Appraisal of global rainfall forecasting models on heavy rainfall days over the Guinea Savanna Zone, Nigeria

Abstract

Numerical models are vital tools in forecasting rainfall globally. In most cases, these models underperform. This has formed the basis for this research which was aimed at the appraisal of global rainfall forecasting numerical models on heavy rainfall days over the Guinea Savanna Zone, Nigeria (GSZN) which served as the study area. Nine (9) meteorological stations were chosen from the study area for the purpose of data collection due to their long history of data. These stations included Makurdi, Lokoja, Ibi, Ilorin, Lafia, Abuja, Minna, Jos and Kaduna. Secondary data were used for the study. These included the observed daily rainfall data obtained from the Nigerian Meteorological Agency (Nimet), Oshodi, Lagos and rainfall forecasts obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) as well as the United Kingdom Met Office (UKMet). Data were presented in figures as well as tables and analysed using the probability of detection (POD), root mean square error (RMSE), mean absolute error (MAE) and rainfall intensities. The results indicated that both global numerical models appraised performed well in categorical rain forecasting over the study area, but under-estimated the total rainfall on the nine (9) events appraised. It was therefore suggested among others that both models be dynamically downscaled to take into consideration local peculiarities over Nigerian domain and local numerical model such as the artificial neural network (ANN) be developed to forecast various degrees of rainfall intensities.