and late rainy-season months. High-frequency forcing is also stronger in June-July than in August-September. This observation is relevant because one hypothesized factor in the feedback, the African Easterly Jet (see below), is operative mainly in the latter period.
The potential feedback mechanisms involve the effect of such factors as soil moisture and dust generation on surface fluxes and atmospheric heating. The dynamic link to the Sahel could be the mid-level African Easterly Jet. The African Easterly Jet is a consequence of the meridional temperature gradient in the region, which in turn is produced by the strong thermal contrast between the dry Sahel-Saharan region and the forests and Atlantic Ocean to the south. It is analogous to the mid-latitude westerly jet stream, in that it provides the instability and energy for the development and maintenance of rain-bearing systems in the Sahel. It has a maximum around 650 or 700 mb, a level where the temperature structure is strongly affected by surface-generated dust. The intensity of the dust, in turn, is highly inversely correlated with rainfall in the Sahel.
The contrast in surface fluxes between wet and dry years (Lare and Nicholson, 1993) is sufficiently large to produce the observed changes of jet intensity between wet and dry years (Newell and Kidson, 1984). These in turn are sufficiently large to influence the jet's dynamics, and are thus capable of altering such variables as the number, size, and intensity of disturbances in the region.
Numerous studies have firmly established that:
decadal-scale rainfall fluctuations have been quite marked in Sahelian Africa;
these fluctuations tend to have a continental spatial scale (although less extreme and persistent than in the Sahel) with distinct teleconnection patterns;
the characteristics of modern decadal anomalies are also apparent in earlier centuries; and
the 1950s and, to a lesser extent, 1960s were continentally "wet," while the 1980s and, to a lesser extent, 1970s have been continentally "dry."
Among the major areas in which research is needed are the causes of the decadal-scale fluctuations of African rainfall, particularly the continentally coherent patterns; the role land-surface feedback may play in the decadal-scale anomalies in the Sahel; and the teleconnections between African rainfall anomalies on the decadal-scale and global climate anomalies.
Our understanding of decadal-scale fluctuations could be greatly enhanced by comparative analysis of the 1950s and more recent decades. Unfortunately, many of the necessary data are routinely available only for about the last 20 years. However, a great many data, especially upper-air data, do exist for the 1950s and 1960s, but in obscure archives where they are not readily available. Effort should be devoted to creating usable data sets for earlier periods, especially the 1950s.
Our knowledge of African climate fluctuations on the scale of centuries is considerably more vague. In order to establish these regional fluctuations' causal mechanisms and their relationships to global climate fluctuations, they must be more precisely delineated in time and space. This will require the preparation of better proxy data sets. Few attempts have been made to use, for example, tree-ring or varve chronologies. Also, the innovative paleoclimatic techniques that have been developed recently might be appropriate here as well. Creation of these proxy data sets for Africa should be a high priority.