Accurate estimates of fossil fuel carbon dioxide (FFCO2) emissions are a critical component of local, regional, and global climate agreements. Current global inventories of FFCO2 emissions do not directly quantify emissions at local scales, instead spatial proxies like population density, nighttime lights, and powerplant databases are used to downscale emissions from national totals. We have developed a high-resolution (hourly, 1km2) bottom-up Anthropogenic Carbon Emissions System (ACES) for FFCO2, based on local activity data for the year 2011 across the Northeastern U.S. We compare ACES with three widely used global inventories, finding significant differences at regional (20%) and city scales (50-250%). At a spatial resolution of 0.1°, inventories differ by over 100% for half of the grid cells in the domain, with the largest differences in urban areas and oil and gas production regions. Given recent US federal policy pull-backs regarding greenhouse gas emissions reductions, inventories like ACES are crucial for US actions, as the impetus for climate leadership has shifted to city and state governments. The development of a robust carbon monitoring system to track carbon fluxes is critical for emissions benchmarking and verification. We show that existing downscaled inventories are not suitable for urban emissions monitoring, as they do not consider important local activity patterns. The ACES methodology is designed for easy updating, making it suitable for emissions monitoring under most city, regional, and state greenhouse gas mitigation initiatives, in particular for the small and medium-sized cities that lack the resources to regularly perform their own bottom-up emissions inventories.