This proposal describes a project to investigate the role of mesoscale land-atmosphere interactions - both their control by large-scale meteorological conditions and their aggregate impact at these large scales - in the coupled land-atmosphere system. The goal is to determine the impact of these scale interactions on warm season convective precipitation in the continental U.S., with a focus on links between soil moisture and rainfall. The underlying hypothesis is that nonlinear interactions between scales may be a source of important feedbacks that might help control the joint evolution of the land and atmosphere. The strategy will be to use a state-of-the-art, high-resolution coupled land-atmosphere modeling system to investigate the following primary research questions: How do precipitation systems driven by large-scale dynamics scale down to mesoscale heterogeneity in soil moisture and surface fluxes, and what is the evolution of this heterogeneity over monthly to seasonal timescales? How does this evolving mesoscale surface heterogeneity scale up to influence convective clouds and precipitation at larger scales, and how important is the aggregate impact of these mesoscale effects over a larger-scale region? Can we identify feedbacks between the large-scale and mesoscale processes? How do these downscaling and upscaling processes and feedbacks vary intraseasonally, as a function of synoptic dynamical regime? How do these downscaling and upscaling processes and feedbacks vary interannually, as a function of variations in hydrological regime (e.g., dry vs. wet years)? The results of this investigation are expected to contribute to a deeper and more unified picture of land-atmosphere coupling across spatial scales from mesoscale to synoptic, and across timescales from hourly to seasonal. Recent research suggests that land-atmosphere interactions and feedbacks, e.g., links between soil moisture and precipitation, are an important control on continental-scale hydrometeorlogy, and furthermore that the spatial heterogeneity of surface characteristics, not just their mean values, may also be important. Current large-scale models (e.g., GCMs) may not adequately capture all the relevant interactions, thus hampering their predictive ability. A possible factor is that processes associated with subgrid-scale surface heterogeneity in such models may be important. However, the extent to which small-scale surface heterogeneity, e.g., on the order of 10 km, plays a role at larger spatial scales is unknown, and is currently a topic of debate in the meteorological community. The work proposed here is intended to provide additional insight into this issue. Because the proposed model simulations incorporate both high resolution and seasonal run times, it is hoped that they will be able to bridge some of the gaps in space and time that are forced by the tradeoffs, arising from computing limitations, between domain size, resolution, and simulation length, that have so far hindered investigation of these questions. Improving our understanding of the factors that control continental rainfall, both its mean value and spatial and temporal variability, particularly at long lead times such as monthly to seasonal, is crucial for applications such as managing water resources and planning for weather-related emergencies. Furthermore, estimating changes in regional climate and hydrology resulting from the combined influence of changing atmospheric composition and natural or anthropogenic land cover change will require improved representations of land-atmosphere interactions in models. The improved understanding that it is hoped will emerge from the proposed work is expected to have broader implications in these areas. Finally, the work proposed here is expected to contribute to graduate study and graduate research opportunities, through the PI's affiliation with the Department of Environmental Sciences at Rutgers University, and to outreach to the general public, through the PI's affiliation with the Center for Environmental Prediction, also at Rutgers University.