Journal Article
Rainfall and dry spell occurrence in Ghana: trends and seasonal predictions with a dynamical and a statistical model
Talardia Gbangou; Fulco Ludwig; Erik van Slobbe; Wouter Greuell; Gordana Kranjac-Berisavljevic
Theoretical and Applied Climatology · Vol. 141, Issue 1-2 · pp. 371-387 · 2020
Abstract
Improved information on the distribution of seasonal rainfall is important for crop production in Ghana. The predictability of key agro-meteorological indices, namely, seasonal rainfall, maximum dry spell length (MDSL) and dry spell frequency (DSF) was investigated across Ghana (with an interest on the coastal savannah agro-ecological zone). These three variables are relevant for local agricultural water management. A dynamical model (i.e. European Centre for Medium-Range Weather Forecasts (ECMWF) System 4 seasonal forecasts) and a statistical model (i.e. response to sea surface temperatures (SSTs)) were used and analysed using correlation and other discrimination skill metrics. ECMWF-System 4 was bias-corrected and verified with 14 local stations’ observations. Results show that differences in variability and skills of the agro-meteorological indices are small between agro-ecological zones as compared to the differences between stations. The dynamic model System 4 explains up to 31% of the variability of the MDSL and seasonal rainfall indices. Coastal savannah exhibits the highest level of discrimination skills. However, these skills are generally higher for the below and above normal MDSL and seasonal rainfall categories at lead time 0. Similarity in skills for the agro-meteorological indices over the same zones and stations is found both for the dynamical and statistical models. Although System 4 performs slightly better than the statistical model, especially, for dry spell length and seasonal rainfall. For dry spell frequency and longer lead time dry spell length, the statistical model tends to perform better. These results suggest that the agro-meteorological indices derived from System 4′ updated versions, corrected with local observations, together with the response to SST information, can potentially support decision-making of local smallholder farmers in Ghana.