An alternative explanation for the increase in dust event frequency is that changes have occurred in the atmospheric mobilisation and transport processes which are manifestations of the regional synoptic climatology. Bell and Lamb (1994) and Lamb et al. (1998) associate drier conditions in the Sahel with a decrease in the frequency of intense, well-organised convective disturbances, otherwise known as squall lines or disturbance lines (DLs). These features of the regional climate occur in the spring and summer, and are responsible for most of the rainfall in the region. As the number of intense events has decreased, the frequency of weak, disorganised disturbances has increased. Such weak events may well mobilise significant quantities of dust which is not subsequently removed from the atmosphere by rainfall washout (wet deposition). Thus an increase in the proportion of weak DLs is likely to result not only in decreased rainfall, but also in enhanced concentrations of atmospheric dust and dust transport in spring and summer, and in an increase in the frequency of dust mobilisation events which are not associated with rainfall. N'Tchayi et al. (1997) describe the increase in dustiness in the Sahel as a year-round phenomenon, and also state that the season of maximum dustiness has shifted in some locations (such as Gao in eastern Mali) from spring to summer. Prospero and Nees (1986) describe increased summer dust transport to the western Atlantic in dry years; such transport is generally at a maximum in summer.
The relationship between summer dust concentrations and the type of
atmospheric variability associated with DL occurrence was examined for
the period 1984-1993 using IDDI and NCEP/NCAR reanalysis data.
Gridded daily wind data (at a resolution of 2.5 degrees latitude by 2.5 degrees longitude) were pooled for the period July-September (JAS) for each year from 1984-1993, such that each grid-square was represented by a timeseries of 92 wind values. For each year, the standard deviation of each series was calculated, resulting in an annual field of standard deviation (sigma) values. Each sigma value represents the variability in the summer zonal wind within a particular grid-cell. 162 grid cells were analysed, representing the area 8.75 - 23.75 degrees north; 21.25 degrees west - 46.25 degrees east. [This area was chosen as it includes the zonal range of the African Easterly Jet (dynamically associated with the generation of easterly waves) and the genesis regions and trajectories of the DLs that are responsible for most Sahelian rainfall and for dust mobilisation (Tetzlaff and Peters, 1988; Rowell and Milford, 1993; Diedhiou et al., 1998b)]. A series of annual box plots was then plotted in order to visualise the distribution of sigma values for each year, in terms of their median, maximum, minimum and upper and lower quartile values (Figure 1).
Mean July-September (JAS) IDDI values were averaged over the region
10 - 20 degrees north and 18 degrees west - 45 degrees east (encompassing
the Sahel and southern Sahara); the annual mean IDDI values were plotted
alongside the standard deviation box plots (Figure 1), and were also correlated
with the various statistics (e.g. median, lower quartile) representing
the sigma values.
There is a striking correlation (r = 0.94, significant at the 0.01 per cent level) between the mean JAS IDDI values for the Sahel and southern Sahara and the variability of the 700 hPa zonal wind as represented by the median sigma values in Figure 1. High median values will result from the frequent passage of easterly waves; high maximum values may be caused by a single intense wave event. The mimimum and maximum sigma values are poorly correlated with both the median and IDDI values, indicating that the dustiest summers are those in which the most easterly wave activity occurs (as opposed to those containing the strongest or weakest wave events). The dustiest summer occurred in 1988, the wettest year in the 1984-93 period. Correlations are significant between IDDI values and rainfall (r = 0.66, significant at the 3 per cent level), and between between the median sigma values and rainfall (r = 0.67, significant at the 3 per cent level). While results from correlations between such short timeseries should be treated with caution, the very high correlation between the IDDI and the median sigma values, and the existence of a plausible physical relationship, suggest that the primary determinant of summer dust production in the Sahel and southern Sahara is the frequency of easterly waves, which generate the convective disturbances associated with both rainfall and dust production. The fact that this correlation is higher than that between the median sigma values and rainfall suggests that easterly wave activity is more likely to generate dust than to generate rainfall, and supports the hypothesis that such activity is often associated with the formation of DLs which mobilise and transport dust but which are too weak to produce precipitation.
Figure 1: Distribution of standard deviations representative of easterly wave activity (each standard deviation represents the variability in the daily 700 hPa zonal wind values over JAS within a 2.5 x 2.5 degree grid cell), plotted alongside mean JAS IDDI values spatially averaged over the Sahel and southern Sahara.
The above relationships hold if the same analysis is performed for August alone (r = 0.75, significant at the 1 per cent level, between IDDI and sigma values), but breaks down if the period is extended to include the months from May to October (a much more liberal definition of the wet season, containing the monsoon onset months).
These results support the notion that dust production is determined predominantly by meteorological processes, at least in the summer months. However, it is unclear whether the relationships between dust, easterly wave activity and rainfall would be the same in a wetter regime such as that prevalent throughout the 1950s and much of the 1960s. An extension of the IDDI dataset to the present day will facilitate further investigation of these relationships over a period extending into the 1990s, during which there has been some amelioration of Sahelian aridity. It is suspected that, during periods of more abundant rainfall, variability in the 700 hPa zonal wind will be more closely related to rainfall amounts and less closely related to dust concentrations/IDDI values, although this is highly speculative. Increased sensitivity of dust concentrations to easterly wave activity could be due either to changes in the land surface, or to changes in the type of DLs associated with easterly waves as described earlier in this section. As there is no convincing evidence for large scale changes in the land surface (particularly so in the extremely sparsely populated northern half of the area over which the IDDI data were averaged), the most convincing explanation for increases in summer dust production is one involving a decrease in the ratio of dust removed from the atmosphere by rainfall to dust mobilised from the land surface by wind and carried aloft by convective disturbances. It is proposed that at least some of the increase in annual dust event frequency may be explained in terms of changes in the atmospheric mobilisation, transport and removal mechanisms, and that changes in the relative strengths of these mechanisms are a manifestation of regional climate change. While there is likely to be some modulation of dust production by the coupled parameters of rainfall and vegetation cover, the assumption that dust production has increased primarily as a result of changes in land-surface properties must be questioned.