Empirical analysis of dry spells for agricultural applications in West Africa

Abstract

Recurring droughts and decreased agricultural productivity during the last two decades in West Africa point to the need for a clearer understanding of the length of dry spells, their frequencies and their probabilities. The simplest calculations of dry spells for general applications involve computation of the probabilities of maximum and conditional dry spells exceeding a user-specified threshold value from a given calendar date. For more precise applications in agriculture, it is important to consider the different periods after sowing a crop, since sowing dates in the semiarid West African regions vary from year to year. Using the specific definition of onset of rains in each year as the sowing date, the length of dry cells was calculated from the historical rainfall data. Probability distribution of time to the next wet day and the percentage frequencies of dry spells were computed for successive days after sowing (DAS) a crop. Dry-spell analysis showed a pronounced drop in the drought risk for cereal crops from the panicle initiation phase (20 DAS) to the following phase (60 DAS). The relationships between mean annual rainfall and average frequency of dry spells for the selected locations in West Africa showed distinct patterns and permit the prediction of the frequency of dry spells from annual rainfall totals. Applications of the dry-spell analysis for the choice of a crop/variety, supplements irrigation, and crop water requirements have been described with examples.