Solar forecasts are essential for grid-connected solar photovoltaics (PV) as penetration increases. This increase leads to increased grid variability and uncertainty that must be managed by power system operators and/or PV plant owners. Better solar forecasting tools contribute to facilitating this management. This work examines two spatio-temporal approaches for short-term forecasting of global horizontal irradiance using gridded satellite-derived irradiance as experimental support. The first approach is a spatio-temporal vector autoregressive (STVAR) model combined with a statistical process for op-timum selection of input variables. The second is an existing operational cloud motion vector (CMV) model, a deterministic approach. An evaluation of the predictive performance of these models is investigated for a case study area in the Caribbean Islands. This region is characterized by a large diversity of microclimates and land/sea contrasts, creating a challenging solar forecasting context. Using scaled persistence as a reference, we benchmark the performance of the two spatio-temporal models over an extended 220×220 km domain, and for three specific, climatically distinct locations within this domain. We also assess the influence of intra-day solar resource variability on model performance. Exploiting this observation could lead to better forecast performance by harnessing the strengths and minimizing the weaknesses of both models for different conditions/locations. In a subsequent investigation, a blended model CMV/STVAR will be developed, by combining the strengths of a purely physical approach and those of a purely statistical approach. Operationally, such an approach would mesh with operational industry-targeted forecast services that exploit gridded satellite remote sensing resources.