. (b) Rainfall change (RCP8.5 minus historical) regressed onto the historical bias difference index [as in (a)] among the models. Shading indicates the regression coefficient and stippling indicates the relationship exceeding the 95% confidence level using a two-tailed t test.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
Figure 10 shows the projected zonal wind changes (RCP8.5 vs historical) at 150, 600, and 850 hPa regressed onto the bias difference index. At 150 hPa, there are significant easterly anomalies over the tropical Atlantic and Africa. These anomalies indicate an enhancement of the upper-tropospheric tropical easterly jet (TEJ), which is located around 5°–10°N, and can enhance the Sahel rainfall via reinforcing the upper-level divergence (Pattanaik and Satyan 2000; Grist and Nicholson 2001; Nicholson 2009). In the middle troposphere, a dipole pattern is shown over North Africa. This dipole pattern can be considered as a northward shift of the African easterly jet (AEJ), which is located climatologically around 15°N. In addition, significant 850-hPa westerly anomalies over the tropical North Atlantic and North Africa indicate a strengthening of the West African monsoon westerlies. Both the northward-shifted AEJ and the stronger lower-tropospheric monsoon westerlies are favorable to the increase of easterly vertical zonal wind shear, promoting the moisture convergence, easterly wave activity, and precipitation in the Sahel (Cook 1999; Thorncroft and Blackburn 1999; Nicholson 2009; Dong et al. 2014). These results suggest that projected Sahel rainfall changes, in association with the historical rainfall biases, are dynamically consistent.
Projected zonal wind change (difference between RCP8.5 and historical) at (a) 150, (b) 600, and (c) 850 hPa regressed onto the historical bias difference index (as in Fig. 9a) among the models. Positive values represent westerly anomalies. Shading indicates the regression coefficient and stippling indicates the relationship exceeding the 95% confidence level using a two-tailed t test.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
d. Discussion on the possible mechanism
All the above-mentioned results demonstrate the connection between the future Sahel rainfall projections and present rainfall simulation biases in South Asia and the western North Pacific. There would be two approaches to establish this connection: one is that the present rainfall biases affect the future rainfall projections, and the other is that some other factors affect both the present biases and future projections. It is hard to determine the physical processes through which the present rainfall biases affect the future rainfall projections because of the extensive temporal and spatial differences between these two ends. Regarding the second approach, two factors of models may be the candidates: convective parameterization and resolution. The convective parameterization is widely known as a factor responsible for model uncertainty and diversity. In addition, the resolution may also remarkably affect the rainfall pattern at the regional or global scales (e.g., Goswami and Goswami 2017; Huang et al. 2018). Therefore, in the following we examine the possible impacts of these two factors, respectively.
Among various schemes of convective parameterization, the most widely used ones are the Arakawa–Schubert (Arakawa and Schubert 1974) scheme, the Zhang and McFarlane (1995) scheme, the Gregory and Rowntree (1990) scheme, and the Tiedtke (1989) scheme (Table 1). Out of the total 34 models, there are 26 models using these four schemes. It should be noted that the models mixing the Tiedtke (1989) scheme and the Emanuel (1991, 1993) scheme (i.e., models IPSL-CM5A-LR and IPSL-CM5B-LR) are simply categorized into the Tiedtke (1989) scheme. Then, these 26 models are divided into two groups: group A includes the models of the Arakawa–Schubert (Arakawa and Schubert 1974) scheme and the Zhang and McFarlane (1995) scheme, and group B includes the models of the Gregory and Rowntree (1990) scheme and the Tiedtke (1989) scheme. These two groups are found to closely relate to both the historical rainfall bias difference and future Sahel rainfall projection (Figs. 11a,b). The group A models clearly tend to show positive historical rainfall bias difference and project stronger enhancement of Sahel rainfall in the future. By contrast, the group B models show negative historical rainfall difference and project weaker change of future Sahel rainfall. This result suggests that convective parameterization is a crucial factor affecting both the historical rainfall bias difference and future Sahel rainfall projection, and thus induces the historical bias–future projection relationship. Furthermore, we also found that convective parameterization can also contribute to the out-of-phase relationship between South Asia and WNP rainfall biases: the group A models overestimate historical rainfall simulation in South Asia, while the group B models overestimate rainfall simulation in the WNP (Figs. 11c,d).
(a) Simulated historical rainfall bias differences (bias in South Asia minus bias in the WNP) for individual models. (b) Projected rainfall changes over the Sahel for individual models. (c) Simulated historical rainfall bias in South Asia. (d) Simulated historical rainfall bias in the WNP. Unit: mm day−1. Blue bars represent the models using the Arakawa–Schubert (Arakawa and Schubert 1974) scheme and the Zhang and McFarlane (1995) scheme, and sienna bars represent the models using the Gregory and Rowntree (1990) scheme and the Tiedtke (1989) scheme.
Citation: Journal of Climate 32, 4; 10.1175/JCLI-D-17-0846.1
The other candidate, that is, the resolution of models, however, is not found to have a close relationship to both the historical rainfall bias difference and future Sahel rainfall projection. Following Goswami and Goswami (2017), we adopted the resolution of the atmospheric models to represent the resolution of coupled models, and defined the resolution index as the product of latitude and longitude grid numbers. We have not found that the above-defined model resolution is closely related to both the historical rainfall bias difference and future Sahel rainfall projection (figure not shown), and the resolution index is weakly correlated to both the historical difference index and future Sahel rainfall projection index among the models, with the correlation coefficients being −0.13 and −0.15, respectively.
5. Summary
This study investigates the differences in Sahel summer rainfall projections by comparing the RCP8.5 experiment with the historical experiment of 34 CMIP5 GCMs. Large uncertainties exist in the projections of Sahel rainfall among all models. The averaged intermodel standard deviation of rainfall changes over the Sahel region (0.83 mm day−1) is much stronger than the MME mean of projected rainfall changes (0.33 mm day−1). Two sets of representative five-member ensemble results (wetting and drying models) present different rainfall change directions over the Sahel. The wetting models project a rainfall increase and northward expansion of the rainband at its north margin. In contrast, the drying models project a decrease and southward contraction. Besides, projections of evaporation change, P − E change, and surface temperature change also show large differences between the wetting and drying models. The wetting models project evaporation and P − E increase, while the drying models project evaporation and P − E decrease in the future. Meanwhile, the wetting and drying models both project surface temperature increase, but the increase in the drying models is stronger than that in the wetting models.
Future projections of Sahel rainfall are closely related to the model present simulation biases of remote regions, that is, South Asia and the WNP. The models with an overestimated precipitation in South Asia tend to project wet conditions in the Sahel, while the models overestimating precipitation in the WNP project dry conditions in the Sahel. The correlation coefficient between Sahel rainfall changes and the historical bias differences (bias in South Asia minus that in the WNP) is 0.81 among the models. In addition, evaporation and surface temperature changes also have a close relationship with the historical simulation biases. The models with more positive historical bias differences tend to project more evaporation and lower surface temperature increase in the Sahel.
These present bias–future projection relationships are used to calibrate Sahel rainfall projections. The accordingly calibrated results show a substantial increase in both the rainfall and P − E projections (0.69 and 0.46 mm day−1) in the Sahel, especially in the central part. These are much stronger than the MME-mean projections (0.33 and 0.21 mm day−1). Our findings of the future–present bias relationship can lead to more reliable projections of Sahel rainfall, since the model biases can be estimated much more reliably by observational or reanalysis data. This is encouraging under the circumstances of a large uncertainty in future Sahel rainfall projections shown by current models. This result implies that more attention and efforts should be paid to reducing the model biases in climatological precipitation over the South Asian monsoon region and the WNP to achieve a realistic projection of Sahel rainfall.
Our further analyses suggest that the diversities in both the historical rainfall bias and Sahel rainfall projections are related to the different convective parameterization schemes among models. The models that use the Arakawa–Schubert (Arakawa and Schubert 1974) scheme and the Zhang and McFarlane (1995) scheme overestimate historical rainfall simulation in South Asia, tend to result in a positive historical rainfall bias difference, and project stronger enhancement of Sahel rainfall in the future. By contrast, the models that use the Gregory and Rowntree (1990) scheme and the Tiedtke (1989) scheme overestimate historical rainfall simulation in the WNP, result in a negative historical difference, and project weaker rainfall change or even drier conditions in the Sahel in the future. These results suggest that convective parameterization is a crucial factor in affecting both the historical rainfall bias difference and future Sahel rainfall projection, and thus ties into the historical bias–future projection relationship. Furthermore, it is worth noting that the underlying physical reasons for the impact from different convective schemes in future projections and the physical links between Sahel rainfall and the remote simulation biases remain unclear, and need our further investigations.
Acknowledgments
We thank the four anonymous reviewers for their constructive comments. We also thank Lijuan Li, Hailong Liu, and Xiaocong Wang for their discussion on convective parameterization. This work was supported by the National Natural Science Foundation of China (Grants 41320104007, U1502233, and 41775083).
REFERENCES
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.
Arakawa, A., and W. H. Schubert, 1974: Interaction of a cumulus cloud ensemble with the large-scale environment, Part I. J. Atmos. Sci., 31, 674–701, https://doi.org/10.1175/1520-0469(1974)031<0674:IOACCE>2.0.CO;2.
Betts, A. K., 1986: A new convective adjustment scheme. Part I: Observational and theoretical basis. Quart. J. Roy. Meteor. Soc., 112, 677–691, https://doi.org/10.1002/qj.49711247307.
Biasutti, M., 2013: Forced Sahel rainfall trends in the CMIP5 archive. J. Geophys. Res. Atmos., 118, 1613–1623, https://doi.org/10.1002/jgrd.50206.
Biasutti, M., I. M. Held, A. H. Sobel, and A. Giannini, 2008: SST forcings and Sahel rainfall variability in simulations of the twentieth and twenty-first centuries. J. Climate, 21, 3471–3486, https://doi.org/10.1175/2007JCLI1896.1.
Bougeault, P., 1985: A simple parameterization of the large-scale effects of cumulus convection. Mon. Wea. Rev., 113, 2108–2121, https://doi.org/10.1175/1520-0493(1985)113<2108:ASPOTL>2.0.CO;2.
Brönnimann, S., A. M. Fischer, E. Rozanov, P. Poli, G. P. Compo, and P. D. Sardeshmukh, 2015: Southward shift of the northern tropical belt from 1945 to 1980. Nat. Geosci., 8, 969–974, https://doi.org/10.1038/ngeo2568.
Burpee, R. W., 1972: The origin and structure of easterly waves in the lower troposphere of North Africa. J. Atmos. Sci., 29, 77–90, https://doi.org/10.1175/1520-0469(1972)029<0077:TOASOE>2.0.CO;2.
Caminade, C., and L. Terray, 2010: Twentieth century Sahel rainfall variability as simulated by the ARPEGE AGCM, and future changes. Climate Dyn., 35, 75–94, https://doi.org/10.1007/s00382-009-0545-4.
Chen, M., P. Xie, J. E. Janowiak, and P. A. Arkin, 2002: Global land precipitation: A 50-yr monthly analysis based on gauge observations. J. Hydrometeor., 3, 249–266, https://doi.org/10.1175/1525-7541(2002)003<0249:GLPAYM>2.0.CO;2.
Chiang, J. C. H., and A. R. Friedman, 2012: Extratropical cooling, interhemispheric thermal gradients, and tropical climate change. Annu. Rev. Earth Planet. Sci., 40, 383–412, https://doi.org/10.1146/annurev-earth-042711-105545.
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.
Cook, K. H., and E. K. Vizy, 2006: Coupled model simulations of the West African monsoon system: Twentieth- and twenty-first-century simulations. J. Climate, 19, 3681–3703, https://doi.org/10.1175/JCLI3814.1.
Dai, A., P. J. Lamb, K. E. Trenberth, M. Hulme, P. D. Jones, and P. Xie, 2004: The recent Sahel drought is real. Int. J. Climatol., 24, 1323–1331, https://doi.org/10.1002/joc.1083.
DelGenio, A. D., and M.-S. Yao, 1993: Efficient cumulus parameterization for long-term climate studies: The GISS scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 46, Amer. Meteor. Soc., 181–184, https://doi.org/10.1007/978-1-935704-13-3_18.
Diallo, I., M. B. Sylla, F. Giorgi, A. T. Gaye, and M. Camara, 2012: Multimodel GCM-RCM ensemble-based projections of temperature and precipitation over West Africa for the early 21st century. Int. J. Geophys., 2012, 972896, https://doi.org/10.1155/2012/972896.
Dong, B., and R. T. Sutton, 2015: Dominant role of greenhouse-gas forcing in the recovery of Sahel rainfall. Nat. Climate Change, 5, 757–760, https://doi.org/10.1038/nclimate2664.
Dong, B., R. T. Sutton, E. Highwood, and L. Wilcox, 2014: The impacts of European and Asian anthropogenic sulfur dioxide emissions on Sahel rainfall. J. Climate, 27, 7000–7017, https://doi.org/10.1175/JCLI-D-13-00769.1.
Donner, L. J., 1993: A cumulus parameterization including mass fluxes, vertical momentum dynamics, and mesoscale effects. J. Atmos. Sci., 50, 137–151, https://doi.org/10.1175/1520-0469(1993)050<0889:ACPIMF>2.0.CO;2.
Emanuel, K. A., 1991: A scheme for representing cumulus convection in large-scale models. J. Atmos. Sci., 48, 2313–2335, https://doi.org/10.1175/1520-0469(1991)048<2313:ASFRCC>2.0.CO;2.
Emanuel, K. A., 1993: A cumulus representation based on the episodic mixing model: The importance of mixing and microphysics in predicting humidity. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 46, Amer. Meteor. Soc., 185–194, https://doi.org/10.1007/978-1-935704-13-3_19.
Evan, A. T., C. Flamant, C. Lavaysse, C. Kocha, and A. Saci, 2015: Water vapor–forced greenhouse warming over the Sahara Desert and the recent recovery from the Sahelian drought. J. Climate, 28, 108–123, https://doi.org/10.1175/JCLI-D-14-00039.1.
Folland, C. K., T. N. Palmer, and D. E. Parker, 1986: Sahel rainfall and worldwide sea temperatures, 1901–85. Nature, 320, 602–607, https://doi.org/10.1038/320602a0.
Fontaine, B., P. Roucou, and P.-A. Monerie, 2011: Changes in the African monsoon region at medium-term time horizon using 12 AR4 coupled models under the A1b emissions scenario. Atmos. Sci. Lett., 12, 83–88, https://doi.org/10.1002/asl.321.
Gaetani, M., C. Flamant, S. Bastin, S. Janicot, C. Lavaysse, F. Hourdin, P. Braconnot, and S. Bony, 2017: West African monsoon dynamics and precipitation: The competition between global SST warming and CO2 increase in CMIP5 idealized simulations. Climate Dyn., 48, 1353–1373, https://doi.org/10.1007/s00382-016-3146-z.
Giannini, A., R. Saravanan, and P. Chang, 2003: Oceanic forcing of Sahel rainfall on interannual to interdecadal time scales. Science, 302, 1027–1030, https://doi.org/10.1126/science.1089357.
Goswami, B. B., and B. N. Goswami, 2017: A road map for improving dry-bias in simulating the South Asian monsoon precipitation by climate models. Climate Dyn., 49, 2025–2034, https://doi.org/10.1007/s00382-016-3439-2.
Gregory, D., and P. R. Rowntree, 1990: A mass flux convection scheme with representation of cloud ensemble characteristics and stability-dependent closure. Mon. Wea. Rev., 118, 1483–1506, https://doi.org/10.1175/1520-0493(1990)118<1483:AMFCSW>2.0.CO;2.
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, https://doi.org/10.1175/1520-0442(2001)014<1337:ASOTDF>2.0.CO;2.
He, S., S. Yang, and Z. Li, 2017: Influence of latent heating over the Asian and western Pacific monsoon region on Sahel summer rainfall. Sci. Rep., 7, 7680, https://doi.org/10.1038/s41598-017-07971-6.
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, https://doi.org/10.1073/pnas.0509057102.
Huang, D., P. Yan, J. Zhu, Y. Zhang, X. Kuang, and J. Cheng, 2018: Uncertainty of global summer precipitation in the CMIP5 models: A comparison between high-resolution and low-resolution models. Theor. Appl. Climatol., 132, 55–69, https://doi.org/10.1007/s00704-017-2078-9.
Janicot, S., V. Moron, and B. Fontaine, 1996: Sahel droughts and ENSO dynamics. Geophys. Res. Lett., 23, 515–518, https://doi.org/10.1029/96GL00246.
Kawase, H., M. Abe, Y. Yamada, T. Takemura, T. Yokohata, and T. Nozawa, 2010: Physical mechanism of long-term drying trend over tropical North Africa. Geophys. Res. Lett., 37, L09706, https://doi.org/10.1029/2010GL043038; Corrigendum, 37, L21706, https://doi.org/10.1029/2010GL045530.
Knutti, R., R. Furrer, C. Tebaldi, J. Cermak, and G. A. Meehl, 2010: Challenges in combining projections from multiple climate models. J. Climate, 23, 2739–2758, https://doi.org/10.1175/2009JCLI3361.1.
Li, G., S.-P. Xie, and Y. Du, 2016a: A robust but spurious pattern of climate change in model projections over the tropical Indian Ocean. J. Climate, 29, 5589–5608, https://doi.org/10.1175/JCLI-D-15-0565.1.
Li, G., S.-P. Xie, Y. Du, and Y. Luo, 2016b: Effects of excessive equatorial cold tongue bias on the projections of tropical Pacific climate change. Part I: The warming pattern in CMIP5 multi-model ensemble. Climate Dyn., 47, 3817–3831, https://doi.org/10.1007/s00382-016-3043-5.
Li, H., H. Wang, and Y. Yin, 2012: Interdecadal variation of the West African summer monsoon during 1979–2010 and associated variability. Climate Dyn., 39, 2883–2894, https://doi.org/10.1007/s00382-012-1426-9.
Lu, J., and T. L. Delworth, 2005: Oceanic forcing of the late 20th century Sahel drought. Geophys. Res. Lett., 32, L22706, https://doi.org/10.1029/2005GL022980.
Maidment, R. I., R. P. Allan, and E. Black, 2015: Recent observed and simulated changes in precipitation over Africa. Geophys. Res. Lett., 42, 8155–8164, https://doi.org/10.1002/2015GL065765.
Monerie, P.-A., B. Fontaine, and P. Roucou, 2012: Expected future changes in the African monsoon between 2030 and 2070 using some CMIP3 and CMIP5 models under a medium-low RCP scenario. J. Geophys. Res., 117, D16111, https://doi.org/10.1029/2012JD017510.
Monerie, P.-A., P. Roucou, and B. Fontaine, 2013: Mid-century effects of climate change on African monsoon dynamics using the A1B emission scenario. Int. J. Climatol., 33, 881–896, https://doi.org/10.1002/joc.3476.
Monerie, P.-A., E. Sanchez-Gomez, and J. Boé, 2017: On the range of future Sahel precipitation projections and the selection of a sub-sample of CMIP5 models for impact studies. Climate Dyn., 48, 2751–2770, https://doi.org/10.1007/s00382-016-3236-y.
Nicholson, S. E., 2009: On the factors modulating the intensity of the tropical rainbelt over West Africa. Int. J. Climatol., 29, 673–689, https://doi.org/10.1002/joc.1702.
Nicholson, S. E., A. K. Dezfuli, and D. Klotter, 2012: A two-century precipitation dataset for the continent of Africa. Bull. Amer. Meteor. Soc., 93, 1219–1231, https://doi.org/10.1175/BAMS-D-11-00212.1.
Park, J. Y., J. Bader, and D. Matei, 2016: Anthropogenic Mediterranean warming essential driver for present and future Sahel rainfall. Nat. Climate Change, 6, 941–945, https://doi.org/10.1038/nclimate3065.
Pattanaik, D. R., and V. Satyan, 2000: Fluctuations of tropical easterly jet during contrasting monsoons over India: A GCM study. Meteor. Atmos. Phys., 75, 51–60, https://doi.org/10.1007/s007030070015.
Praveen, V., R. S. Ajayamohan, V. Valsala, and S. Sandeep, 2016: Intensification of upwelling along Oman coast in a warming scenario. Geophys. Res. Lett., 43, 7581–7589, https://doi.org/10.1002/2016GL069638.
Rodríguez-Fonseca, B., and Coauthors, 2015: Variability and predictability of West African droughts: A review on the role of sea surface temperature anomalies. J. Climate, 28, 4034–4060, https://doi.org/10.1175/JCLI-D-14-00130.1.
Rowell, D. P., 2001: Teleconnections between the tropical Pacific and the Sahel. Quart. J. Roy. Meteor. Soc., 127, 1683–1706, https://doi.org/10.1002/qj.49712757512.
Rowell, D. P., 2003: The impact of Mediterranean SSTs on the Sahelian rainfall season. J. Climate, 16, 849–862, https://doi.org/10.1175/1520-0442(2003)016<0849:TIOMSO>2.0.CO;2.
Schneider, U., A. Becker, P. Finger, A. Meyer-Christoffer, B. Rudolf, and M. Ziese, 2011: GPCC full data reanalysis version 6.0 at 0.5°: Monthly land-surface precipitation from rain-gauges built on GTS-based and historic data. GPCC, accessed 9 April 2018, https://doi.org/10.5676/DWD_GPCC/FD_M_V7_050.
Schott, F. A., S.-P. Xie, and J. P. McCreary, 2009: Indian Ocean circulation and climate variability. Rev. Geophys., 47, RG1002, https://doi.org/10.1029/2007RG000245.
Sheen, K. L., D. M. Smith, N. J. Dunstone, R. Eade, D. P. Rowell, and M. Vellinga, 2017: Skilful prediction of Sahel summer rainfall on inter-annual and multi-year timescales. Nat. Commun., 8, 14966, https://doi.org/10.1038/ncomms14966.
Sylla, M. B., E. Coppola, L. Mariotti, F. Giorgi, P. M. Ruti, A. Dell’Aquila, and X. Bi, 2010: Multiyear simulation of the African climate using a regional climate model (RegCM3) with the high resolution ERA-interim reanalysis. Climate Dyn., 35, 231–247, https://doi.org/10.1007/s00382-009-0613-9.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, https://doi.org/10.1175/BAMS-D-11-00094.1.
Thorncroft, C. D., and M. Blackburn, 1999: Maintenance of the African easterly jet. Quart. J. Roy. Meteor. Soc., 125, 763–786, https://doi.org/10.1002/qj.49712555502.
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779–1800, https://doi.org/10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.
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, https://doi.org/10.1126/science.286.5444.1537.
Zhang, G. J., and N. A. McFarlane, 1995: Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Center General Circulation Model. Atmos.–Ocean, 33, 407–446, https://doi.org/10.1080/07055900.1995.9649539.
Zhou, L., H. Chen, W. Hua, Y. Dai, and N. Wei, 2016: Mechanisms for stronger warming over drier ecoregions observed since 1979. Climate Dyn., 47, 2955–2974, https://doi.org/10.1007/s00382-016-3007-9.