FORECASTING THE OCCURRENCE OF DRY SPELL DURING THE GROWING SEASON IN NIGERIA: A REGRESSION MODEL APPROACH

Abstract

This study examines the relationship between meteorological factors and dry spell occurrences during growing seasons in Nigeria. Using historical data from 2018 to 2023, we analyzed six geopolitical regions, each represented by one state. We collected and analyzed historical meteorological data (rainfall, temperature, and humidity) from the six states. Correlation analysis identified significant relationships between variables. We derived a Temperature Range variable from maximum and minimum temperatures and employed Ordinary Least Squares (OLS) regression and machine learning regression models to predict rainfall patterns. Significant correlations exist between rainfall, temperature, and humidity. Temperature Range and Relative Humidity effectively predict future rainfall patterns. Regional temperature variations were observed, with high temperatures prevailing in northern regions and low temperatures
characterizing southern regions. A strong negative correlation exists between Temperature Range and annual rainfall. Furthermore, regions with lower temperature ranges exhibit higher humidity, leading to increased rainfall and reduced dry spells. Notably, machine learning regression models
better OLS regression due to relaxed normality assumptions. This study enhances understanding of dry spell predictions in Nigeria's growing seasons, providing valuable insights for agricultural planning and climate resilience strategies.