Each spring many plants put on new leaves and/or open their flowers creating a “green-wave” that can be tracked using phenological data. Various phenological datasets can be used to study spring onset at continental to global scales. Here we present a novel exploratory analysis where we link two multi-decadal and high-spatial resolution datasets: temperature-based phenological indices and land surface phenological metrics derived from satellite images. Our exploratory analysis, illustrated with data for the conterminous US, focuses on identifying regions with similar spring onset, and on mapping the coherence between these phenological products. Our results show that the spring onset patterns captured by the satellite are more complex than the ones identified using temperature-based phenological indices. They also highlight areas with stable and unstable spring onsets (i.e., areas that tend to remain or change of phenoregion from year to year). Finally, our results reveal that temperature-based indices are both positively and negatively correlated with the phenological information that can be derived from satellites. This opens the door to the definition of rules to integrate multi-source phenological data. To cope with the computational challenges of analyzing big geospatial rasters, we executed our analysis on a cloud platform running Apache Spark and various of its extensions (e.g., Geotrellis, SparkMLlib). This platform performed well and allowed the execution of user-tailored analyses. Hence, we believe that our computational platform paves the path towards the efficient analysis of global vegetation phenology at very high spatial resolution and, more generally, to the analysis of the ever-increasing collections of geospatial data about our planet.