Continental‐Scale Validation of VMF3‐ and ERA5‐Derived Precipitable Water Vapour Against Ground‐Based GNSS Observations Across Africa

Accurate precipitable water vapour (PWV) estimates are essential for weather forecasting, climate monitoring, and hydrological applications, particularly across observation‐sparse regions. This study presents a continental‐scale validation of PWV derived from the Vienna Mapping Function 3 (VMF3) tropospheric products and the fifth‐generation ECMWF reanalysis (ERA5) against ground‐based Global Navigation Satellite System (GNSS) observations over Africa. Daily PWV estimates from 27 International GNSS Service (IGS) stations spanning January 2015–August 2022 were analysed. IGS was adopted as the primary observational reference after demonstrating excellent consistency with the independently processed Nevada Geodetic Laboratory (NGL) product (bias = −0.27 mm; root mean square error (RMSE) = 1.32 mm; Pearson correlation coefficient ( r ) = 0.9965; concordance correlation coefficient (CCC) = 0.9963). Performance was evaluated using bias, RMSE, relative RMSE, r , CCC, and Bland–Altman analysis at station, regional, and continental scales, complemented by seasonal, moisture‐dependent, and geographic stratifications, an analytical retrieval‐uncertainty budget, and homogeneity testing. Continentally, both products reproduced GNSS PWV with high fidelity, with ERA5 showing a marginal overall advantage (bias = −0.19 mm; RMSE = 1.64 mm; rRMSE = 6.4%; r  = 0.9945; CCC = 0.9944) over VMF3 (−0.34 mm; 1.71 mm; 6.6%; 0.9942; 0.9939), consistent with the NGL cross‐check. Performance was geographically heterogeneous, however, indicating no uniform regional superiority. Absolute errors generally increased with atmospheric moisture load, whereas relative errors increased significantly with elevation and absolute latitude (Spearman’s ρ up to + 0.77; p   < 0.001). Deseasonalised continental PWV exhibited a significant increasing tendency over the study period (Sen’s slope 0.28–0.34 mm yr ⁻¹ ; Mann–Kendall p   < 0.003) and a common ~2 mm upward mean shift around July 2019 (Pettitt and Standard Normal Homogeneity Test (SNHT), p   < 0.001), although the short record precludes attribution to long‐term climate change. These results establish VMF3 and ERA5 as reliable and temporally continuous complements to GNSS PWV over Africa, supporting improved weather and climate applications in data‐sparse regions.