E. Linacre and B. Geerts


Charney's hypothesis

There is a controversy about the advance of deserts in the world (1). There is a widespread belief that the Sahara desert is advancing into the Sahel region, for instance. The Sahel is a narrow band of West Africa between 15 -18° N, between the Sahara to the north and savannah (grass and open forest) and equatorial forest to the south. It extends from Senegal at the coast at about 15° W, across Mali and Niger, to about 15° E. It receives rainfall during a short but active wet season, from late June to mid September. It is covered by grassland and supports a pasture-based society which traditionally moved meridionally following the rains. Its northern limit may be defined by the 200 mm/a isohyet.

Is the Sahara extending into the Sahel? And if so, is this because of fluctuations of rainfall (total amount, rainfall intensity, duration of wet season, …) or is it largely the result of human activities, such as overgrazing or the removal of trees for firewood? There are also the questions: Do deserts create droughts? Do droughts create deserts? In other words, is there a positive climate feedback, which accelerates land degradation? A now classic paper by Jules Charney in 1975 (2) speculated that overgrazing in the Sahel leads to less vegetation, which raises the ground’s albedo, so that less solar radiation is absorbed and the Earth’s surface becomes cooler. It should be noted that the atmosphere above the Sahara experiences continues radiative cooling, because the dry air, free of clouds, absorbs very little of the longwave radiation upwelling from the ground. This radiative cooling is naturally compensated by subsidence heating, and the subsidence sustains the dry air, cloudlessness, and arid surface conditions. According Charney, overgrazing would enhance the radiative loss, which would foster subsidence within the troposphere, leading to drier conditions in the Sahel, and therefore less plant growth during the wet season. Less vegetation means a higher albedo. So we have positive feedback and a self-aggravating process, culminating in desertification, a process of land degradation that destroys its productivity. Charney’s hypothesis was supported by experiments with a very simple GCM, in which he changed the surface albedo from 20% to 30% (3).

The problem of overgrazing in the Sahel is as acute now as it was in the 1960's, yet there is no clear rainfall trend in the Sahel. The period 1930-'60 was slightly wetter than 1960-'90 in most parts of the Sahel. More significant than any trend is the occurrence of dry and wet periods, each lasting several years. The Sahel enjoyed a notably wet decade in the 1950’s, which was followed by a drought in the 70’s and 80’s (1). However, land productivity was fully recovered around 1990. So Charney's hypothesis cannot be confirmed.

To reject Charney's hypothesis, one needs to examine whether the albedo has really increased in the Sahel. Satellite-estimated albedo of the Sahel between 1983-1988 was about 35% in (dry) January and 31% in (wet) July, a difference of about 4%. The seasonal variation greatly exceeded any overall change during the period. This points to there being no irreversible change towards desert. It is concluded that the formation of desert is not a single self-aggravating process, but is complex, reflecting changes of both climate and human activities. There may be incomplete recovery after a dry period, or changes in the composition of the vegetation.


Desertification trends are evaluated by means of the ‘normalised difference vegetation index’ (NDVI). This index, based on satellite data, quantifies the amount of vegetation (1). It is a figure for the ‘green-ness’ of the surface, the ratio of the measured reflectances of red light (i.e. 0.55-0.68 micron wavelength) and near-infra-red (0.73-1.1) in solar radiation. NDVI is strongly correlated with the biological productivity of an area. In the Sahel there is close agreement of the shifts of NDVI and rainfall boundaries during 1980 - 1995, i.e. the NDVI/rainfall relationship remained about constant. In other words, there was no progressive ‘march’ of desert over more fertile areas, no one-way ratchet effect due to deserts causing droughts.

A question related to the question about deserts causing drought, is that of large lakes inducing rain. Nicholson et al. (1) mentioned proposals to flood the Kalahari desert of Botswana to increase the rainfall regionally. The topic was discussed indirectly in Notes 10.I and 10.J, where reasons are given for expecting no influence of forestation on rainfall, unless done over large areas; even then, the enhanced rainfall is insufficient to support the manmade forest.

A separate effect of desertification, apart from any possible influence on rainfall, is an increase of soil erosion and dust storms (4), as shown in Fig 1. This shows the close relationship between rainfall and dust.

Fig 1. The frequency of dust storms at Gao (in the Sahel), compared with annual rainfall anomalies (from (1), after (4)). The anomaly unit is the regionally averaged departure from the long-term mean, divided by the standard deviation. The dust occurrence is expressed in terms of the number of days when there is dust haze.




(1) Nicholson, S.E., C.J. Tucker and M.B. Ba 1998. Desertification, drought and surface vegetation: an example from the West African Sahel. Bull. Amer. Meteor. Soc., 79, 815-29.

(2) Charney, J.G. 1975. Dynamics of deserts and drought in Sahel. Quart. J. Royal Meteor. Soc, 101, 193-202.

(3) Xue, Y. and J. Shukla 1993. The influence of land surface properties on Sahel climate. J. Climate, 6, 2232-45.

(4) N’Tchayi, M.G., J.J. Bertrand and S.E. Nicholson 1997. The diurnal and seasonal cycles of desert dust over Africa north of the equator. J. Appl. Meteor. 36, 868-82.