Anthropogenic carbon dioxide (CO2) emissions dominate the atmospheric greenhouse gas radiative forcing budget. However, these emissions are poorly constrained at the regional and seasonal scales. Here, we use a combination of tall tower CO2 mixing ratio and carbon isotope ratio observations and inverse modeling techniques to constrain anthropogenic CO2 emissions within a highly heterogeneous agricultural landscape near Saint Paul, Minnesota, in the Upper Midwestern United States. The analyses indicate that anthropogenic emissions contributed 6.6, 6.8 and 7.4 mmol mol-1 annual CO2 enhancements (i.e. departures from the background values) in 2008, 2009, and 2010, respectively. Oil production, the energy industry, and residential emissions contributed 2.9 (42.5%), 1.4 (19.8%), and 1.1 mmol mol-1 (15.8%) of the total anthropogenic enhancement over the three-year period. The total anthropogenic signal was further partitioned into CO2 emissions from fuel oil, natural gas, coal, gasoline, and diesel consumption using inverse modeling and carbon isotope ratio analyses. The results indicate that fuel oil and natural gas consumption accounted for 83% of the anthropogenic CO2 sources. Here, the a posteriori CO2 emission from natural gas was 60.4 ± 24.0% (a priori 20.0%), while the a posteriori CO2 emission from fuel oil was 25.2 ± 3.8% (a priori 50.8%) during winter— suggesting a more important role of residential heating. The modeled carbon isotope ratio of the CO2 source (-30.6 ± 0.4‰) was relatively more enriched in 13C-CO2 compared to that derived from Keeling plot analyses (-37.5‰ to -35.0‰), supporting that natural gas consumption was underestimated for this region.