Convective clouds are central to Earth’s water and energy balances, and are a major source of uncertainty in climate projections on decadal to centennial timescales. The challenge for Earth system models is not only to represent small-scale processes in convective clouds, but also their interactions with other changing components of the Earth system. In this seminar I will present observational and modeling evidence of interactions between convective clouds, soil moisture, and vegetation, and their impacts on agricultural drought and heat-extremes. I will discuss single-column and global climate model experiments that explore multiple pathways by which soil moisture and vegetation can affect climate. The results point to sensitivities of simulated clouds and climate to model representation of surface water and energy fluxes that (in the presence of feedbacks) can yield large biases in climate prediction. The complexity of these feedbacks motivates new approaches that can integrate observations and complex models across spatial and temporal scales, to better inform land-surface and cloud parameterizations in Earth system models. Toward this goal, I will discuss ongoing and planned efforts to combine ground-based and satellite observations with high-resolution (cloud-permitting and LES) model experiments.