The recognized importance of soil moisture (SM) in land-atmosphere interactions has driven efforts to develop higher resolution datasets that better capture such interactions. In this study, ESA Climate Change Initiative MEDIUM RESOLUTION SM (0.1°) and the retrievals from contributing sensors are evaluated over Africa. Daytime land surface temperature (LST) from thermal infrared imagery can identify spatial variability in SM and is used here for evaluation of 0.1° (and reference coarse 0.25°) SM products. Among the different passive (AMSR2, SMAP and SMOS), active (ASCAT), MEDIUM RESOLUTION (merged), and reanalysis (ERA5) SM products evaluated, ASCAT was best at capturing spatial variability at 0.1° (and for most surface conditions at 0.25°). The MEDIUM RESOLUTION 0.1° dataset is dominated by passive sensors and shows poorer spatial correlation than ASCAT SM. ERA5-Land SM showed almost no skill in capturing mesoscale spatial patterns, and ERA5 (which benefits from ASCAT SM data assimilation) also performed comparatively poorly. We discuss two applications of medium resolution SM using the best performing SM dataset. We use 0.1° ASCAT SM to rank different precipitation datasets in their ability to capture medium resolution accumulated rainfall patterns (~12 hrs). Rain-over-Africa (a deep learning-based dataset) that solely relies on thermal infrared radiation clearly outperforms more established products. ERA5 precipitation shows poor skill, which largely explains the poor ERA5 SM estimates. A second application explores the preliminary use of SM in nowcasting convective storms (cores) in western Africa. We find that adding 0.1° ASCAT SM to the baseline nowcasting model (which uses thermal infrared brightness temperatures only) improved the prediction accuracy of the nowcasts at lead time of 4 hrs. Results show that adding SM resulted in an increase in predicted probability values for true events, particularly across regions of dry SM anomalies and high spatial gradients in SM climatology. These results highlight the potential of better prediction of rainfall through incorporation of medium resolution SM, while further work is needed to leverage the advantages of multiple sensors in creating a 0.1° SM dataset for applications that require high temporal resolution information.