The continuous shadows of passing clouds create rapid (seconds to minutes) variations in the solar radiation reaching the land. These cloud flecks induce fast fluctuations in radiative forcing that trigger a cascade of interacting processes across a wide range of spatial and temporal scales acting in soil, vegetation and atmosphere. Understanding and quantifying these processes require detailed knowledge of soil physics, plant biology, atmospheric composition, radiation, turbulence, boundary-layer and cloud dynamics. The challenges are twofold: first, there is a knowledge gap since little is known on how these processes interact across scales and different domains. Second, our weather and climate models are progressively able to resolve smaller and shorter spatial and temporal scales. Thus, they enter the domain where the fast cloud induced interactions act on the surface energy, water and carbon dioxide fluxes which, in turn, feedbacks on cloud formation. Consequently, appropriate parameterizations of these processes need to be designed and tested. In CloudRoots, we aim to develop a unique combined observation - simulation system that allows to quantify all necessary interactions between soil, vegetation and atmosphere, from the size of the stomata (10 - 100μm) to the size of the atmospheric boundary layer (~1 km), while accounting for second to minute scale dynamics as well as diurnal and seasonal cycles. CloudRoots innovates by integrating recent advances in the atmospheric science field. Observationally, we plan new measurement of one-minute fluxes of the water and carbon stable isotopes by combining laser spectroscopy with the scintillometer technique. Numerically, we plan high-resolution experiments using large-eddy simulations including three-dimensional radiation perturbations by clouds and canopy. These novel elements are methodologically divided in two interlinked sub-projects; one is focused on processes and the other on scaling. The processes project (PhD1) is oriented to soil-vegetation-atmosphere interactions in which there will be a prominent role of surface observations. By building on these observations, the scaling project (PhD2) derives relationships of the land processes to larger spatiotemporal scales using a hierarchy of models. The scaling laws provide a basis for parameterizations that will be systematically evaluated in grassland and forest ecosystems of the Ruisdael Observatory in the Netherlands. Once established and evaluated, our scaling relations will be studied in a boreal and a tropical forest sites. CloudRoots will lay the observational and theoretical foundation necessary to include spatiotemporal scale processes in weather and climate forecasts and improve predictions of extreme events.