Abstract:
Rainfall, a critical source of freshwater, is a vital resource on Earth. Improving rainfall forecasting is essential for helping communities worldwide prepare for the impacts of heavy rainfall, ultimately saving lives and reducing infrastructure damage. This study focuses on the development of a forecasting model for heavy rainfall in the Guinea Savanna Zone of Nigeria. Daily secondary data, including wind parameters (direction and speed), relative humidity, rainfall, and thunderstorms (predictors), were collected from January 1, 1981, to December 31, 2015. Data were sourced from the Nigerian Meteorological Agency and the European Centre for Medium–Range Weather Forecasts (ECMWF). Through parametric and non-parametric tests, the characteristics of rainfall in the study area were analyzed. A Multiple Linear Regression (MLR) model was established to assess the relationship between predictors and outcomes. Probability of Detection (POD) was used to evaluate existing numerical rainfall forecasting models against actual heavy rainfall events.
Furthermore, an Artificial Neural Network (ANN) based heavy rainfall forecasting model was developed and applied to predict heavy rainfall (accumulated rainfall of 50 mm and above per day) for the year 2019. The Gumbel extreme probability theory was employed to validate the model. Results indicated an overall upward trend in rainfall and heavy rainfall occurrences, with inter-annual variability. The long-term mean annual rainfall for the study area was calculated at 1220.5 mm, with the highest mean monthly rainfall of 263.5 mm observed in August and the lowest of 1.9 mm in December. The year 1983 recorded the lowest mean rainfall of 1020.3 mm, whereas 2014 marked the highest mean of 1467.2 mm.
Analysis of total regional monthly heavy rainfall revealed November with the lowest total (01) and August with the highest (257). Thunderstorms exhibited a strong correlation with heavy rainfall. Both the ECMWF and UKMet rainfall forecast models underestimated 24-hour rain accumulation. The developed model demonstrated competence in forecasting heavy rain within the study area. Consequently, it is recommended that the Nigerian Meteorological Agency (NiMet) adopt local heavy rainfall forecasts and disseminate results to local communities. Special emphasis should be placed on thunderstorm forecasts. Establishing additional meteorological stations in educational institutions and local government headquarters is advised, along with enhanced focus on short-term rain predictions. It is suggested to conduct similar studies in other ecological zones of Nigeria to develop tailored heavy rainfall forecasting models.
ADVANCEMENT OF RESILIENT RAINFALL PREDICTION MODEL FOR THE GUINEA SAVANNA ZONE IN NIGERIA.  GET MORE, ACTUARIAL SCIENCE PROJECT TOPICS AND MATERIALS
