Prediction uncertainties in the Energy Exascale Earth System land model (ELM) are caused in part by uncertain parameters related to ecosystem processes that control fluxes of carbon and energy. To quantify this uncertainty, 10 model parameters were varied across a 275-member ensemble of global ELM simulations performed at 2x2 degree spatial resolution using satellite phenology. A temporally and spatially resolved surrogate model of gross primary productivity and latent heat flux was then created using a dimension-reduction technique. Global sensitivity analysis performed using the surrogate model indicates different parameters drive model prediction uncertainty depending on time of year, location and environmental conditions. In warmer and drier climates, parameters controlling stomatal conductance and rooting depth distribution are strong drivers of productivity, while in colder climates phenology and temperature sensitivity parameters are more important. Finally, we perform a calibration on the surrogate model using Bayesian methods to demonstrate how ELM parameters and predictions may be improved using gridded observation datasets.