Desertification and climate change

Mick Kelly and Mike Hulme address the complex and often uncertain links between climate change, prolonged aridification or desiccation, and desertification. They pay particular attention to the case of the African Sahel.

The authors are staff members of the Climatic Research Unit at the University of East Anglia (Norwich, UK).

The definition of desertification adopted by the United Nations Conference on Environment and Development in 1992 is land degradation in arid, semi-arid and dry sub-humid areas resulting from various factors including climatic variations and human activities. This definition cites climate variation as a direct causal factor and it implicitly links climate change and the assessment of the extent of desertification.

Since arid, semi-arid and dry sub-humid areas are climatically defined, any change in climate which results in an expansion or contraction of these areas will alter the extent of the area in which desertification can be considered to occur.

For example, if an arid area converts to hyper-arid because of climate change, then the area in which desertification may occur will decrease. Hyper-arid areas are not included in the accepted definition. If a humid area converts to sub-humid, then the potential area within which desertification may occur will increase.

Determining the precise contribution of climate variation to the problem of desertification is not, however, an easy matter.

Climate change does alter the frequency and severity of drought in various parts of the world and can cause desiccation. It does not necessarily follow, though, that drought and desiccation will, by themselves, induce, or even contribute to, desertification in dryland regions. Whether or not this occurs depends upon the nature of resource management in these areas. Against a backdrop of management failure, climate change can certainly aggravate dryland degradation.

Photo: Mike Hulme

Separating out the interrelated impacts of climatic and human factors may be difficult but some progress has been made. For example, a satellite index of active vegetative cover for the Sahara has been derived by Tucker and his colleagues at the NASA Space Flight Center in Maryland. The index reveals marked interannual variations in the extent and quality of surface vegetation in this dryland area.

A substantial proportion of this variation is attributable to climate effects, specifically rainfall variations. Removing this component reveals a progressive increase of 41,000km ² a year in the area of the Sahara Desert during the 1980s.

A satellite index of the extent of the Sahara (circles), a rainfall deficit index for the region (bars), and the residual trend in the extent of the Sahara when rainfall effects are removed (squares).

This trend could be the result of the cumulative impact of a series of dry years on vegetation recovery. Alternatively, it may well be due to deterioration in vegetation cover caused by human activity.

Clearly, the relative contribution of human activity and climate variation to dryland degradation will vary from region to region and from time to time. Identifying the relative role of these factors in order to identify the most appropriate response in any particular situation is a pressing challenge.

The assessment is complicated by the fact that desertification itself may cause climate change. By modifying surface characteristics, dryland degradation can lead to reductions in surface soil moisture and so make more energy available to increase air temperature in the areas most affected. How widespread are the resulting changes in climate likely to be?

Balling Jr of Arizona State University has recently suggested that the increase in surface air temperature in desertified regions caused by changes in land cover characteristics is sufficient to significantly contaminate the global-mean temperature record.

Balling may be correct in principle but we believe that he has overstated his case and contend that his analysis provides no convincing evidence that a substantial desertification signal is present in the global- mean temperature record.

First, his identification of areas which are severely desertified and non-desertified is derived from a map of desertification prepared for the United Nations Conference on Desertification back in 1977 and now superseded on the basis of more accurate data collected for the recent UNEP assessment.

Second, the warming signal of +0.5 ° C over a 100-year period identified by Balling in his comparison of trends in desertified and non-desertified areas is based on a small sample of the total area affected by land degradation.

Balling extrapolates from this limited sample to the global scale by suggesting that over 30% of the global land surface, representing about 90% of the total dryland area, is prone to this desertification warming signal. The global assessment of desertified areas by UNEP in 1992 estimated, however, that only 20% of dryland regions were seriously degraded. This suggests that only 6% of the total global land surface may exhibit a strong desertification warming signal.

Finally, the differential warming between desertified and neighbouring non-desertified areas found by Balling may indicate that these areas differ in their sensitivity to climate variability and may not all be attributable to desertification itself.

The warming of desertified areas may well have been great enough to produce a measurable effect on global-mean temperature over recent decades. But the influence appears limited compared to that of factors such as enhancement of the greenhouse effect. We consider that the effect of desertification on global-mean temperature is unlikely to have exceeded +0.05 ° C a century.

We return to the question of the influence of desertification on climate later in this article when we discuss the specific case of the Sahel. First, though, we consider a more certain link between dryland degradation and global environmental change.

There is no doubt that desertification plays a role in altering the sources and sinks of greenhouse gases, contributing to global warming.

Dryland degradation is likely to limit the local carbon sink by reducing the carbon stored in ecosystems and, as vegetation dies and soil is disturbed, carbon emissions will increase. Emissions of other greenhouse gases might also be affected by desertification. For example, there may be greater methane production in poorly fed cattle in degraded areas. On the other hand, dry soils are methane sinks so desertification might reduce atmospheric concentrations.

The net contribution of dryland degradation to global warming is difficult to quantify at this time. It is reasonable to conclude, though, that the global carbon release associated with dryland degradation is not more than a few per cent of the total greenhouse forcing associated with the major greenhouse gases.

Like many other processes, the role of desertification is not substantial on a global scale compared to, say, the contribution of fossil fuel use. Nevertheless, in terms of the carbon budgets of the countries most affected, slowing or reversing the process of desertification could play a major part in reducing their national contribution to global warming, offsetting emission growth in other sectors.

Surface air temperature difference (degrees Celsius) between five years with low Sahelian rainfall and five years with high Sahelian rainfall. There is a marked temperature gradient between the northern and southern Atlantic Ocean.

We now consider the case of the African Sahel, the area where the linkages between desertification and climate change are most clear.

The recent 25-year period of desiccation in the African Sahel represents the most substantial and sustained change in rainfall for any region of the world within the period of instrumental measurements.

Current ideas about the causes of the desiccation have crystallized around two central themes:

  • internal feedback mechanisms within Africa associated with land cover changes, such as desertification; and
  • global circulation changes associated with particular patterns of heat distribution (surface temperature) within the oceans.

The idea that modification of land cover characteristics in dryland regions could affect regional rainfall was first proposed in the 1970s by Otterman and by Charney.

An initial change in land cover characteristics occurs associated with desertification. This may involve vegetation change or removal or a deterioration in soil quality, factors which affect the amount of soil moisture.

The land cover change then accelerates as rainfall is suppressed. This in turn increases moisture stress on vegetation, lowers soil moisture levels and further reduces rainfall amounts, closing the feedback loop.

If land cover changes can account for a significant proportion of the rainfall decline in the Sahel, then it is the complex matrix of processes leading to desertification in recent decades that is responsible for the desiccation.

Climate modelling experiments have clearly shown that large-scale conversion of land surface characteristics can generate climate change on the local to regional spatial scale.

There is, however, a real difficulty in attributing the recent desiccation in the Sahel to land cover changes on the basis of these studies:

  • first, the observational evidence for the marked changes in surface albedo in dryland regions introduced into most model experiments remains weak; and
  • second, the observed changes have been localized in extent and often short-term rather than widespread and sustained as assumed in the model studies.
These experiments are important in understanding how the various physical systems are interlinked. Because of their lack of realism, though, it is dangerous to draw from their results the conclusion that observed land cover changes have accounted for observed rainfall changes in the recent past. A surer way to proceed is with so-called simulation experiments based on real-world data. These experiments use observed changes in surface characteristics, such as the actual soil moisture conditions in a given year, and then examine whether the model reproduces the observed rainfall anomaly of that year.

The most impressive set of simulation experiments has been completed at the UK Met. Office. The conclusion from these experiments is that ocean temperature forcing dominates the effects of land surface moisture feedback. Land surface feedback can play a part in generating self-sustaining drought but the role of this mechanism is secondary to that of variability within the wider climate system.

These experiments underline the importance of a link between Sahelian drought and sea surface temperatures in the neighbouring Atlantic Ocean (and further afield) first proposed in the late 1970s.

Recent work by Folland and his colleagues at the UK Met. Office has confirmed that higher temperatures south of the equator and lower temperatures north of the equator are associated with lower rainfall over much of north Africa.

The physical basis of this relationship appears to lie in a disturbance to the atmospheric circulation in the Atlantic/Africa sector induced by the underlying sea surface temperature pattern. This circulation change then affects Sahelian rainfall.

The outstanding question concerns the initial cause of the temperature contrast north and south of the equator.

The ocean temperature pattern may well be a manifestation of natural climate variability. For example, it could be the result of a reduction in the northward transport of heat in the Atlantic Ocean, possibly caused by freshening of the surface waters of the northern North Atlantic. Further investigation of this possibility is needed.

But there is equally convincing, though equally speculative, evidence to suggest that there may be a link with global warming. The very different effect of sulphur emissions on warming rates in the Northern and Southern Hemisphere and/or the variable ocean response (which is slower in some areas than in others) could have created the observed temperature contrast.

While the link between Sahel rainfall and the temperature contrast across the equator is well-established, the evidence concerning the initial cause of the temperature contrast, whether that be natural or human in origin, must be considered circumstantial.

Varying degrees of uncertainty surround the linkages between climate change and desertification and desiccation. Further work is needed to clarify the relative role of the various factors. Nevertheless, uncertainty should not be allowed to obscure the fact that there are intrinsic links that cannot be ignored.

Flowchart showing linkages between desertification, climate change, management failure and related factors.

The flowchart summarizes the complex links between the various factors considered in this article. The solid lines indicate causal relationships about which we can be confident. The dotted lines indicate relationships about which there is uncertainty with regard to their precise importance or even existence.

The main linkages are:

  • Desertification is, first and foremost, the result of resource management failure.
  • This failure is the product of both local factors (such as population pressure and inequity) and external factors (the state of the global economy, commodity prices, the burden of debt, and so on).
  • The process may be aggravated by climate change: in particular, by prolonged aridification or desiccation.
  • Desiccation itself could be the result of natural mechanisms with the climate system such as the influence of ocean temperature patterns.
  • Desiccation may also be caused by desertification, through surface-atmosphere interaction, or it may possibly be the result of global warming.
  • Finally, dryland degradation contributes to global warming through its effect on the sources and sinks of greenhouse gases.

It is important to note that the relative importance of these various factors may change in the future. For example, if there is sustained degradation of a substantial dryland area, the significance of desertification as a causative agent may well increase. On the other hand, climate modelling suggests that rainfall over the Sahel may decrease further as global warming develops.

Recent model simulations of the effects of global warming indicate increased rainfall in most areas. But, over the Mediterranean, northern Africa and a large part of the Sahel, annual rainfall decreases. The reduction is most marked over the southwestern margins of the Sahara, Mauritania and parts of northern Mali and Niger.

Predicting the regional response to global warming is not a simple matter. But, if this is an accurate reflection of the impact, the role of greenhouse gas emissions in reducing rainfall in the Sahel could become a major factor in the future.

Uncertainties concerning the underlying causes of desertification have already hampered previous international efforts to combat the problem. It is inevitable that negotiations regarding the desertification convention will be rendered more complicated because of these same uncertainties.

Improving understanding of the nature and causes of dryland degradation must be considered a high priority if the international response to desertification is to be effective.

Further information

Mick Kelly and Mike Hulme, Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, UK. Email:

On the Web

A listing of theme sites covering drylands and desertification is available in the Cyberlibrary.


This article is based on a briefing document produced for the British Overseas Development Administration (ODA). The views expressed are those of the named authors alone and should not be taken to represent the views of the ODA. A more detailed discussion of the same issues can be found in an article in Environment, July/August, 1993.