Evaluation of WRF-Chem-Simulated Carbon Dioxide Concentrations in the Baltimore–Washington Metropolitan Area

In urban areas, high concentrations of carbon dioxide (CO₂) are emitted and transported, leading to both an increase in global concentrations and concern for the future of Earth’s climate. In recent years, several urban testbeds such as Indianapolis and LA have been established in an attempt to quantify the greenhouse gas (GHG) emissions from metropolitan areas. The NIST Northeast Corridor GHG observation network, which started operation in late 2015, is one of such urban testbeds centered around the Baltimore, MD – Washington, DC metropolitan area (NEC-BW). We present a high-resolution (1km) WRF-Chem CO₂ simulation for the month of February 2016 over the NEC-BW area. We compared the simulated CO₂ to observations from four urban/suburban sites as well as at a rural background site. Four different anthropogenic emissions inventories are used to provide an ensemble of model simulated concentrations: FFDAS, ODIAC, Vulcan, and EDGAR. WRF is also coupled with the VEGAS model to provide the biospheric flux and uses CarbonTracker-NRT for boundary conditions of the background CO₂ concentration. The amplitude of the simulated CO₂ diurnal cycle varied from within 5 ppm to as much as 50+ppm as observed at the four urban/suburban sites on synoptic time scales. Modeled and observed CO₂ concentrations are in good agreement overall to the first order, within approximately 1% of each other when averaged over the entire month. Differences can vary widely however at a given hour, depending on the prevailing synoptic weather condition, time of day, and other meteorological factors such as planetary boundary layer height and winds. However, despite the large differences in total emissions from the four different inventories for the area of interest, the predicted concentration values are dominated by synoptic-scale weather patterns and errors in the simulated meteorology.