MULTIVARIATE ANALYSIS OF METEOROLOGICAL REGIMES AND SOLAR IRRADIANCE VARIABILITY IN DOSSO, SAHELIAN NIGER: IMPLICATIONS FOR PHOTOVOLTAIC APPLICATIONS

Solar irradiance in the Sahel covaries with several meteorological variables, making pooled bivariate correlations vulnerable to seasonal confounding. This study characterizes the joint meteorological structure associated with daytime global solar irradiance in Dosso, Niger, using field observations from May and August 2021. Sixty-two daily records were derived from hourly measurements of irradiance, ambient temperature, relative humidity and wind speed. After quality control, Pearson correlations were evaluated for the pooled dataset and separately by month; standardized variables were then analyzed using principal component analysis (PCA) and k-means clustering. Mean irradiance decreased from 822.75 W m⁻² in May to 651.52 W m⁻² in August, while mean temperature decreased from 38.23 to 29.85°C and relative humidity increased from 35.96 to 74.15%. Across all 62 days, irradiance correlated positively with temperature (r = 0.658) and negatively with relative humidity (r = -0.611), but the within-month relationships differed markedly. The first two principal components explained 91.28% of total standardized variance. PC1 (64.24%) represented a hot-dry-high-irradiance versus cool-humid-low-irradiance gradient, whereas PC2 (27.04%) was dominated by wind speed. A two-cluster k-means solution achieved a silhouette coefficient of 0.469 and separated all May observations from all August observations. The separation describes two sampled seasonal regimes rather than a universal climatic classification. The results demonstrate that season-aware multivariate analysis provides a more defensible interpretation of short solar-resource records than pooled single-variable correlations and can guide the selection of representative meteorological scenarios for photovoltaic studies in data-sparse Sahelian regions.