Assessment of ground-based atmospheric observations for verification of greenhouse gas emissions from an urban region

Abstract

International agreements to limit greenhouse gas emissions require verification to ensure that they are effective and fair. Verification based on direct observation of atmospheric greenhouse gas concentrations will be necessary to demonstrate that estimated emission reductions have been actualized in the atmosphere. Here we assess the capability of ground-based observations and a high-resolution (1.3 km) mesoscale atmospheric transport model to determine a change in greenhouse gas emissions over time from a metropolitan region. We test the method with observations from a network of CO(2) surface monitors in Salt Lake City. Many features of the CO(2) data were simulated with excellent fidelity, although data-model mismatches occurred on hourly timescales due to inadequate simulation of shallow circulations and the precise timing of boundary-layer stratification and destratification. Using two optimization procedures, monthly regional fluxes were constrained to sufficient precision to detect an increase or decrease in emissions of approximately 15% at the 95% confidence level. We argue that integrated column measurements of the urban dome of CO(2) from the ground and/or space are less sensitive than surface point measurements to the redistribution of emitted CO(2) by small-scale processes and thus may allow for more precise trend detection of emissions from urban regions.

Fig. 1.

Average hourly observed CO₂ concentration at the downtown site in 2006. (A) Averaged by hour of the day from winter (Dec.–Feb.) and summer (June–Aug.) months, and from nearly the entire year (April–Dec.) at the background site. (B) Observed CO₂ from the whole year (Left y axis) versus average hourly CO₂ emissions estimated from the Vulcan inventory for a 0.5° × 0.5° area encompassing the Salt Lake Valley (Right y axis). (C and D) Distribution of observed CO₂.

Fig. 2.

Hourly observed and modeled CO₂ concentrations for two weeks in October 2006 (A) at the downtown site, and (B) averaged by hour of the day at the downtown, neighborhood, and junior high sites.

Fig. 3.

Hourly modeled versus observed CO₂ at three sites for a two-week time period in October 2006 resulting from (A) high-resolution and (B) baseline model configurations. Solid lines are standard major axis regression lines and dashed lines are one-to-one shown for reference.

Fig. 4.

Quantile–quantile plots of hourly modeled versus observed CO₂ at the downtown site for two-weeks in October 2006 from (A) baseline and (B) high-resolution models. Model values unscaled and scaled by the two optimization methods are shown. For the high-resolution model, scaling factors by the two optimization methods are near identical, so the two scaled model distributions are nearly indistinguishable.

Fig. 5.

Simulated partial column-averaged XCO₂ (ppm) enhancements above background up to 3 km above SLC and the surrounding area on October 18, 2006 at 15 h (A) and 23 h (B) MST. (D and E) Simulated CO₂ enhancement near the surface, 50 m above the ground, for the same times and locations as in A and B. (F and G) Vertical slices through the areas of maximum XCO₂ enhancement in the urban domes. (C) Topography in kilometers above sea level. The downtown and rural measurement sites are marked with Xs for reference. Lines in A–E show the positions of the transects plotted in F and G. Note that the two upper-left panels have the same scale, but the four lower panels do not.