Version classiqueVersion mobile

Rural societies in the face of climatic and environmental changes in West Africa

 | 
Benjamin Sultan
, 
Richard Lalou
, 
Mouftaou Amadou Sanni
, 
et al.

Part I. Recent and future climate change in West Africa. Obvious features, uncertainties and perceptions

Chapter 3. Climate projections in West Africa: the obvious and the uncertain

Abdoulaye Deme, Amadou Thierno Gaye et Frédéric Hourdin

Texte intégral

Introduction

1A serious drought hit West Africa in the 1970s (Nicholson, 1980; Held et al., 2005) whose consequences for agriculture—the main activity of the people in the region—made a serious contribution to late development. Rainfall in the Sahel seems to have recovered since the 2000s, although it is not possible to say whether the trend will last during the coming decades (Paeth and Hense, 2004). This period also coincided with warming that probably resulted from the increase in greenhouse gases (GHG), thus making knowledge of the evolution of rainfall a crucial subject. Indeed, decades of drought punctuated by famines, displacements of populations and economic collapse were replaced by or compounded with heatwaves, flooding in many cases, new outbreaks of certain vector diseases with strong impacts on the health of populations.

2In addition, as part of the ongoing climate change, the rainfall regime in West Africa is among those with the greatest uncertainty. This is shown simply by examining the results of the CNRM and the IPSL that were part of IPCC’s CMIP3 (Climate Model Intercomparison Project 3) and that give contrasting precipitation projections for West Africa towards the end of the 21st century (Joussaume et al., 2007). This is corroborated by all the models participating in the CMIP3 (whose results form the basis of the IPCC 4th report AR4); half of these indicate an increase in precipitations in the Sahel and the other half give the opposite scenario (Christensen et al., 2007). Likewise, Cook and Vizy (2006) have shown that only a small number of the general circulation models (GCMs) used for the 4th IPCC report were capable of providing a satisfactory representation of the main characteristics of the West African Monsoon (WAM): precipitation structure, meridian atmospheric circulation, the main interannual variability modes, etc. The 4th IPCC report thus concluded in 2007 that coupled climate models were not yet capable of simulating the climate in West Africa with precision (Randal et al., 2007). However, regional and global atmospheric models forced with the temperatures observed at the surface of the sea (SST) in projects such as AMMA-MIP (African Monsoon Multidisciplinary Analyses—Model Intercomparison Project; Hourdin et al., 2010), WAMME (West African Monsoon Modeling and Evaluation; Xue et al., 2010) and CORDEX-Africa (Coordinated Regional Climate Downscaling Experiment; Jones et al., 2011, Nikulin et al., 2012) gave better results for the WAM although there is still considerable bias in precipitation and meridian circulation. This is the challenge that CMIP5 aimed at taking up and whose work forms the basis of the 5th IPCC report, AR5. Thus some 20 modelling groups from about 10 countries within the framework of CMIP5 conducted a series of coordinated, standardised experiments (Taylor et al., 2012) and several questions raised still remain. Are there still Sahel rainfall projection divergences between the models? Do the GCMs reproduce the return of rainfall in recent years? What capacity do GCMs have for reproducing the main features of the WAM? What are the climate projections for West Africa in the years to come, over the next century? What evidence and uncertainties are observed in climate projections for West Africa?

3These questions are addressed here using the philosophy of the major lines of the Escape project, that is to say by performing a diagnosis of what happened in the past, what is happening today and what will happen in the light of the possible consequences of climate variability and change for resources and vulnerable economic sectors. The aim is thus to be able to provide the most accurate and objective information for the political authorities and the population to enable them to implement mitigation strategies in the face of a late rainy season, extreme climatic events (heatwaves, floods, etc.) and adaptation strategies for populations and economies in the face of climate change.

4We first describe the data used for the evaluation of the CMIP5 models available for the study. The second part includes analysis of the capacity of CMIP5 models to reproduce certain key features of the WAM. The third part is a discussion of climate projections (via rainfall and temperature data) for West Africa in the coming decades and especially those for the end of the 21st century. The obvious features and uncertainties of these projections are covered in the fourth section, with the chapter being completed by conclusions and prospects.

Data and methods

5Numerical simulations, re-analysis data and observations are used to reply to the three questions of the introduction. In study of the climate and its variability, it is essential to possess long time series with few missing figures for the zone studied; the re-analyses and satellite observations used in this work have these advantages.

CMIP5 simulations

6Use was made of available simulations from a dozen models that participated in CMIP5. The climate change scenarios (RCP2.6, RCP4.5 and RCP8.5) were compared with simulations of the past climate (referred to as ‘historical’) to study the response of the West African climate to increased atmospheric CO2. Simulations with prescribed SSTs, called AMIP simulations, were also analysed to assess the state of the art of the representation of the WAM in the GCMs. The simulations used are available at http:/cmip-pcmdi.llnl.gov/cmip5/index.html.

Reanalysis and observations

7CMIP5 climate simulations were compared with several datasets consisting of reanalysis and observations: (1) the first set consists of NCEP/NCAR rainfall and temperature (Kalnay et al., 1996) at a resolution of approximately 280 km. This dataset drawn from the reanalysis of all the observation data available (from meteorological stations, weather buoys, radiosondes, aerial observations, etc.) using a high-performance integration system covers the period 1957-1996; (2) the second dataset is drawn from ERA-Interim reanalysis (Simmons et al., 2008) by the European Centre for Medium-Range Weather Forecasts. The data are constructed using a method similar to those of NCEP/NCAR and cover the period 1958-1997 at a resolution of some 160 km; (3) the third validation dataset is version 2 of GPCP (Global Precipitation Climatology Project) precipitation (Huffman et al., 2001). GPCP precipitation was obtained by combining ground rain gauge data and indirect satellite observations. The data used are for the period 1997-2006 with resolution of 100 km for daily data and 280 km for monthly data; (4) CMAP (CPC Merged Analysis of Precipitation) precipitation figures (Xie and Arkin, 1997) form the last dataset for the validation of CMIP5 models and were also obtained by combining rain gauge and satellite observations; (5) CRUTS2.1 (Climatic Research Unit TimeSeries) data (Mitchell and Jones, 2005), based on observations from more than 4,000 meteorological stations, cover the period 1901 to 2012; (6) TRMM rainfall data (Huffman et al., 2007) were the sixth dataset used to validate the CMIP5 simulations.

The representation of several features of the WAM in the CMIP5 models

8The key features of the WAM for which the models have clear weaknesses are examined here. This is to allow better assessment of the robustness of their projections—discussed in subsequent chapters.

The seasonal cycle of the WAM

9The seasonal cycle of the monsoon is the most important feature for local populations whose main activity is rainfed farming. It displays alternation between a dry season and a rainy season during which the Sahel receives most of its precipitation. Rainfall structure in the rainy season is set around the Intertropical Convergence Zone (ITCZ), with three regional maximums: one centred in the western part of the continent towards 8° N in the Fouta Djalon uplands, a second towards 10° E near Mount Cameroon and a third near the Ethiopian high plateaux towards 40° E (not shown in the figures). The two maximums are fairly well reproduced by the ERA-Interim, CMAP and GPCP observations and their mean (Fig. 1a). GPCP gives comparatively lower values while the ERA-Interim maximums are generally greater than those of CMAP. The Russian model INMC4 stands out by its too marked under-estimation of the maximums of Fouta Djalon (Fig. 1b) while the models ACCESS1-0, ACCESS1-3, CMCC-CM, FGOALS-g2, IPSL-CM5A-LR and IPSL-CM5A-MR find a maximum in the west rather than over the sea. However, the two last French models (IPSL-CM5A-LR/MR) clearly show the maximum at Mount Cameroon. The Japanese model MIROC5 exaggerates the westward spatial extent, showing a zone running from 20°W to 10°E. But the models BCC-CSM1-1, BNU-ESM (although they underestimate the maximum at Mount Cameroon), CNRM-CM5, CSIRO-Mk3-6.0 and IPSL-CM5B-LR give a fairly accurate picture of the maximums at Fouta Djalon and Mount Cameroon. Finally, in spite of the dispersion in JAS (July–August–September) precipitation climatology, the ensemble mean of the 13 models is a fairly good representation of the maximums in West Africa and on Mount Cameroon.

10Figure 2 shows the surface temperature climatology of the JAS season for 1989-2002, observations (2a) and CMIP5 models (2b). CRU observation data show the temperature gradient in West Africa, with low temperatures in the southern part (to approximately 13°N) and high temperatures in the north, marked by a maximum in Algerian Sahara between 25° and 28°N. NCEP and Era-Interim reanalyses give a fairly clear picture of the high temperature zone: the first model gives a lower, more extensive maximum than that of CRU, while the second gives a high maximum with greater spatial coverage. Half of the CMIP5 models give too strong an underestimation of the maximum temperature in the Sahara (ACCES1-3, INMCM4, IPSL-CM5A-LR, IPSL-CM5B-LR) or place it further south at the frontier between Mauritania and Senegal (ACCESS1-0, ACCESS1-3, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR). A second group of models (BCC-CSM1-1, BNU-ESM, CMCC-CM, FGOALS-g2) gives a fairly good picture of this maximum, but more in the north of Mauritania and Mali. Finally, only two models (CSIRO-Mk3-6.0 and MIROC5) show the Sahara maximum well, with figures comparable to those of CRU, but overestimate the maximum in the east towards 20° E (between Niger and Chad), while the CNRM-CM5 model underestimates it but positions it well. As for precipitation, it is noted that the mean of all the CMIP5 models gives a fairly good picture of maximum temperatures in the Sahara and further east in the continent towards 20° E, but with slightly lower values.

Figure 1. 1989-2005 precipitation climatology for the JAS (July – August – September) rainy season: comparison of observations (a) and of CMIP5 models (b).

Figure 2. 1989-2002 temperature climatology for JAS (July – August – September) rainy season: comparison of observations (a) and of CMIP5 models (b).

Figure 3. 1989-2002 temperature climatology for the MAM (March-April-May) dry season: comparison of observations (a) and CMIP5 models (b).

11We studied the behaviour of the models during the dry season from the angle of the questions raised by the ACASIS project (Alerte aux canicules au Sahel et à leurs impacts sur la santé, Alert on Sahelian heatwaves and their impact on health), which has just started. Surface temperature climatology in MAM (March-April-May) is shown in Figure 3 for 1989-2002, with comparison of observations (3a) and the CMIP5 models (3b). Observations show overall that the high temperatures are found in the Sahel during the dry season, with three maximums: one towards the west (along the frontier between Senegal and Mali), a second around Niamey in the centre of the Sahel and a third towards 20°E in Chad. There is practically complete agreement between CRU and ERA-Interim while NCEP does find the position of these maximums but with much smaller values. Some CMIP5 models underestimate these maximum temperatures too markedly (IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR) or do not show them at all (INMCM4). A second group of models (BCC-CSM1-1, CNRM-CM5, FGOALS-g2) gives a somewhat poor representation of the structure of temperature climatology in MAM, with the maximum in the west found in the right position but missing that of the centre of the Sahel (BCC-CSM1-M1, FGOALS-g2) or underestimating temperature maximums (CNRM-CM5). A third group of models (ACCESS1-0, ACCESS1-3, BNU-ESM, CMCC-CM, CSIRO-Mk3-6.0, MIROC5) shows the three MAM temperature maximums fairly well even though some are underestimated.

Onset of the monsoon

12We have described the importance of the seasonal cycle of WAM for the local populations. In fact the seasonal cycle of the WAM that is accompanied by the northward shift of the ITCZ does not take place steadily but in the form of a rapid movement lasting about 10 days that is referred to as the ‘onset of the monsoon’ (Sultan and Janicot, 2000, 2003; Le Barbé et al., 2002). This feature is very important for the population of the Sahel as it signals the start of ‘useful’ rain for farmers (Ati et al., 2002). However, Cook and Vizy (2006) showed that for the 4th IPCC report a third of the models did not provide good simulation of the WAM and in particular did not shift the ITCZ to the continent during the northern summer. Other authors (Hourdin et al., 2010, Roehrig et al., 2013) showed that in an AMIP mode (in which the SSTs are forced by observations), the models give better simulation of the shift of the ITCZ to the continent.

13This is why the seasonal pattern of precipitation averaged for 10°W-10°E is shown in Figure 4 for CMIP5 models in an AMIP experiment and compared with GPCP and TRMM observations. The three phases of the WAM are shown clearly by the GPCP data. In the first phase, peak precipitations remain centred south of 5°N until the end of May and precipitation then decreases before increasing strongly near and north of 10°N, indicating the beginning of the rainy season in the Sahel (second phase) and reaches a peak in August. In the third phase—from August to October—GPCP shows a southward return of maximum precipitation, corresponding to the second rainy season in the Guinea region. The CMIP5 models examined show almost the same results as those of Cook and Vizy (2006): more than half do not show the onset of the monsoon sufficiently. Indeed 9 of the 13 models studied give poor coverage of the onset (BCC-CSM1-1, CMCC-CM), or a very staggered second phase, that is to say the rainy season in the Sahel (BNU-ESM), or a first phase (maximum rainfall on the Guinean coast) that is too far north (FGOALS-g2, MIROC5) or too weak (INMCM4, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR). Finally, only the models ACCESS1-0, ACCESS1-3, CNRM-CM5 and CSIRO-Mk3-6.0 succeed in reproducing more or less the three phases of intra-seasonal variability of the WAM, with ACCESS1-0 and CNRM-CM5 performing less well.

Figure 4. 1996-2008 seasonal evolution of mean precipitation between 10°W and 10°E (in mm per day) given by CMIP5 models and GPCP observations. The seasonal cycle has been smoothed by means of a 10-day sliding average.

What are the climate projections for West Africa in the 21st century?

14The 4th IPCC report (CMIP3 simulations) concluded that global warming was very likely to occur as a result of the increase in greenhouse gases in the atmosphere (Meehl et al., 2007). However, there was a divergence between models in the precipitations in certain regions and in particular in the Sahel (Cook, 2008). The question is therefore that of knowing what progress has been contributed by the work of CMIP5, based on the 5th IPCC report.

Rainfall projections for West Africa in the 21st century

15Figure 5 gives a succinct view of the changes in precipitation in West Africa in the 21st century in a so-called ‘realistic’ scenario, rcp4.5, for the period 2011-2040. Most of the CMIP5 models studied (8-10 out of 13) show for the coming 30 years rainfall projections that are rising in the Sahel east of 10°W, and falling west of 10°W (except for BCC-CSM1-1 and INMCM4) and falling in the Maghreb (except for ACCESS1-3, BNU-ESM, CSIRO-Mk3-6.0, FGOALS-g2, MIROC5). For equatorial Africa during the same period, most of the models (9 out of 13) show rising projections while the models BCC-CSM1-1, BNU-ESM, CSIRO-Mk3-6.0 and INMCM4 tend to show the opposite. The ensemble mean of the models studied displays rising rainfall projections for the Sahel (east of 10°W) and equatorial Africa and falling projections in the Sahel (west of 10°W) and the Maghreb.

16For the second half of the 21st century (figure not shown) most of the models (ACCESS1-0, ACCESS1-3, BCC-CSM1-1, CNRM-CM5, FGOALs-g2, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR, MIROC5) confirm an increase in rainfall in the central and eastern Sahel, increased rainfall in equatorial Africa and a decrease in the western Sahel and the Maghreb (with a difference between the models in the Maghreb regions concerned). Two models (CSIRO-Mk3-6.0 and INMCM4) stand out for this period by projecting decreases in rainfall in practically the whole of West Africa; two other models (BNU-ESM and CMCC-CM) give opposite results for the central and eastern Sahel and equatorial Africa. The last 30 years of the 21st century (2071-2100, figure not shown) confirm (in 10 models out of 13) upward rainfall projections in the eastern and central Sahel and in part of equatorial Africa and decreasing rainfall projections in the western Sahel and the Maghreb. The models do not agree clearly about the Maghreb. CSIRO-Mk3-6.0 and INMCM4 models also confirm their singularity by giving downward rainfall projections for the whole of West Africa, while the BCC-CSM1-1 model practically divides the continent in two: decreasing rainfall in the west and increasing rainfall in the east.

Figure 5. The difference in rainfall between the future (scenario rcp4.5) and the present (historical) climate in West Africa in JAS (July-August-September) in 12 CMIP5 models and their ensemble mean for the period 2011-2040.

17Scenario rcp4.5 shows, for the majority of the CMIP5 models studied, rainfall projections for the 21st century and during the July-August-September rainy season compared to 1961-2000: (1) an increase in the central and eastern Sahel, in part of the equatorial zone (0-10°N) and in the Maghreb; (2) a decrease in precipitation in the western Sahel, Gabon and the Congo (5°S-0°).

18Figure 6 shows the changes in precipitation in West Africa in the 21st century in comparison with present precipitation in the case of the ‘pessimistic’ scenario rcp8.5 for 2011-2040. Precipitation projections for the first 30-year period give markedly contrasted results. First, most of the models studied (BNU-ESM, CNRM-CM5, FGOALS-g2, IPSL-CM5A-LR, IPSL-CM5A-MR, MIROC5) give increasing rainfall projections for the central and eastern Sahel and a decreasing projection for the western Sahel, while two models (BCC-CSM1-1 and CSIRO-Mk3-6.0) indicate the opposite. Then three models (ACCESS1-0, ACCESS1-0 and CMCC-CM) give rising rainfall projections for the entire Sahel while two others (INMCM4 and IPSL-CM5B-LR) tend to indicate a decrease. For equatorial Africa, more than half of the models studied (BCC-CSM1-1, BNU-ESM, CSIRO-Mk3-6.0, FGOALS-g2, INMCM4, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR) show decreasing rainfall projections while five models (ACCESS1-0, ACCESS1-3, CMCC-CM, CNRM-CM5, MIROC5) indicate the opposite. Finally, almost all the models studied (except for MIROC5) give decreasing rainfall projections for the Maghreb even if they are not in agreement about the zones concerned.

19Only one model (FGOALS-g2) makes increasing rainfall projections for the entire Sahel for the second half of the 21st century (2041-2070; figure not shown). A single model again (CSIRO-Mk3-6.0) differs by giving decreasing rainfall projections of more than 1.5 mm per day for the entire Sahel. A large majority of the models (ACCESS1-0, ACCESS1-3, BCC-CSM1-1, BNU-ESM, CMCC-CM, CNRM-CM5, CSIRO-Mk3.6.0, FGOALS-g2, INMCM4, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR, MIROC5), that is to say 10 out of 13, give decreasing rainfall projections in the western Sahel and increasing projections in the central and eastern Sahel. In this category, two models (BCC-CSM1-1 and INMCM4) give greater extension in the western Sahel concerned by a decrease in precipitation: this covers an area from Niamey (2.5°E) to Senegal. The projections of almost all the models indicate a decrease in rainfall in the Maghreb, although some of these models (ACCESS1-3, BNU-ESM, CNRM-CM5, FGOALS-g2) exclude Algeria and eastern Morocco. For the southern part of equatorial Africa (Gabon, Congo and the Democratic Republic of the Congo), 8 out of 13 of the models (ACCESS1-0, BCC-CSM1-1, BNU-ESM, CSIRO-Mk3-6.0, INMCM4, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR) project a decrease in rainfall. However, the projections of several models (ACCESS1-3, CMCC-CM, CNRM-CM5, FGOALS-g2, MIROC5) tend to indicate an increase in rainfall in the above zone.

20The projections of the models for the last decades of the 21st century (2071-2100; figure not shown) show broad agreement on a decrease in precipitation in the Maghreb: only the model BNU-ESM indicates the opposite. It is true that some models (CNRM-CM5, FGOALS-g2, MIROCS) also indicate a decrease in the eastern Maghreb and in Libya. The projections of a great majority of the models (ACCESS1-0, ACCESS1-3, CMCC-CM, INMCM4, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR, MIROC5) show a decrease in rainfall in the western Sahel and an increase in the central/eastern Sahel. For this period there are also models (BNU-ESM, CNRM-CM5, FGOALS-g2) that give projections for increased rainfall throughout the Sahel and two models (BCC-CSM1-1 and CSIRO-Mk3-6.0) that project the opposite. Finally, for the south of equatorial Africa (Gabon, Congo and the Democratic Republic of the Congo), a large majority of the models (ACCESS1-0, ACCESS1-3, BNU-ESM, CSIRO-Mk3-6.0, INMCM4, IPSL-CM5A-LR, IPSL-CM5A-MR, IPSL-CM5B-LR) gives projections for a decrease in rainfall, while three others (CMCC-CM, CNRM-CM5, FGOALS-g2) indicate the opposite. Two models (BCC-CSM1-1 and CSIRO-Mk3-6.0) stand out for this zone by giving opposite projections for the Guinean coast and the south of equatorial Africa.

Figure 6. Difference in rainfall between the future (scenario rcp8.5) and the present (historical) climate in West Africa in JAS (July-August-September) in 12 CMIP5 models and their ensemble mean for 2011-2040.

21In an rcp85 scenario, the ensemble mean of the CMIP5 models studied show rainfall projections for the 21st century for the July-August-September rainy season compared to the period 1961-2000. This shows (1) an increase in the eastern Sahel, with very high figures for Burkina Faso, southern Niger, southern Chad and the Sudan towards the last decades of the 21st century, (2) a decrease in the western Sahel (Senegal, the Gambia and Guinea) and figures becoming very high towards the final decades of the century in the Casamance, the Gambia and the whole of Guinea; (3) a decrease in the Gabon, the Congo and the Democratic Republic of the Congo (in the southern part of equatorial Africa); (4) a decrease in the Maghreb affecting the whole of Morocco in the first period and remaining in the northern part of the zone (Morocco, Algeria, Tunisia and Libya) towards the last decades.

Temperature projections for West Africa in the 21st century

22July-August-September rainy season projections for the 21st century are compared to the period 1961-2000, in an rcp4.5 scenario (figure not shown): temperatures rise in West Africa in almost all the models. Only one (BNU-ESM) gives downward temperature projections—throughout the 21st century—for the central Sahel. This model is joined by FGOALS-g2 for the period 2071-2100. The projection of rising temperatures divides West Africa into two zones: (1) north of 12.5°N where the temperature difference is 2°C for the first 30 years, reaching some 2.5°C towards the end of the 21st century; (2) south of 12.5°N where the temperature difference is 1°C during the period 2011-2040 and then reaches 2°C in 2071-2100.

23In a pessimistic scenario rcp8.5, the projection of temperatures compared to those of the recent past is shown in Figure 7. Projections for the first period (2011-2040) give increases reaching 2.5-°C-3°C north of 15°N in all the models and 1°C for all the models (except for BNU-ESM that tends to give downward temperature projections in the central and eastern Sahel) south of 15°N. This warming trend in projections is confirmed by all the models, both north of 15°N (where 3.5°C is reached) and south of 15°N (2°C) during the period 2041-2070 (figure not shown). However, in this southern part of the region the warming of the entire Guinean coast is distinctly more moderate. Finally, warming is accentuated in the final decades of the 21st century, accentuating the trend (figure not shown): this varies from 3°C south of 15°N (including the Guinean coast this time) to 4-5°C northwards.

24A remarkable feature of these temperatures—even in an ‘unrealistic’ rcp2.6 scenario—is that the projections indicate rises in practically the whole of West Africa (except for Libya and part of northern Niger, Mali and Chad) of as much as 0.5°C during the ongoing decades 2011-2040 (figure not shown). Furthermore, this upward trend should reach 1°C at the end of the 21st century and concern the whole of West Africa. This has led us to focus on the dry season as Fontaine et al. (2013) showed recently that an increase in heatwaves could occur then (see also Chapter 1).

25This is why we show in Figure 8 the temperature projections for March-April-May in comparison with present temperatures (1961-2000) for a scenario considered to be realistic, that is to say rcp4.5. Increases of up to 1.5°C are observed in 2011-2040 in part of the central and northern Sahel (Algeria, Niger, northern Burkina Faso, Mali and northern Senegal), with agreement between the models examined (Fig. 8a). The increase is accentuated in the second half of the 21st century (figure not shown) reaching 2°C everywhere in West Africa and exceeding this in Algeria, Niger and Mali. Finally, the end of the 21st century will be marked by an acceleration of this rising trend (with 2.5°C exceeded everywhere in northern West Africa) and its spread as far as the southern part of the continent to about 10°N, thus affecting Guinea, Côte d’Ivoire, the whole of Burkina Faso and even northern Nigeria (Fig. 8b). These results conform the relevance of the questions raised by the ACASIS project in the analysis of heat wave phenomena during the dry period in West Africa.

Figure 7. The difference in temperature between the future (scenario rcp8.5) and the present (historical) climate in West Africa in JAS (July-August-September) in 12 CMIP5 models and their group average for the period 2011-2040.

Figure 8a. Difference in temperature between the future (scenario rcp4.5) and the present (historical) climate in West Africa in MAM (March-April-May) in 12 CMIP5 models and their group average for the periods 2011-2040.

26The precipitation projections are different for the Sahel between the west (15°W-5°W) with a decrease and the east (10°E-35°E) with an increase and a transition zone (5°W-10°E) in which the models diverge. In contrast, as regards temperatures, warming seems to be solid and display considerable amplitude. Likewise, the warming is so great in the dry season that the occurrence of frequent heatwaves is very plausible, as is mentioned in the 5th IPCC report (see also Chapter 1).

Figure 8b. Difference in temperature between the future (scenario rcp4.5) and the present (historical) climate in West Africa in MAM (March-April-May) in 12 CMIP5 models and their group average for the period 2071-2100.

What are the obvious features and uncertainties of climate projections for the 21st century in West Africa?

27The results of the CMIP5 models examined and analysed in the first paragraph show that they give a reasonable picture of certain key features of the WAM. Confidence can therefore be awarded to these models for their ability to provide more or less reliable climate projections for West Africa. The results above seem to show that confidence in the models would be higher for temperatures than for precipitations.

28The aim in this paragraph is to reply to this assertion by evaluating—for both parameters—which results are robust and, for results that are not robust, to see what uncertainties are associated with them. For this, we first designed simple diagnoses inspired by Roehrig et al. (2013) to compare the means of all the models, their dispersion and their agreement of these scales. We then took the characteristic regions of West Africa where the results seemed to diverge and determined the evolution during the 21st century of the mean values of anomalies in precipitation and temperature, together with their dispersion.

Precipitation

29Figure 9 shows the mean changes, spread and sign consistency between models for rainfall in JAS, for scenario rcp4.5 and for the periods 2011-2040 (9a), 2041-2070 (9b), 2071-2100 (9c), for scenario rcp8.5 and for the same periods respectively (fig. 9d, 9e et 9f).

30The ensemble mean of the models with an rcp4.5 scenario indicates an overall increase in precipitation during the period 2011-2040, in the eastern Sahel (from Chad to Bamako, towards 10°W), a decrease in the western Sahel (from Bamako to Senegal) and a decrease in the Maghreb (Morocco, Tunisia and northern Libya). Very marked variations are seen in standard deviation (Fig. 9a), reaching 2-2.5 mm per day, especially in the Gulf of Guinea as far as Liberia and Sierra Leone, thus showing considerable uncertainty with regard to this increase in rainfall. Agreement between models with regard to the increase in the Sahel reaches 80%, whereas that for the western Sahel varies from 60% (south Senegal and Guinea-Bissau) to 40% (Bamako-Senegal). Agreement as to the decrease in precipitation is weak (30-40%) in the Maghreb as well. No noteworthy changes in ensemble mean and dispersions in models in the Sahel are noted in an rcp8.5 scenario (Fig. 9d). In contrast, agreement among models changes in the eastern Sahel: it stays at 80% in the most easterly part (Niger-Chad) but decreases to 40-60% in Burkina Faso and Mali and to 40% in the western Sahel.

31Few changes are observed in the ensemble mean in the half of the 21st century and in an rcp4.5 scenario (Fig. 9b), except for the spread of the decrease in the Maghreb (reaching Mauritania, Algeria and almost the whole of Libya) and its accentuation in the Casamance and Guinea. However, an increase in dispersion in the eastern Sahel and a change in the agreement between models in this zone are noted. Agreement remains strong (80%) in Niger and Chad but decreases strongly (40-60%) for Bamako-Niger; in contrast, it increases in the western Sahel, reaching 50-60% in a large part of Senegal and 40-60% in the northern Maghreb. The rcp8.5 scenario (Fig. 9e) confirms the increase in agreement for Senegal (60%) and in much of the Maghreb.

32Rainfall projections for the mean of all the models tend to display a decrease in precipitation in the Maghreb at the end of the 21st century in an rcp4.5 scenario (Fig. 9c) but vary very little for the Sahel in comparison with the beginning and middle of the 21st century. However, agreement between models decreases in the eastern Sahel (Niger-Chad) while remaining fairly high (60%). In contrast, it remains stable in the western Sahel and the Maghreb, but covers the whole of these areas, unlike the case in the preceding period. Dispersion increases considerably in the Sahel, reaching 2.5 mm per day almost everywhere in an rcp8.5 scenario (Fig. 9. f). But agreement on the difference in rainfall divides the Sahel into three zones: the west (15° W-5° W) with 40-60% of the models, the centre (5° W-10° E) with few models (20-30%) agreeing with each other and the extreme eastern part (10° E-30° E) with agreement between 60-80% of the models.

Figure 9. Ensemble mean, spread and sign consistency of the CMIP5 models in the evolution of rainfall in JAS, for scenario rcp4.5 and the periods 2011-2040 (a), 2041-2070 (b) and 2071-2100 (c) (d-f for scenario rcp8.5).

33The following features of these results seem obvious in the two scenarios studied: (1) differences in the Sahel for rainfall projections in the 21st century, with a western part (Bamako-Senegal) displaying a decrease in precipitation and an eastern part (10° E-30° E), from Niger to the Sudan, with a somewhat opposite trend; (2) the special case of a transition zone from 5°W to 10°E (Bamako to the centre of Niger), where the models do not display clear agreement with regard to the decrease in precipitation.

Temperatures

34Figure 10 shows the ensemble mean, inter-model dispersion and agreement between models for the variation of temperature in JAS in scenario rcp4.5 for the periods 2011-2040 (10a), 2041-2070 (10b) and 2071-2100 (10c) and for scenario rcp8.5 for the same periods (Fig. 10d, 10e and 10f). The results seem robust for rcp4.5 and the three periods covered and feature warming everywhere, accentuated north of 15°N in the first period and even reaching 10°N in the middle and at the end of the 21st century, with levels always greater than 2°C (Fig. 10a-c). Likewise, projection dispersions are broad (from 1.5°C in the first decades to 2.5°C in 2041-2071) in a Sahelian strip between 10° N-20° N and 0°-30° E. This dispersion is less marked in this strip in the western part of the continent (15°W - 0°). The warming observed is common to the models except in zones of broad dispersion but where agreement between the models nevertheless exceeds 80%. The configurations are similar in an rcp8.5 scenario (Fig. 10d-f) but with an almost doubling of warming in comparison with the figures of the rcp4.5 scenario.

35For discussion of projection uncertainties, we therefore considered the zones identified as having broad dispersions in the models (rather than examining West Africa as a whole). Annual mean anomalies were calculated (in comparison with the 20th century) in rainfall and temperature projections with their dispersions.

36Figure 11 shows the evolution in time from 1900 to 2100 of the annual temperature and precipitation anomaly of the ensemble mean of the twelve MIP5 models studied in the western Sahel for the rcp4.5 and rcp8.5 scenarios. Warming in a cp4.5 scenario seems to have been distinct since the 1980s, reaching nearly 1°C in 2010, almost doubling in 30 years before stabilising at about 3°C towards 2070 (Fig. 11a). The uncertainty with regard to this warming (+/-0.2°C) is fairly small in the 2010s and then increases very rapidly to +/-1.2°C towards the end of the 21st century. In an rcp8.5 scenario (Fig. 11c), warming is practically identical to that of the rcp4.5 scenario until 2040 when it reaches 2.5°C (2°C in the rcp4.5 scenario) but then increases very rapidly: it is 4°C in 2070 and then 6°C at the end of the 21st century.

Figure 10. Ensemble mean, spread and sign consistancy of the CMIP5 models in the evolution of temperature in JAS, for scenario rcp4.5 and the periods 2011-2040 (a), 2041-2070 (b) and 2071-2100 (c) (d-f for scenario rcp8.5).

Figure 11. Evolution of the temperature anomaly (a, c) and of the rainfall anomaly (b, d) of the mean of all the CMIP5 models studied between the 20th century and the 21st century in the western Sahel zone (10-20°N, 15°W-5°W, indicated by a red rectangle), for the rcp4.5 and rcp8.5 scenarios.
The dotted lines represent the anomaly, the pale area represents 1-sigma and the dark area 2-sigma.

37The uncertainty of this warming is much the same as in the rcp4.5 scenario until 2040, but it is more than double (+/-2.8°C) towards the end of the 21st century. A great difference between the warming in the rcp4.5 and rcp8.5 scenarios is that it stabilises from 2070 in the first case but continues to increase fairly rapidly in the second. Although it is variable, the rainfall anomaly in an rcp4.5 scenario (Fig. 11b) is generally positive from the beginning of the 20th century until the 1970s, when it becomes negative—coinciding with the Sahel drought. After a slight return to positive anomalies towards 2010, a decrease in precipitation tends to take shape, lasting until the end of the 21st century. This rainfall anomaly range fluctuates around +/-80 mm in 2040-2070, reaching +/-120 mm towards the end of the 21st century. In an rcp8.5 scenario (Fig. 11d), the rainfall anomaly is slightly positive but with strong variability for 1990-2010. It then becomes negative again from 2040 until the end of the 21st century. The amplitude of variations is distinctly greater than that of the rcp4.5 scenario, varying between +/-90 mm in 2010-2040, between +/-120 mm in 2040-2070 and between +/-230 mm in 2070-2100.

Figure 12. Evolution of the temperature anomaly of the mean of all the CMIP5 models studied, covering the 20th and 21st centuries in the eastern Sahel zone (10-20° N, 10° E-35° E; indicated by a red rectangle), for the rcp4.5 and rcp8.5 scenarios.
The dotted lines represent the anomaly, the pale area represents 1-sigma and the dark area 2-sigma.

38From 1980, warming is just as distinct in the eastern part of the Sahel in an rcp4.5 scenario (Fig. 12a) and displaying practically the same evolution as in the western zone. However, variation amplitudes are slightly greater at +/-0.8°C towards 2040 (against 0.6°C for the west) and +/-1.4°C from 2070 (against 1.2°C). In an rcp8.5 scenario (Fig. 12c), the evolution of the temperature anomaly is very similar to that of the western zone. However, variation amplitudes are slightly greater. The rainfall anomaly in the eastern Sahel in an rcp4.5 scenario displays an upward trend from the 1990S until 2040 (Fig. 12b), with strong variability. From 2040, although it remains positive, the rainfall anomaly oscillates around 30 mm until the end of the 21st century. The amplitude of variations increases very rapidly from 2000, oscillating around +/-140 mm in 2040 (+/-100 mm in the western zone) reaching +/-160 mm towards 2100 (against +/-120 mm). The amplitude of the variation of the rainfall anomaly in the eastern zone is thus distinctly greater than that in the western zone. In an rcp8.5 scenario (Fig. 12d), the increasing rainfall trend is greater and faster from 2040 (30 mm), reaching 100 mm towards 2100. The amplitude of variations (+/-140-300 mm) is distinctly greater than that of the western zone during the same periods.

Conclusion

39The results of simulations using 12 models participating in the CMIP5 have show that although these models have made considerable progress in representing the features of the WAM, their climate projections for West Africa have advanced little in comparison with those of CMIP3. However, the indication of warming in West Africa seems strong, reaching a mean of 3°C in an rcp4.5 scenario and twice this figure in an rcp8.5 scenario, that is to say 10% to 60% greater than average global warming. The dispersion of projections given by the models, as in CMIP3, is always substantial for temperatures but marked above all for precipitation. Although models are practically unanimous with regard to temperature in the warming forecast, variations of amplitude of 1.8-4.2°C in an rcp4.5 model and 3.5-8.5°C in the western Sahel (15° W-5° W) are reached; the figures are slightly higher in the eastern Sahel (10° E-35° E). This uncertainty with regard to the amplitude of variations of warming remains a large challenge to be taken up for the health implications involved. For precipitation, the projections given by the models lead to more questions than answers. What seems to be abundantly clear is that precipitation decreases in the western Sahel and increases in the eastern Sahel. This observation for the western Sahel is common to an increasing number of models as the 21st century proceeds: 40% in 2011-2040, 60% in 2041-2070 and over 80% in the final period. In contrast, even though agreement between models is fairly close for the eastern Sahel, precipitation decreases by 80% in the first period to 70% in the last period. The rcp4.5 scenario seems to show oscillations around a mean value in both the east and the west (positive in the first zone and negative in the second) implying great inter-annual variability, while the rcp8.5 scenario shows a distinct upward trend in precipitation anomaly in the east, reaching 100 mm at the end of the 21st century. The amplitudes of the variations in precipitation are fairly large, ranging in the western Sahel between 80 and 120 mm in the rcp4.5 scenario and between 90 and 230 mm in the rcp8.5 scenario. These amplitudes are slightly greater in the eastern zone. The existence of a zone from 5°W to 10°E (covering western Mali, Burkina Faso, northern Nigeria and eastern Niger) is observed where agreement between the models is small with regard to changes in precipitation in the two scenarios.

40In short, although the CMIP5 models studied can give a good picture of the essential features of the WAM, it seems difficult to use their precipitation projections to make direct use of them to anticipate climate changes and their impacts as regards rainfall. This does not seem to be the case for temperatures, for which signs found are robust.

Bibliographie

References

Ati O. F., Stigter C. J., Olapido E. O., 2002
A comparison of methods to determine the onset of the growing season in northern Nigeria.
Int. J. Climatol., 22 (6): 731-742.

Chou C., Neelin J. D., Chen C. A., Tu J. Y., 2009
Evaluating the rich-get-richer mechanism in tropical precipitation change under global warming.
J. Climate, 22: 1982-2005.

Christensen J. H. et al., 2007
«Regional climate projection». In Solomon S., Qin D., Manning M., Chen Z., Marquis M. C., Averyt K. B., Tignor M., Miller H. L. (eds): Climate Change 2007: The Physical Science Basis, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA (R), Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, 996 p.

Cook K. H., 2008
Climate science: the mysteries of Sahel droughts. Nat. Geosci., 1: 647-648.

Cook K. H., Vizy E. K., 2006
Coupled model simulations of the West African monsoon system: twentieth-and twenty-first-century simulations.
J. Climate, 19: 3681-3703.

Fontaine B., Janicot S., Monerie P.-A., 2013
Recent changes in air temperature, heat waves occurrences, and atmospheric circulation in Northern Africa. J. Geophys. Res. Atmos., 118: 1-17, doi:10.1002/jgrd.50667,2013.

Held I. M., Delworth T. L., LU J., Findell K. L., Knutson T. R., 2005
Simulation of Sahel drought in the 20th and 21th centuries. Proc. Natl. Acad. Sci. U.S.A., 102: 17891-17896.

Hourdin et al., 2010
Amma-Model Intercomparison Project. Bull. Amer. Meteor. Soc., 91 (1): 95-104.

Huffman G. J., Adler R. F., Morrissey M. M., Bolvin D. T., Curtis S., Jouyce R., Mc-gavock B., Susskind J., 2001
Global precipitation of one-degree daily resolution from multisatellite observations. J. Hydrometeor., 2: 36-50.

Huffman G. J., Bolvin D. T., Nelkin E. J., Wolff D. B., Gu G., Hong Y., Bowman K. P., Stocker E. F., 2007
The TRMM Multisatellite Precipitation Analysis (TMPA): Quasi-global, multiyear, combined-sensor precipitation estimates at fine scales. J. Hydrometeor., 8 (1): 38-55.

Jones C. G., Giorgi F., Asrar G., 2011
The Coordinated Regional Downscaling Experiment: Cordex. An international downscaling link to CMIP5.
CLIVAR Exchanges, 56, 16 (2): 34-400.

Joussaume S., Armand D., Delecluse P., Seguin B., Journé V., Delmas R., Gillet M. (eds), 2007
Les recherches françaises sur le changement climatique. Institut national des sciences de l’univers, 19 p., disponible sur http://www.insu.cnrs.fr/a2059,recherches-francaises-changement-climatique-2007.html

Kalnay E. et al., 1996
The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc., 77 (3): 437-471.

Le Barbé L., Lebel T., Tapsoba D., 2002
Rainfall variability in West Africa during the years 1950-90.
J. Climate, 15: 187-202.

Meehl G. A., Covey C., Delworth T., Latif M., Mcavaney B., Mitchell J. F. B., Stouffer R. J., Taylor K. E., 2007
The WCRP CMIP3 multimodel dataset: a new era in climate change research. Bull. Amer. Meteor. Soc., 88: 13833-1394.

Mitchell T. D., Jones P. D., 2005
An improved method of constructing a database of monthly climate observations and associated high-resolution grids. Int. J. Climatol., 25 (6): 673-712, doi:10.1002/joc.1181.

Nicholson S., 1980
The nature of rainfall fluctuations in subtropical West Africa.
Mon. Wea. Rev., 108: 473-487.

Nikulin G. et al., 2012
Precipitation climatology in an ensemble of Cordex-Africa regional climate simulations.
J. Climate, 25 (18): 6057-6078.

Paeth H., Hense A., 2004
SST versus climate change signals in West Africa rainfall: 20th-century variations and future projections.
Climatic Change, 65: 179-208.

Randal D. A. et al., 2007
«Climate models and their evaluation». In Solomon S., Qin D., Manning M., Chen Z., Marquis M. C., Averyt K. B., Tignor M., Miller H. L. (eds): Climate Change 2007: The Physical Science Basis, Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA (R), Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, 996 p.

Roehrig R., Bouniol D., Guichard F., Hourdin F., Redelsperger J-L., 2013
The present and future of the West African Monsoon: a process-oriented assessment of CMIP5 simulations along the AMMA transect. J. Climate, 26: 6471-6505.

Simmons A., Uppala S., Dee D., Kobayashi S., 2008
ERA-Interim: New ECMWF reanalysis products from 1989 onwards. ECMWF Newsletter, 110.

Sultan B., Janicot S., 2000
Abrupt shift of the ITCZ over West Africa and intra-seasonal variability. Geophys. Res. Lett., 27: 3353-3356.

Sultan B., Janicot S., 2003
The West African monsoon dynamics. Part II: The «preondet» and «onset» of the summer monsoon.
J. Climate, 16: 3407-3427.

Taylor K. E., Stouffer R. J., Meehl G. A., 2012
An overview of CMIP5 and the experiment design.
Bull. Amer. Meteor. Soc., 93 (4): 485-498, doi:10.1175/BAMS-D-11-00094.1.

Xie P., Arkin P. A., 1997
Global precipitation: a 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs.
Bull. Amer. Meteor. Soc., 78: 2539-2558.

Xue Y., De Sales F., Lau W. K.-M., Boone A., Feng J., Dirmeyer P., Guo Z., Kim K. M, Kitoh A., Kumar V., Poccard-Leclercq I., Mahowald N., Moufouma-Okia W., Pegion P., Rowell D. P., Schemm J., Schubert S. D., Sealy A., Thiaw W. M., Vintzileos A., Williams S. F., Wu M-Li C., 2010
Intercomparison and analyses of the climatology of the West African Monsoon in the West African Monsoon Modeling and Evaluation project (WAMME) first model intercomparison experiment.
Clim. Dyn., doi:10.1007/s00382-010-0778-2

Table des illustrations

Légende Figure 1. 1989-2005 precipitation climatology for the JAS (July – August – September) rainy season: comparison of observations (a) and of CMIP5 models (b).
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-1.jpg
Fichier image/jpeg, 488k
Légende Figure 2. 1989-2002 temperature climatology for JAS (July – August – September) rainy season: comparison of observations (a) and of CMIP5 models (b).
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-2.jpg
Fichier image/jpeg, 570k
Légende Figure 3. 1989-2002 temperature climatology for the MAM (March-April-May) dry season: comparison of observations (a) and CMIP5 models (b).
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-3.jpg
Fichier image/jpeg, 539k
Légende Figure 4. 1996-2008 seasonal evolution of mean precipitation between 10°W and 10°E (in mm per day) given by CMIP5 models and GPCP observations. The seasonal cycle has been smoothed by means of a 10-day sliding average.
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-4.jpg
Fichier image/jpeg, 490k
Légende Figure 5. The difference in rainfall between the future (scenario rcp4.5) and the present (historical) climate in West Africa in JAS (July-August-September) in 12 CMIP5 models and their ensemble mean for the period 2011-2040.
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-5.jpg
Fichier image/jpeg, 415k
Légende Figure 6. Difference in rainfall between the future (scenario rcp8.5) and the present (historical) climate in West Africa in JAS (July-August-September) in 12 CMIP5 models and their ensemble mean for 2011-2040.
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-6.jpg
Fichier image/jpeg, 425k
Légende Figure 7. The difference in temperature between the future (scenario rcp8.5) and the present (historical) climate in West Africa in JAS (July-August-September) in 12 CMIP5 models and their group average for the period 2011-2040.
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-7.jpg
Fichier image/jpeg, 413k
Légende Figure 8a. Difference in temperature between the future (scenario rcp4.5) and the present (historical) climate in West Africa in MAM (March-April-May) in 12 CMIP5 models and their group average for the periods 2011-2040.
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-8.jpg
Fichier image/jpeg, 438k
Légende Figure 8b. Difference in temperature between the future (scenario rcp4.5) and the present (historical) climate in West Africa in MAM (March-April-May) in 12 CMIP5 models and their group average for the period 2071-2100.
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-9.jpg
Fichier image/jpeg, 400k
Légende Figure 9. Ensemble mean, spread and sign consistency of the CMIP5 models in the evolution of rainfall in JAS, for scenario rcp4.5 and the periods 2011-2040 (a), 2041-2070 (b) and 2071-2100 (c) (d-f for scenario rcp8.5).
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-10.jpg
Fichier image/jpeg, 595k
Légende Figure 10. Ensemble mean, spread and sign consistancy of the CMIP5 models in the evolution of temperature in JAS, for scenario rcp4.5 and the periods 2011-2040 (a), 2041-2070 (b) and 2071-2100 (c) (d-f for scenario rcp8.5).
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-11.jpg
Fichier image/jpeg, 530k
Légende Figure 11. Evolution of the temperature anomaly (a, c) and of the rainfall anomaly (b, d) of the mean of all the CMIP5 models studied between the 20th century and the 21st century in the western Sahel zone (10-20°N, 15°W-5°W, indicated by a red rectangle), for the rcp4.5 and rcp8.5 scenarios.The dotted lines represent the anomaly, the pale area represents 1-sigma and the dark area 2-sigma.
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-12.jpg
Fichier image/jpeg, 204k
Légende Figure 12. Evolution of the temperature anomaly of the mean of all the CMIP5 models studied, covering the 20th and 21st centuries in the eastern Sahel zone (10-20° N, 10° E-35° E; indicated by a red rectangle), for the rcp4.5 and rcp8.5 scenarios.The dotted lines represent the anomaly, the pale area represents 1-sigma and the dark area 2-sigma.
URL http://books.openedition.org/irdeditions/docannexe/image/12325/img-13.jpg
Fichier image/jpeg, 200k

© IRD Éditions, 2017

Conditions d’utilisation : http://www.openedition.org/6540

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search