Evaluation of Sunshine Correlation with Rainfall, Rain Rate and Relative Humidity for Solar Photovoltaic Generation Forecasting and Grid Integration in Sokoto State, Nigeria

Abstract

This study quantifies the statistical relationships among four meteorological variables: sunshine duration, rainfall amount, rainfall rate, and relative humidity using six years of monthly observational data (2015–2020, n = 72) sourced from the Nigerian Meteorological Agency (NiMet). The statistical analyses included a one-way analysis of variance (ANOVA) F-test, Pearson Correlation Coefficient (PCC) analysis, and linear regression trend modelling. The results show that the annual mean sunshine duration is 10.235 hr/day (3,740 hr/yr), corresponding to an estimated Global Horizontal Irradiance (GHI) of 2,431 kWh/m2/yr, indicating that Sokoto State has a favourable solar resource. One-way ANOVA confirms that the monthly means of all four variables differ significantly, indicating a strong seasonal structure in the dataset. All annual ANOVA results are non-significant (p > 0.05). PCC analysis reveals significant negative correlations between sunshine duration and rainfall (r = − 0.7179, p < 0.001), and between sunshine duration and relative humidity (r = − 0.6342, p < 0.001), confirming that moisture parameters serve as reliable inverse proxies of solar resource availability in Sokoto’s Sahel climate. Rainfall and rain rate are nearly collinear (r = 0.9515, p < 0.001), while rainfall and relative humidity share a strong positive association (r = 0.8209, p < 0.001), together forming a physically coherent wet-season moisture cluster. The annual mean rainfall is low, at 24.45 mm/yr, concentrated almost entirely in the June–September window. Year-on-year trends indicate a downward tendency for sunshine (Overall Trend) and rainfall, while rain rate and relative humidity exhibit slight upward tendencies, although none of the annual ANOVA results are statistically significant. A model excluding sunshine was developed; it explains 52% of the monthly variance and achieves a MAPE of 5.1% on the data. \(P_{\text {forecast }}=8616.24-137.157 R-3.758 R H\left(\frac{k W h}{\text { day }}\right.\) per installed \(\left.M W p\right)\).