Forecasting ecosystem response to climate change is critical for guiding policy-making but challenging due to: complicated relationships between microclimates and regional climates; species’ responses that are driven by extremes rather than averages; the multifaceted nature of species’ interactions; and the lack of historical analogs to future climates. Given these challenges, even model systems such as the Galapagos Islands, a world-famous biodiversity hotspot and World Heritage Site, lack clear forecasts for future environmental change. Here, we developed a novel non-parametric method for simulating the ecosystem futures based on observed vegetation productivity (NDVI) during analogous weather observed historically. Using satellite images taken from the past to piece together a simulated future, we projected that productivity of terrestrial vegetation of the Galapagos will increase over the next century by approximately one standard deviation archipelago-wide, with increases largest during the wet season (January to June), and in the arid zones. This greening may impact a variety of ecological and evolutionary processes, species of conservation concern, and agricultural practices. Our straightforward approach can be applied to many other regions, particularly those with rapid ecosystem responses to stochastic inter-annual climatic fluctuations that provide appropriate climate analogs for forecasting.