Using daily (monthly) rainfall data from 167 (254) stations across West Africa with at least 80% data availability for the 31-year period 1980-2010 and the gridded African Rainfall Climatology Version 2 (ARC2) for the period 1983-2010, linear trends in yearly and monthly rainfall totals were investigated. Measures of the Expert Team on Climate Change Detection and Indices (ETCCDI) and two rainy season onset and retreat definitions were employed to assess the corresponding trends in frequency and intensity of daily rainfall and changes to monsoon season length. A rotated Empirical Orthogonal Function analysis yielded two homogeneous rainfall regions, the Sahel and Guinea Coast, in terms of interannual to decadal rainfall variability, and this led to analysis of station data and Standardised Precipitation Index for the two regions. Results show that the majority of stations in the Sahel between the West Coast and 15 ∘ E shows a statistically significant positive rainfall trend for annual totals. The August-October period exhibits the largest rainfall recovery in the Sahel and the date of the retreat of the rainy season significantly moved later into the year by 2 days decade −1 . The recovery is reflected both in more rainy days associated with longer wet spell duration and more extreme rainfall events. Trends along the Guinea Coast are weak and non-significant except for extreme rainfall related indices. This missing significance is partly related to the hiatus in rainfall increase in the 1990s, but also to the larger interannual rainfall variability. However, the tendency towards a more intense second rainy season suggests a later withdrawal of rains from the West African subcontinent. ARC2 trends are broadly consistent where ground calibration was undertaken, but are dubious for Nigeria and Ghana, and especially for the Guinea, Jos and Cameroon Line highlands due to missing gauge data.

New empirical long-range schemes for the prediction of dates of onset and cessation and of the monthly and annual amounts of rainfall are developed for Kano, in the West African Sahel, using only surface synoptic data. They are based on variations in equivalent potential temperature, q e , which occur as a result of the seasonal, monthly and daily variations of moisture in the summer monsoon flow over West Africa. Agricultural activities may begin about 72 days after the day the anomalies of q e (i.e. q%) first become positive for at least 15 days, essentially signifying the beginning of adequate moisture supply associated with a well established monsoon flow.The new schemes ensure that both the cessation date and the annual amount of rainfall can be predicted prior to the onset of the rains, thus providing, in conjunction with the onset date, very important and useful information for reliable and effective planning of agricultural and water resource activities. Performance tests using an 11-year independent data set indicate that the schemes possess reliable skill.Because the weather over Nigeria is very typical of the entire West African region, being affected by the same wind regime and weather phenomena, these prediction schemes will provide tremendous assistance for enhanced and sustainable agriculture, as well as for efficient water resources management, if extended to the whole area. Furthermore, the methods have the important advantage that, bearing in mind the fact that the majority of West African countries have very sparse, if indeed, any upper-air data, the surface synoptic data needed for their use are readily available in all of the countries.

ABSTRACT:Empirical studies have shown that a necessary condition for abundant rainfall in any year at Kano in Nigeria is a pre-rainy season moisture build-up for 6-8 weeks at the surface or below 800 hPa. The International Centre for Theoretical Physics (ICTP) regional climate model (RegCM3) is used to study the mechanism and variability of the inflow of this moisture over West Africa over a 21-year period . The study focused on three dry (1983, 1989 and 1990) and three wet years (1980, 1988 and 1996) and uses rainfall and other surface data such as specific humidity, over Nigeria for the same years for model validation. Comparison of RegCM3 simulation of rainfall and moisture patterns over West Africa with both CRU and Nigerian stations observations shows that the model performed very well with respect to the observations over the region, particularly over the Sahel and Savana zones.The composites of moisture inflow and wind anomalies for the wet years are found to be associated with areas of boundary layer moisture build-up, overlain by relatively drier (but still moist) air driven by strong west to southwest wind vector anomalies at 700 hPa (African Easterly Jet, AEJ, level). These strong wind anomalies, which occur south of the AEJ between May and September and transport moisture towards it, seem unconnected with the boundary layer winds and are not found in the dry composites. In sharp contrast, the dry composites are characterized by negative moisture anomalies in the boundary layer overlain by drier than normal air associated with northerly to easterly wind anomalies at the AEJ level. Secondly, while the ITD reached the same northernmost limit of about 20°N in August during both dry and wet periods, the AEJ advanced faster across West Africa, reaching a mean position of 16.5°N in August during wet years but 15°N in dry years. Also, on a monthly and inter-annual basis, moisture adequacy and variability analysis show that, for dry years, moisture build-up is delayed but starts early and strongly during wet years. Furthermore, rainfall does not begin properly until about 2.5 months after the start of the moisture build-up. Finally, RegCM3 predicted rainfall correlates very well (r ≥ 0.6) with observed rainfall for stations within Nigeria.

Using thunderstorm and rainfall information from five Nigerian stations for a five year period, the separate contributions of line squalls, thunderstorms and ordinary monsoons to the total rainfall and their variations with latitude are investigated. Total and thunderstorm rainfall decreases with increasing latitude but monsoon precipitation decreases exponentially while line squall rainfall is at a maximum around 9"N. Rainfall from thunderstorms shows a single annual peak in July/August at stations north of about 8"N whereas line squall precipitation exhibits a double maximum for all stations south of 12"N. It is also shown that, in the mean, line squalls are the most important rain producing systems, giving 47.6 per cent of the mean total annual precipitation compared with 39 per cent and 13.4 per cent from thunderstorms and ordinary monsoon rain, respectively.