Modelling climate change in West African Sahel rainfall (1931-90) as an artifact of changing station locations

Since the major droughts in the West African Sahel during the 1970s, it has been widely asserted that mean annual summer rainfall has declined since the late 1960s. Explanation of this persistent regional drying trend was important for famine early-warning and global climate models. However, the network of rainfall stations changed considerably during that recent period of desiccation. Furthermore, it was difficult to reconcile the calculation of a simple mean value for a region known to have a complex spatial and temporal rainfall pattern. A simple model separated the Sahel into 'wet' and 'dry' regions. This model was inverted against mean annual summer rainfall for the Sahel between 1931 and 1990. Model predictions were found to be insensitive to initial starting conditions. The optimized parameters explained 87% of the variation in observed mean annual summer rainfall. The model predicted the mean annual rainfall in the wet 'coastal' and dry 'continental' regions of the Sahel to be 973 mm and 142 mm respectively. Consequently, the predicted long-term mean annual summer rainfall was 558 mm, 15% greater than that of the observed long-term mean (417 mm). The mean annual summer rainfall for the region was corrected by removing the influence of changing station locations over the study period. No persistent decline was found in mean annual summer rainfall, which suggested that the perceived drying trend was an artifact of the crude statistical aggregation of the data and historical changes in the climate station networks. The absence of a decline in rainfall questioned the validity of' the hypotheses and speculations for the causes of the drying trend in the region and its effects on global climate change. It also increased the likelihood that changes over time in other regional and global climate station networks have influenced the performance and interpretation of global climate models. © 2004 Royal Meteorological Society.