Drought in the Sahel | SpringerLink

26 downloads 2607 Views 375KB Size Report
Abstract. The Sahel region of West Africa is well known as a region of environmental degradation. The reported incidence of desertification has been challenged ...
GeoJournal 48: 299–311, 1999. © 2000 Kluwer Academic Publishers. Printed in the Netherlands.

299

Drought in the Sahel C.T. Agnew1 & A. Chappell2 1 School of Geography, Manchester University,

Oxford Road, Manchester, M13 9PL, U.K., e-mail: [email protected]; Institute of Environmental Systems, Department of Geography, University of Salford, Manchester, M5 4WT U.K., e-mail: [email protected]

2 Telford

Received 6 May 1999; accepted in revised form 21 February 2000

Key words: desertification, desiccation, drought, drought classes, meteorological drought, Sahel

Abstract The Sahel region of West Africa is well known as a region of environmental degradation. The reported incidence of desertification has been challenged but persistent and widespread drought is still widely accepted. Drought, defined solely as a function of rainfall, is believed to have commenced in the early 1970s and continued through to the present. It is usually defined as a meteorological phenomenon and standardised rainfall anomalies are employed to indicate the severity of negative departures from the ‘norm’. There are several difficulties with this approach. The period of standardising rainfall has changed from 1931–1960 to 1961–1990 but the impacts on drought occurrence have not been fully determined. The spatial aggregation of rainfall anomalies may mask important local variation and the purely statistical approach to defining drought takes little account of human impact. The first two issues, averaging period and spatial aggregation, are investigated through an analysis of rainfalls in Continental Sahel (Bukina Faso, Mali and Niger). A new classification of drought classes is suggested. Despite the clear evidence of negative rainfall anomalies for rainfalls aggregated across the Sahel region, it is found that the averaging period has a significant impact on our perceptions of the occurrence of what can be considered to be meteorological drought according to the definition employed and that there is significant spatial variation.

Introduction The Sahel region in West Africa is renown for its twin environmental problems of drought and desertification. The region has been described by Sivakumar and Wallace (1991, p. v) “as one of the harshest climatic regions of the world, with low and highly variable rainfall, high soil and air temperatures, high evaporative demand, and poor soils”. All recent decades have witnessed reports on drought in the Sahel which is seen by many authors to be both widespread and persistent. In the 1970s a downward rainfall trend was noted by several climatologists including Lamb (1974) and Winstanley (1973). Concern for the inhabitants of the region grew during the early 1970s and eventually the ‘Sahel drought’ became an international relief effort with the FAO (11/5/1973) announcing “In some areas there now appears serious risk of imminent human famine and virtual extinction of herds vital to nomad populations”. In the same year a UNICEF telex recorded “drought for four to five years, . . . desert is moving southward . . . problem of survival . . . drought stricken regions.” Several years later the Guardian newspaper (10 October 1977) reported that drought had struck again and noted the call from the Mauritanian Government for assistance from the international community to save human lives from death. The same concerns continued through the 1980s and a decade later Copans (1983) estimated that there had been 100,000 drought related deaths in the Sahel. A downward

trend in rainfall was observed by many (Hare, 1984; MacDonald, 1986) with Tooze (1984) reporting that the drought in the Sahel has not yet ended. Tickell (1986) (then head of the Overseas Development Agency) went even further and announced that the documentary record for the last 70 years (in the Sahel) shows a slight decline of rainfall from 1955 and acute drought since 1968. Not all believed in a persistent drought but the view of Flohn (1987) that drought started in the late 1960s and never really ended is characteristic of the prevailing perception for the region. The same image of a drought ravaged landscape, sometimes with an advancing desert front, can be found in the 1990s literature. Pritchard (1990, p. 246) “The desert is advancing partly because of recurring cycles of drought”; Odingo (1992, p. 6) “After a 20 year series of droughts, the Sudano-Sahelian region remained the most permanently vulnerable area.”; Nicholson and Palao (1993, p. 371) “The Sahelian region of West Africa is well known for the extreme droughts it experiences. The current one has prevailed since the late 1960s”; D’Amato and Lebel (1998, p. 956) “the prolonged drought that has struck the Sahel for 25 years now”; and finally Zheng and Eltahir (1998, p. 2078) “Observations from West Africa indicate a significant decline in rainfall levels since the early 1960s”. In contrast 1994 was one of the wettest years in Niger since the late 1960s, and high rainfalls have been recently observed, but it is too soon to make any comment about

300 their significance. Doubts about the extent and persistence of drought have been raised earlier than this. Toulmin in 1988 reported on remarkably good rains in the Sahel while Glantz (1987) lists several sources that noted an end to the Sahel drought in the 1970s. Wijkman and Timberlake (1985) go even further in claiming that human activity is to blame for famine and death not a reduction in rainfall. A view that was promoted by Franke and Chasin (1980) in their analysis of peanut production, and elaborated upon by Glantz (1994) in his book aptly entitled Drought Follows the Plow. Norse (1994, p. 134) also suggests that climate change is not the only factor and argues “food insecurity has stemmed from institutional factors”. Others have also questioned the impact of drought on food supply, (Agnew, 1995; Garcia, 1981; Kelly and Buchanan-Smith, 1994; Olsson, 1993). Despite some contradictory evidence, the prevailing view is however that the Sahel is a drought stricken region. This is in contrast to the second major type of environmental degradation in the region, the notion of desertification, sometimes coupled to an expanding Sahara, (Hulme and Kelly, 1993; LeHouerou, 1996; Raynaut, 1997; UNEP, 1992). Many authors have challenged the idea of an advancing Sahara (Binns, 1990; Mainguet, 1991; Warren and Agnew, 1988) as Thomas (1993, p. 323) states ‘the concept of an advancing desert front used to infer desertification and the spread of the Sahara is incorrect’. Desertification is much more than merely the expansion of existing deserts and can be defined in terms of land degradation (Bruins and Berliner, 1998; Thomas and Middleton, 1994), but the widespread occurrence of land degradation in the Sahel region has also been questioned (Agnew and Warren, 1996; IUCN, 1989; Mainguet, 1999; Warren, 1998). Yet relatively few have questioned the incidence of drought. This paper then seeks to critically examine the evidence upon which claims of 25 to 30 years of Sahelian drought has been based. It is believed that this is an important issue as attention that is being paid to drought and expenditure on possible solutions or relief may be inappropriate. It is also timely as a new 30 year period (1961 to 1990) has been identified for the establishment of climatic conditions in the region. Although this has been investigated by Hulme (1992) he did not consider the implications for drought occurrence. This paper then looks at the occurrence of drought in the Sahel region over the period 1961 to 1990 based on a climatological assessment. We use annual rainfall characteristics to initially assess the occurrence of drought and then investigate the implications of changes in temporal and spatial variability. However before embarking upon this analysis we need to establish the area of the Sahel upon which this investigation is based, explain the rationale of using annual rainfall totals and the definitions of drought that we will employ. Where is the Sahel? The Sahel region is not well demarcated, the name meaning the ‘edge of desert’ but this is an ambiguous definition. It is normally taken to be the arid West African countries

from Senegal to Chad but some also include Sudan to the East (Hulme, 1996) while Sivakumar and Wallace (1991) employ a definition based upon a 150 day growing season which covers the Gambia and Guinea to the west and extends the southern boundary into Nigeria, Togo and Benin. Some examples based on isohyets are,

Annual precipition. The Sahel lies: FAO < 300 mm (Copans, 1983). Agnew (1982) 200 to 700 mm. Hulme (1992) 100 to 600 mm. WMO 300 to 750 mm (vegetation) (Davy et al., 1976). WMO 100 to 700 (landforms) (Davy et al., 1976). Some obvious contrasts arise such as whether or not the Sahel is based upon a pastoral region where annual rainfall is less than 300 mm or embraces permanent dryland cultivation where annual rainfall is greater than 500 mm? Alternatively the boundaries have been drawn using latitude and longitude:

Latitude Sivakumar and Wallace (1991) used 10 to 15◦ N (based on 60 to 150 growing days). Ba et al. (1995) employed 9 to 30◦ N and further divided the Sahel into two regions using longtitude 5◦ W, Region 1 to the West includes Mauritania, Senegal, Guinea, Gambia and part of Mali; Region 2 extends eastwards to Chad (Figure 1). Nicholson and Palao (1993) used latitudes of 10 to 20◦ N. They found West Africa was heterogeneous with respect to rainfall variability and using a principal components analysis of annual rainfalls divided it into three regions, the Guinea coast (up to latitude 10◦ N; the West Coast (Senegal, Mauritania, Gambia and Guinea) and the Sahel (10 to 20◦N) including, Mali, B. Faso, Niger, Chad. Others have since reported a similar spatial demarcation (Janicot et al., 1998) with the ‘Continental Sahel’ being recognised as a rainfall region. Therefore the analysis here investigates further the latitudinal grouping within the Continental Sahel area focusing upon Niger, Mali and B. Faso rainfalls. The study area is shaded in Figure 1.

Why use annual rainfalls? Several agroclimatic studies have already been carried out for the region that demonstrate the importance of annual water supply and plant growth, (Cocheme and Franquin, 1967; Davy et al., 1976; Sivakumar, 1986, 1989). Ahlcrona (1988) related seasonal rainfalls to millet yields in the Sudan, the IUCN (1989) showed the strong association between rainfall and livestock numbers and the 1991 Niamey international workshop was based on the assumption that (Sivakumar and Wallace, 1991, p. v) “Water availability is a major constraint limiting food crop production”. The advantage of using rainfall to investigate drought occurrence is that data are available while other variables such

301

Figure 1. The Continental Sahel Region used in this study covering parts of Mali, Burkina Faso and Niger (shaded). The boundaries of Continental Sahel follow the demarcation by Ba et al. (1995) with a Southern boundary of 9◦ N, a Western boundary of 5◦ W and a vague Northern limit supposedly the edge of the desert. In reality the Northern boundary is fixed by the availability of rainfall stations, i.e., Agadez 18.7◦ N and Kidal 18.4◦ N (for a more detailed discussion of the demarcation of bio-climatic boundaries see Bruins and Berliner 1998).

as river flow are increasingly managed by human action, e.g., discharges of the river Senegal show a downward trend but as dams have now been constructed the current record is of limited use. Similarly flows in the River Niger are affected by irrigation in Mali. Rainfall data at a shorter time scale, e.g., daily can be useful for detailed water balance analysis (Agnew, 1989, 1991a) but a recent paper on rainstorm characteristics by D’Amato and Lebel (1998, p. 955) reached the interesting conclusion that “One major result of this work is that the statistics characterising the rain events in the Sahel display little fluctuations, either in space or from year to year, as compared with those observed for the total seasonal rainfall”. Hence seasonal rainfall and especially the number rather than the magnitude of rainfalls is the key source of variation. It has also been shown by Nicholson and Palao (1993) that rain in the month of August is highly correlated with seasonal or annual rainfalls (99% of annual rain falls May to September), e.g., for all stations in Niger, August monthly rainfall explains 65% of annual rainfall but the boreal summer (June, July and August) explains 92%. Annual totals are employed here to investigate the occurrence of meteorological drought years with confidence that they will display any significant rainfall variation. It should be noted that, in the Sahel, drought is best examined as an agroclimatic phenomenon where spatial and temporal variation are important. The author and others (Agnew, 1989, 1990, 1991b, 1995; Bley et al., 1991) have already shown that agroclimatic drought is not as persistent nor as widespread as the claims for meteorological drought occurrence. Hence the need to re-examine the latter, especially in the light of changes in the averaging period.

Definitions of drought Palmer (1965), Beran and Rodier (1985) and Gordon (1993) have argued that the primary characteristic of drought is a reduction in precipitation. There are many other types of definition ranging from agricultural to economic drought (Agnew, 1989) but here we are primarily concerned with meteorological drought resulting from a lack of precipitation. There is a consensus that drought occurs when rainfall is ‘below average’ (Bruins and Berliner, 1998) but Tannehill (1947) argued this should be an abnormal moisture deficiency while others such as Dracup et al. (1980) employ vague terms such as below normal, deficient precipitation and less than average. There are numerous examples of more rigorous definitions which tend to employ either a threshold value e.g, the U.K Met. Office employs 15 consecutive days without rain while in the USSR drought can be defined as 10 days when the total rainfall does not exceed 5 mm, or statistical value e.g., twice the standard deviation below the average rainfall (India) or 70% of normal rainfall (South Africa) (see Agnew and Anderson, 1992; McKee et al., 1994; Wilhite, 1993). Curiously, of the many reports of drought in the Sahel listed above, few quantify the severity of drought in any fashion but display a period negative rainfall anomalies (see Jones and Hulme, 1996) as proof of meteorological drought. This is overly simplistic as it suggests that any period of below average rainfall can be considered a drought year. A more rigorous approach to the assessment of meteorological drought severity is required based on a analysis of past rain-

302 falls but there are three fundamental problems with this type of strategy for the analysis of drought in drylands. • It ignores the impacts of rainfall upon human activities and/or the rest of the environment in a period of economic and political change. • It assumes that a statistical definition of drought can be reliably employed despite the variability of the data and changing climatological base period. • It confuses drought with desiccation. The first point has been dealt with elsewhere (Agnew, 1995) but the last two points require elaboration. The term ‘persistent drought’ can be found in reports on this region but what does this mean in an arid area? If rainfalls are always low then lack of rainfall can hardly be classified as drought. Care then needs to be exercised when using standardised rainfall anomalies rather than absolute reductions in rainfall amounts to assess regional drought in areas that include rainfall stations in hyper-arid locations. Furthermore a distinction should be made between drought, being a short term phenomenon, and desiccation, being a long term increase in aridity. If the climate is changing and entering a drier phase then ‘what is normal’ and ‘what is abnormal’ i.e. drought needs to be re-established based on the ‘new’ climate conditions. Otherwise it is necessary to establish some fixed threshold values against which drought incidence can be evaluated irrespective of the change in climate, e.g., the Palmer Index (Palmer, 1965). Those who analyse drought in the Sahel are now faced with a dilemma as the climatological base period has shifted from an averaging period of 1931 to 1960; to 1961 to 1990. With this a new set of normal/abnormal conditions have been established hence the frequency of a statistically defined meteorological drought is liable to change. It is then necessary to investigate the meaning of meteorological drought in the context of changing base periods, a recent period of desiccation, and the strategy of aggregating regional statistics that range from the hyper-arid to the wetter sudano-sahelian zone.

Analysis of recent meteorological drought in the Sahel What is normal? The use of averages and standard deviations to assess drought can be questioned both on the basis of the annual rainfall frequency distribution, the spatial variability of the data and the network of meteorological stations. Recent examination (Agnew and Chappell, 2000) of impacts due to changes in the Sahelian raingauge network suggest that it was only after 1970 that the network provided sufficient observations for reliable E–W analysis; but the N–S arrangement of rain gauges is still too biased to the south. Davy et al. (1976) expressed some concerns over the use of a normal distribution for rainfall in the Sahel, especially in the drier regions where zero values are found, but then used this approach due to its simplicity. Agnew (1982) and Sivakumar (1986) showed that for shorter periods of 10 day rainfalls the data become highly skewed and follow a gamma function. It is widely assumed however that for periods of

30 plus years normality can be accepted. Figure 2 shows that the mean annual rainfall, calculated over 30 year periods and averaged over the region appears to have been declining for much of the 20th Century and not just during the early 1960s. This raises the question, does the period 1961 to 1990 represent average Sahelian rainfall? Hulme (1992) has already investigated the effect of changing the base years for averaging rainfalls over the Sahel from 1931–1960 to 1961–1990 and the impact on rainfall variability. He argued (p. 690) “since the standard deviation closely co-varies with the mean rainfall, the difference between, or ratio of, two standard deviations is not necessarily the best measure to use when mean rainfall conditions are themselves changing”, and went on to investigate changes in the coefficient of variation between 1931–1960 and 1961– 1990. This is found to increase in the Sahel by less than 5% but more markedly (up to 15%) in Western and Eastern extremes. He also examined changes in the spatial patterns of rainfall change (Sahel becomes drier), seasonality, and time series (Sahel shows recent persistence). He did not however investigate further the impact upon drought occurrence as this is assumed by the persistence of the recent downward trend, nor the appropriateness of using the 1961–1990 base period although it is noted that a wet phase occurred prior to 1961. The coefficient of variation, calculated for 30 year periods, increases for much of the 20th Century (Figure 3) in Continental Sahel from a low of 22.3 to a high of 25.3%. This modest increase is largely an artifact of lower rainfall totals in latter years as the rainfall standard deviation, determined for the same 30 year periods, decline. Hence while the relative variability slightly increases the absolute variability has reduced over the period 1931 to 1990 and 1961–1990 is a period of lower standard deviations but higher coefficients of variation. Given that absolute variability has declined one might assume that the period 1961–90 will provide a more reliable estimate of average rainfalls in this region. However, Figure 4 shows the effect of changing the time period over which the mean rainfall is determined. Taking 1990 as the starting point and averaging over 5 to 60 years, i.e., initially 1986 to 1990 and finally 1931 to 1990; the average rainfall continues to change as the averaging period lengthens until 40 years of observations are included, when a ‘stable’ mean Sahelian rainfall is reached of around 600 mm. This casts some doubt over the use of the 1961–1990 period to represent average Sahelian climate conditions and suggests 1950 to 1990 is more appropriate. Nevertheless we will persevere with this base period as it is now recommended by the WMO (Hulme, 1992). What is abnormal, what is drought? Figure 5 shows the recent pattern of negative rainfall anomalies that has given rise to reports of widespread drought, where a standardised rainfall anomaly for year t (SRAt ) is SRAt = [Pt − Pm ]/σ,

(1)

Pt = precipitation total; Pm = mean precipitation over a specified period, normally 30 years;

303

Figure 2. Continental Sahel mean rainfalls for 30 year averaging periods. Start date refers to the initial year in the 30 year averaging period, i.e., 1931 covers the period 1931 to 60, and 1961 covers the period 1961 to 1990.

Figure 3. Continental Sahel variation of annual rainfall since 1931, using the same meaning of start date as for Figure 2.

304

Figure 4. Mean annual rainfall for the Continental Sahel region, calculated over different averaging periods, starting in 1990. Hence an averaging period of 60 refers to the years 1931 to 1990; while an averaging period of 5 refers to the years 1986 to 1990.

Figure 5. Standardised rainfall anomalies for Continental Sahel calculated for two base periods, 1961–1990 and 1931–1960 (histograms for 1931–1960 have not been plotted because of potential confusion).

305 Table 1. Probabilities for different standardised rainfall anomalies (SRA). (see text for explanation of SRA, magnitudes are in terms of standard deviations hence rainfalls with different annual totals can be directly compared) Standardised rainfall anomaly

Probability of occurence

Drought occurence

Less than −2.00 Less than −1.65 Less than −1.50 Less than −1.28 Less than −1.00 Less than −0.84 Less than −0.50

0.023 0.050 0.067 0.100 0.159 0.201 0.309

Extreme Extreme Severe Severe Moderate Moderate No drought

σ = Standard deviation of precipitation over same averaging period. Seven year moving means have been drawn for SRAs based on two base periods of 1931–1960 and 1961–1990. It is clear that adoption of the more recent period reduces values by 0.72 of a standard anomaly. That is there is a significant reduction in the perception of meteorological drought occurrence and severity through changing the base period. The magnitude of this change varies spatially as it is a function of total rainfall but this averages to a difference of 94 mm of annual rainfall across Continental Sahel (the mean standard deviation in annual rainfall for continental sahel is 132 mm). For comparison the variation of annual rainfalls in the country of Niger for the period 1961 to 1990 have been calculated which show that rainfalls have been less variable over this period. Irrespective of the total amount of annual rainfall, all stations show a confidence interval (95% level of significance) of less than ±64 mm with a mean of ±42 mm. Thus the effect of changing the base averaging period over which rainfall anomalies are calculated produces a difference that is almost twice the variation observed in rainfall over the 1961–1990 period. But what are the implications for drought assessment? Figure 6 shows the rainfall anomalies based only on the 1961–1990 base period. The high occurrence of negative anomalies is evident since the late 1960s but are these all drought years? It was reported above that a negative departure of two standard deviations (equivalent to two standardised rainfall anomalies) in annual rainfall has been suggested as a drought year. But at no time is this value reached for standardised anomalies averaged over the Continental Sahel region, raising two questions. Has the spatial averaging masked local drought conditions and is the choice of two standard deviations appropriate? In the analysis of river flows, low flows are demarcated by use of a 95% threshold (Claussen, 1995; Gustard et al., 1992). Table 1 shows the expected probabilities of different rainfall anomalies based on the assumption the data are normally distributed. It seems sensible therefore to employ the following (plotted in Figure 6): • Extreme Drought SRA < −1.65 (95 percentile). • Severe Drought −1.28 > SRA > −1.65 (90 percentile).

Table 2. Drought frequencies for Continental Sahel Region based on 1931 to 1994 using two different base periods Drought Type

1961–1990

1931–1960

Extreme drought Severe drought Moderate drought 0.0 > SRA > −0.84 0.84 > SRA > 0.0 1.28 > SRA > 0.84 1.65 > SRA > 1.28 SRA > 1.65

1 3 7 12 25 9 4 3

6 6 7 26 18 1 0 0

• Moderate Drought −0.84 > SRA > −1.28 (80 percentile). • No Drought SRA > −0.84 (i.e., no drought 8 years in 10). It should be stressed that these values have no meaning other than an expected frequency of occurrence but this simple exercise at least illustrates the excessively high threshold of using two standardised anomalies to demarcate drought conditions. Figure 7 and Table 2 show the results of using these thresholds. During the period 1930 to 1994, severe drought or worse occurred in four years, moderate drought or worse in 11 years. This appears to be a reasonable frequency of 1.7 years in 10 of meteorological drought and severe conditions around 1 year in 20. These thresholds seem appropriate. The results reveal no significant relationship between aridity and drought frequency although one can detect a tendency of higher drought frequencies in drier regions. The relationship between different types of drought severity and annual rainfall has been investigated with similar results (see Table 3). The majority of stations observe 10 to 15 droughts and yet many record less than 10 and some more than 20 droughts. We can now re-examine Figure 6 using these drought thresholds. The wetter decades before 1970 are evident but even though rainfalls have fallen, inevitably, since then meteorological drought has not been persistent. The incidence of drought is much higher in recent years and the severity of drought has increased but many recent years are well within the range of what can be expected for natural variability around the mean. There are however several curious features of the data. Nicholson and Palao (1993) reported drought in the 1940s yet there is no evidence in the rainfall except for a number of very wet years which may have influenced perceptions at the time. There is general agreement about the intensity of drought in the early 1970s and yet this period does not now seem abnormally dry. Perhaps of greater significance are the line graphs for the individual countries comprising Continental Sahel. Conditions appear to have been far worse in Mali than they did in Niger. Figure 8 shows a compilation of drought frequencies for these countries. The lack of meteorological drought in Niger compared to Mali is clearly visible. Table 3 shows the results for individual stations and no spatial patterns are evident.

306

Figure 6. Standardised rainfall anomalies (1961–1990 base period) for Burkina Faso, Mali and Niger with thresholds for moderate and severe drought (see text for explanation). The histograms show anomalies which are the mean of all the rainfall stations employed.

Figure 7. Drought frequencies for Continental Sahel between 1931 and 1994 using two different base periods.

307 Table 3. Drought frequencies (%) for Continental Sahel Region during period 1931 to 1994 Annual Rainfall

Country

Station

All drought types (%)

Extreme drought (%)

Severe drought (%)

Moderate drought (%)

12 111 198 342 362 399 409 478 493 538 546 597 796 115 142 158 186 186 221 275 336 368 459 463 463 480 493 502 554 569 629 635 690 893 929 953 1117 1124 488 595 761 780 801 830 872 1025 1055 1097

Niger Niger Niger Niger Niger Niger Niger Niger Niger Niger Niger Niger Niger Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Mali Burkina F. Burkina F. Burkina F. Burkina F. Burkina F. Burkina F. Burkina F. Burkina F. Burkina F. Burkina F.

Bilma Agadez N’Guigmi Maine-Soroa Tahoua Tillabery Zinder Birni N’konni Maradi Magaria Niamey Torodi Gaya Kidal Bamba Tombouctou Gao Goundam Menaka Niafunke Hombori Nara Mourdiah Nioro du Sahel Mopti Bandiagara Yelimane Ke macina Djenne Markala Kayes Segou San Koutiala Kita Bamako/Senou Bougouni Sikasso Dori Ouahigouya Koudougou Ouagadougou Dedougou Fada N’gourma Boromo Bobo/Dioulasso Gaoua Banfora agric.

27 11 11 19 17 17 17 17 16 25 17 23 14 22 33 13 16 20 19 28 20 11 22 11 16 22 34 22 23 36 17 22 22 13 19 13 20 9 17 11 23 14 20 17 16 16 17 22

0 2 0 8 2 2 2 5 0 11 3 11 8 2 23 2 3 16 8 20 9 5 14 3 0 11 20 13 16 28 6 6 5 3 8 2 3 3 3 0 11 3 13 5 5 3 5 13

0 3 2 2 6 2 5 5 8 5 2 3 0 9 2 3 5 0 3 2 8 3 3 3 5 0 5 2 2 2 2 5 5 2 0 0 8 2 3 6 2 3 0 5 3 2 3 3

27 6 9 9 9 14 11 8 8 9 13 9 6 11 8 8 8 5 8 6 3 3 5 5 11 11 9 8 6 6 9 11 13 8 11 11 9 5 11 5 11 8 8 8 8 11 9 6

308

Figure 8. Comparison of drought frequencies between 1931 and 1994 for countries in Continental Sahel using thresholds given in the text.

Figure 9. Annual millet production for Burkina Faso, Mali and Niger (from http://apps.fao.org/).

309

Figure 10. Accumulated rainfall anomalies for countries in Continental Sahel from 1931.

Contrast stations where drought has been most frequent such as Magaria (538 mm Niger), Bamba (142 mm Mali) and Koudougou (761 mm B. Faso); with stations where drought is much less common, e.g., N’Guigmi (198 mm, Niger), Nioro (463 mm, Mali) and Ouagadougou (780 mm, B. Faso). Table 3 then illustrates the spatially variable nature of annual rainfalls in this semi arid region. Perhaps now we are beginning to see why conflicting claims can arise from analyses of meteorological drought occurrence in the Sahel. First of all the averaging period that is employed can significantly alter the number of droughts recorded with an almost doubling of drought years by using 1931–1960 compared to 1961–1990 base years. The aggregation of stations, from a region that others have argued is climatologically homogeneous can in places mask local conditions and exaggerate the incidence of drought. Note that 11 moderate or worse droughts were found in Continental Sahel, for the period 1931 to 1994 yet the frequencies for Mali, B. Faso and Niger were 11, 9 and 5, respectively. This surely illustrates the danger, for the purposes of meteorological drought assessment, of treating the Sahel as a homogeneous entity even within the boundaries designated by Ba et al. (1995) and Nicholson and Palao (1993). Furthermore, Figure 9 shows that millet production, the major rainfed crop in the region, has steadily increased during this period thus having a complementary relationship with negative rainfall anomalies. The same is true for total cereal production hence the above analysis of meteorological drought has failed to explain the observed changes in food crop production. While

yields have not been increasing in the same manner the point to be made is that perceptions of meteorological drought, based solely upon a statistical analysis of precipitation, cannot provide a reliable assessment of drought occurrence, a conclusion also reached by Bruins and Berliner (1998).

Conclusions Annual rainfalls, aggregated across the Sahel, have declined since the 1970s. This has promted assertions of widespread and persistent drought in the region. Our investigation of meteorological drought in the Continental Sahel region (Mali, B. Faso and Niger) has revealed two major flaws in previous assessments of drought for this area concerning the use of statistical definitions and the spatial aggregation drought occurrence. We have demonstrated that even in a region that is considered to be climatologically homogeneous that the frequency of drought varies from the regional aggregation, to national groupings, to individual stations. Any statement that refers to ‘Sahel drought’ is then to be approached with caution as at best it is only likely to be accurate for part of the area concerned (see Agnew and Chappell 2000 for a geostatistical evaluation of the dangers of aggregating rainfall data in the Sahel). Statistical assessments of drought are based on identifying abnormal deviations from conditions established over a 30 year base period. Too many previous studies have used the recent downward trend in rainfalls to present all recent decades as being drought affected rather than trying to define

310 conditions of abnormality. Hence the notion of persistent drought in the Sahel has become a paradigm for climatologists with too little critical evaluation of rainfall data that do not fit this pattern. Figure 10 shows a plot of accumulated rainfall anomalies from 1931. It is interesting to note that accumulated deficits are still positive and at no time since 1940 are accumulated deficits negative. The trend towards lower rainfalls is clearly evident but the widespread concern for drought in this region must be seen as a relative change in moisture availability, from high positive accumulated anomalies in the 1960s. In a period of such change, a persistent increase followed by a persistent decrease in rainfalls, the ability of any statistical analysis to demarcate drought and nondrought years must be questioned. Hence it is suggested that areally averaged standardised rainfall anomalies should be not be employed for drought assessment in this region. Alternative approaches include impact assessment where the water availability for a particular activity, e.g., rainfed agriculture is examined, or use of thresholds, e.g., the drought years of 1974 and 1984 are widely reported as being severe. These approaches have the advantages of being independent of climate change and base periods while focusing on impacts rather than relative change.

Acknowledgements The monthly rainfall totals were generously provided by the Climate Research Unit at the University of East Anglia. We would also like to acknowledge the useful comments made by the reviewers of this paper and the cartographic work of Nick Scarle, School of Geography, Manchester University.

References Agnew C.T., 1982: Water availability and the development of rainfed agriculture in S.W. Niger. Transactions of the Institute of British Geographers 7: 419–457 Agnew C.T., 1989: Sahel drought, meteorological or agricultural? International Journal of Climatology 9: 371–382. Agnew C.T., 1990: Spatial aspects of drought in the Sahel. Journal of Arid Environments 18: 279–293. Agnew C.T., 1991a: Evaluation of a soil water balance model for the analysis of agricultural drought in the Sahel. In: Sivakumar M.V.K., Wallace J.S., Renard C. and Giroux C. (eds), Soil Water Balance in the Sudano-Sahelian Zone, IAHS Publication 199, Wallingford, pp. 583–592. Agnew C.T., 1991b: Disaster in the arid realm. In: Slater F. (ed.), Societies: Choices and Environments. Collins Educational, London, pp. 56–79. Agnew C.T., 1995: Desertification, drought and development in the Sahel. In: Binns A. (ed.), People and Environment in Africa. J. Wiley & Sons, Chichester, pp. 137–149. Agnew C.T. and Warren A, 1996: A framework for tackling drought and land degradation. Journal of Arid Environments 33: 309–320. Agnew C.T. & Anderson E., 1992: Water Resources in the Arid Realm. Routledge, London. Agnew C.T. and Chappell A., 2000: Desiccation in the Sahel. In McLaren S. and Kniveton D. (eds), Linking Climate Change to Land Surface Change: Proceedings of TIBG-RGS Conference, Leicester, January 1999. Kluwer Academic Publishers, Dordrecht (in press). Ahlcrona E., 1988: The impact of climate and man on land transformation in central Sudan. Applications of Remote Sensing, Meddelanden fran Lunds Universitets Geografiska Institutioner, Avhandlinger 103.

Ba M.B., Frouin R. and Nicholson S.E., 1995: Satellite derived inter-annual variability of West African Rainfall during 1983–88. Journal of Applied Meteorology 34: 411–431. Beran M.A. and Rodier J.A., 1985: Hydrological aspects of drought. Studies and Reports in Hydrology 39. UNESCO-WMO. Binns T., 1990: Is desertification a myth? Geography 75: 106–113. Bley J., Van der Ploeg R.R., Sivakumar M.V.K. and Allison B.E., 1991: A risk probability map for millet production in southwest niger. In: Sivakumar M.V.K., Wallace J.S., Renard C. and Giroux C. (eds) 19 Soil Water Balance in the Sudano-Sahelian Zone. IAHS Publication 199, Wallingford, pp. 571–581. Bruins H.J. and Berliner, P.R., 1998: Biocimatic aridity, climatic variability, drought and desertification: Definitions and management options. In: Bruins H.J. and Lithwick (eds), The Arid Frontier. Kluwer Academic Publishers, Dordrecht, pp. 97–116. Chappell A. and Agnew C.T, 1999: Geostatistical analysis and simulation of West African rainfall. Proceedings International Geographical Union Commission on Land Degradation and Desertification 6th Meeting. Perth, University of Western Australia, 1999, Sept. 20–28. Clausen B., 1995: Discharge data collection and analysis strategies inlow flow studies. Nordic Hydrology 26: 191–204. Cocheme J. and Franquin P., 1967: An agroclimatology survey of a semiarid area in Africa south of the Sahara. Technical Note 86, WMO, Geneva. Copans J., 1983: The Sahelian drought. In: Hewitt K. (ed.), Interpretations of Calamity. Allen and Unwin, London, pp. 83–97. D’Amato N. and Lebel T., 1998: On the characteristics of the rainfall events in the Sahel with a view to the analysis of climatic variability. International Journal of Climatology 18: 955–974. Davy E.G., Mattei F and Solomon S.I., 1976: An evaluation of the climate and water resources for development of agriculture in the SudanSahelizn zone of West Africa WMO Special Environmental Report No. 9. WMO, Geneva. Dracup J., Lee K. and Paulson E., 1980: On the definition of droughts. Water Resources Research 16 (2): 297–302. Flohn H., 1987: Rainfall teleconnections in Northern and North Eastern Africa. Theoretical and Applied Climatology 38: 191–197. Franke F. and Chasin B., 1980: Seeds of Famine. Allanheld, Osman and Co., New Jersey. Garcia R.V., 1981: Drought and Man. Volume 1. Nature Pleads Not Guilty. Pergamon, Oxford. Glantz M. (ed.), 1987: Drought and Hunger in Africa. Cambridge University Press, Cambridge. Glantz M. (ed.), 1994: Drought Follows the Plow. Cambridge University Press, Cambridge. Gordon A.H., 1993: The random nature of drought: mathematical and physical causes. International Journal of Climatology 13: 497–507. Gustard A., Bullock A. and Dixon J.M., 1992: Low flow estimation in the U.K. Institute of Hydrology Report 108, Wallingford. Hare F.K., 1984: Climate, drought and desertification. Nature and Resources (UNECSO) 20 (1): 2–8. Hulme M., 1992: Rainfall changes in Africa. International Journal of Climatology 12: 685–699. Hulme M., 1996: Recent climatic change in the world’s drylands. Geophysical Research Letters 23 (1): 61–64. Hulme M. and Kelly M., 1993: Desertification and climate change. Environment 35 (6): 39–45. International Union for Conservation of Nature 1989: Sahel Studies. IUCN, Nairobi. Janicot S., Harzallah A., Fontaine B. and Moron V., 1998: West African monsoon dynamics and eastern equitorial atlantic and pacific SST anomalies (1970–88). Journal of Climate 11: 1874–1882. Jones P.D. and Hulme M., 1996: Calculating regional climatic time series for temperature and precipitation: methods and illustrations 16: 361– 377. Kelly M. and Buchanan-Smith M., 1994: North Sudan in 1991: food crisis and the international relief response. Disasters 18 (1): 16–34. Lamb H..H., 1974 The Earth’s changing climate. Ecologist 4: 10–15. Le Houerou H.N., 1996: Climate change, drought and desertification. Journal of Arid Environments 34: 133–185. MacDonald L.H., 1986: Natural resources development in the Sahel: the role of the United Nations System. United Nations University, Japan.

311 Mainguet M, 1991: Desertification: Natural Background and Human Mismanagement. Springer-Verlag, Berlin. Mainguet M., 1999: Aridity: Droughts and Human Development. SpringerVerlag, Berlin. McKee T.B., Doesken N.J. and Kleist J., 1994: Drought monitoring with multiple time scales. Proceedings of Ninth Conference on Applied Climatology. American Meteorological Society, pp. 233–236. Nicholson S.E. and Palao I.M., 1993: A re-evaluation of rainfall variability in the Sahel. International Journal Climatology 13: 371–389. Norse D., 1994: Multiple threats to regional food production:environment, economy, population? Food Policy 19 (2): 113–148. Odingo R.S., 1992: Implementation of the plan of action to combat desertification (PACD) 1987–1991. Desertification Control Bulletin 21: 6–14. Olsson L., 1993: On the causes of famine – drought, desertification and market failure in the Sudan. Ambio 22 (6): 395–403. Palmer W.C., 1965: Meteorological Drought US Weather Bureau Research Paper 45. Washington D.C. Pritchard J.M., 1990: Africa. Longman, Harlow. Raynaut C. (ed.), 1997: Societies and Nature in the Sahel. Routledge, London. Sivakumar M.V.K., 1986: The climate of Niamey Resource Management Programme Progress Report 1. ICRISAT Sahelian Center BP 12404 Niamey. Sivakumar M.V.K., 1989: Agroclimatic aspects of rainfed agriculture in the Sudano-Sahelina zone. In: Proceedings of Workshop on Soil, Crop and Water Management Systems for Rainfed Agriculture in the Sudano-Sahelian Zone, Niamey January 1987. ICRISAT Sahelian Center, Niamey. ICRISAT, Patancheru, AP 502 324 India, pp. 17–38. Sivakumar M.V.K., and Wallace J.S., 1991: Soil water balance in the Sudano-Sahelian zone: need, relevance and objectives of the work-

shop. In: Sivakumar M.V.K., Wallace J.S., Renard C. and Giroux C. Soil water balance in the Sudano-Sahelian zone. IAHS Publication 199, Wallingford, pp. 3–10. Tannehill I.R., 1947: Drought its Causes and Effects. Princeton University Press. Thomas D.G., 1993: Sandstorm in a teacup: Understanding desertification. Geographical Journal 159 (3): 318–331. Thomas D.G. and Middleton N., 1994: Desertification: Exploding the Myth. J. Wiley, Chichester. Tickell C., 1986: Drought in Africa: impact and response. Overseas Development 102: 10. Tooze S., 1984: Sahel drought, call for joint action. Nature 307: 497. Toulmin C., 1988: Smiling in the Sahel. New Scientist 12: 69. UNEP 1992: World Atlas of Desertification, United Nations Environment Programme. Edward Arnold, London. Warren A. and Agnew C.T., 1988: An assessment of desertification and land degradation in arid and semi arid areas. Paper 2, International Institute for Environment and Development, London. Warren A., 1998: Environmental science and desertification. In: Bruins H.J. and Lithwick (eds), The Arid Frontier, Kluwer Academic Publishers, Dordrecht, pp. 117–127. Wijkman A. and Timberlake L., 1985: Is the African drought an act of God or of Man? Ecologist 15 (1/2): 9–18. Wilhite D.A. (ed.), 1993: Drought Assessment, Management and Planning: Theory and Case Studies. Kluwer Academic Publishers Boston, MA. Winstanley D.W., 1973: Rainfall patterns and general atmospheric circulation. Nature 245: 190–194. Zheng X. and Eltahir E.A.B., 1998: The role of vegetation in the dynamics of west african monsoons. Journal of Climate 11: 2078–2096.