Understanding the causes of and predicting future trends in global climate change are major challenges that are of great societal relevance and importance in the 21st century. To date, the most reliably predicted indications of climate change are globally averaged properties of the climate system, such as the global-mean surface temperature. Averaged global temperatures, however, fail to address society-relevant questions, which are often local and regional. Current coupled global climate models (CGCMs) used to simulate climate change typically have rather coarse horizontal spatial resolution (about 100 km). To generate climate information at finer spatial resolution in specific regions of interest (‘downscaling’), regional climate models (RCMs) are often used. At present, most downscaling is only performed when modeling atmospheric and land conditions, with the oceans represented by sea surface temperature boundary conditions derived from the CGCMs. This means that RCMs are strongly affected by the biases present in the CGCM simulations of sea surface temperatures and atmospheric flow patterns. These biases arise because the coarse resolution of CGCMs is insufficient to resolve fine-scale physical processes, such as atmospheric convection, oceanic eddies, and steep orography. In this project, we propose to improve simulations of regional climate change through a hierarchical approach using coupled regional atmospheric and oceanic models, in conjunction with CGCMs. By using a regional ocean model, we will simulate ocean circulation at far greater spatial resolutions, and by coupling it to a regional atmospheric model, we will be able to resolve ocean-atmosphere interactions on subseasonal time scales in the mesoscale regime. As a result, we hope to alleviate many of the systematic biases seen in CGCMs and improve the downscaling process. The Atlantic sector is chosen as an ideal test-bed for this coupled RCM approach because it is characterized by large systematic errors in CGCM simulations, and also because the thermohaline circulation in the Atlantic Ocean is particularly sensitive to climate change. Our proposed research will leverage existing community model development activities, including the global Community Climate System Model (CCSM) and the regional Advanced-Research version of the Weather Research and Forecast (WRF-ARW) model, both developed at the National Center for Atmospheric Research. Additionally, we will be using the Regional Ocean Modeling System (ROMS), developed at Rutgers University and UCLA. The use of these community models will greatly facilitate sharing our results with the broader climate modeling community. We also expect our work to feed back into the CCSM development process. The objective of this project is to advance the DOE Biological and Environmental Research program’s goal to accurately predict future climate on decadal to centennial time scales. Improved mathematical representations of atmospheric and oceanic dynamics and transport in the coupled RCM approach will increase the accuracy of long-term climate change projections, especially at sub-continental scales.