Assessment of Prediction Schemes for Estimating Rainfall Onset over Different Climatic Zones in West Africa

Abstract

This study compares the predictive skills of some commonly used prediction schemes to estimate dates of onset of rainfall in various climatic regions of West Africa. Specifically, the abilities of seven different schemes, which are relatively easy to apply on a large scale, were compared from 1980 to 2014 over the three major climatic regions (Forest, Guinea and Sudan-Sahel savannah) of Nigeria. Three of them are dependent on rainfall data (ogive, daily rainfall probability and Walter-Olaniran methods); three on rain-evapotranspiration relation (Benoit, Anyadike, Kowal and Knabe) and one dependent on equivalent potential temperature (Theta-E). The prediction schemes demonstrated that the onset dates were much earlier (from Julian day 061 to 084) in the south than (146 to 162) the north. The results also showed that the onset dates varied slightly (by ±6 days) from east to west within the same climatic zone. Their performances varied across the stations and zones. However, they generally performed adequately well in the Savannah than the Forest. Particularly, Walter-Olaniran and ogive methods performed best in the Sudan-Sahel with predictive skills of less than ±7 days of actual date of onset. The study concludes that foremost criterion for a choice of any prediction scheme is the ability of the scheme for estimating rainfall onset over a region of interest.