Trends in anthropogenic NOx emissions

High-resolution, routine observations from OMI offer a unique perspective on the spatial and temporal variation of regional-scale anthropogenic NO2 emissions. By comparing trends in OMI observations with those from ground-based measurements and an emissions inventory, we have shown that satellite observations are well-suited for capturing changes in emissions over time.


Average tropospheric NO2 column concentrations (molecules cm-2) from OMI and CARB surface sites for weekdays and weekends for the South Coast region (Los Angeles basin) of California (Russell et al., 2010).


OMI observations indicate that NOx concentrations have decreased significantly in urban regions of the United States between 2005 and 2011, with an average reduction of 32 ± 7%. By examining day-of-week and interannual trends, we find that these reductions can largely be attributed to improved emission control technology in the mobile source fleet; however, we also find that the economic downturn of the late 2000's has impacted emissions.


Average summertime (April–September) OMI BEHR NO2 column densities for 2005 (left) and 2011 (right) (Russell et al., 2012).


NOx Emissions From Fires

Emissions from biomass burning make up roughly 15% of the global NOx budget with substantial variability in annual emissions arising from variability in fire frequency and intensity. Constraining biomass burning emissions has proven difficult due to the large uncertainties in NOx emission factors and global biomass burned. Laboratory studies indicate that biomass burning NOx emission factors can vary greatly depending on individual fire conditions, including fuel type and meteorological conditions. We find that satellites are capable of observing NOx emissions from fires and that observation of fire activity and NO2 can establish statistical properties of NO2 emission coefficients, which are proportional to emission factors. Using observations of fire radiative power (FRP) from MODIS and tropospheric NO­2 column measurements from OMI, we derived NO2 wildfire emission coefficients (g MJ−1) for three land types over California and Nevada from observations of 1,960 fires over the years 2005–2008. Retrieved emission coefficients are 0.279 ± 0.077, 0.342 ± 0.053, and 0.696 ± 0.088 g MJ−1 NO2 for forest, grass and shrub fuels, respectively. These emission coefficients reproduce ratios of emissions with fuel type reported previously using independent methods; however, the magnitudes of these coefficients are lower than prior estimates, which suggests either a negative bias in the OMI NO2 retrieval over regions of active emissions, or that the average fire observed in our study has a smaller ratio of flaming to smoldering combustion than measurements used in prior estimates of emissions. Our results indicate that satellite data can provide an extensive characterization of the variability of fire NOx emissions; 67% of the variability in emissions over this region can be accounted for by using an FRP-based parameterization.

MODIS fire pixels, OMI tropospheric NO2 columns, and wind vectors during the Zaca fire on August 19, 2007. NO2 in the fire plume dwarfs the signal from Los Angeles (white circle) (Mebust et al., 2011).

Agricultural NOx emissions

Soil NOx emissions (SNOx) are highly variable products of microbial activity in soils. Processes governing SNOx are not well understood, but the best correlation on a regional scale is with water-filled pore space, temperature, and N-availability. We use OMI NO2 columns interpreted with a model of SNOx, driven by daily soil temperature and precipitation from the North American Regional Reanalysis (NARR) for 2005–2008 to examine the interannual variability of SNOx over the United States, their magnitude and pulsing behavior, and implications for ozone air quality.   Anomalously large SNOx (~50% greater than the June 2005–2008 mean) are predicted in June 2006 over the agricultural Great Plains due to rain-induced pulsing, triggered by warmer (+ 0–2°C) and drier (+ 0–50%) than average conditions over the region. Mean summertime tropospheric NO2 columns over the agricultural Great Plains ranged from 1.25 – 3×1015 molecules cm-2. In June 2006, columns were 30% higher with a spatial pattern consistent with our predicted SNOx anomaly confirming the presence of large interannual variation in SNOx (Figure 6). In a case-study over agricultural southeastern South Dakota, we examined rain-induced pulsing events in May–July 2006. OMI tropospheric columns reached 4.6 × 1015 molecules cm-2, equivalent to a surface concentration of ~2 ppbv. The modeled peaks occur later than in observations and persist for longer, differences that suggest the model is not capturing the dynamics of pulsed SNOx properly. We use the GEOS-Chem CTM driven by our NARR SNOx model to examine implications for ozone air quality. We find that in June 2006, SNOx enhanced mean 8-hr maximum surface ozone by 5 ppbv, compared with 3 ppbv for 2005–2008, with daily ozone enhancements due to SNOx reaching up to 16 ppbv. These large enhancements suggest that reducing fertilizer use or increasing its efficiency would substantially improve air quality in the central United States.

June mean anomaly in NO2 from a model of soil NOx emissions (left) and from OMI (right), calculated as difference compared to the June 2005–2008 mean (Hudman et al., 2010).

The Spatial Resolution of Satellite-based NO2 Observations

Capturing the spatial variability of NOx is necessary for understanding both ozone and nitric acid formation and the transport of reactive nitrogen due to nonlinear feedbacks of NOx on hydroxyl radical concentration and ozone production. For most of its time in orbit, the operational mode of OMI has been used to infer column NO2 with a footprint at nadir that is 13 km along-track and 24 km across-track. Higher-resolution observations can be achieved if CCD detector elements in the across-track dimension are not binned on-board the instrument, giving an optics limited footprint of approximately 7×13 km2 at nadir. We have found that at this spatial scale, slant column NO2 varies by up to 1×1016 molecules cm-2. The retrieved signal to noise ratio of the spectrum is as high as 20 and is ~5 at the plume edge, similar to the ratio for binned values in the same region. The high-resolution observations are capable of distinguishing three of four large point sources in close proximity around the Rihand Reservoir, a distinction not possible in either the single orbit or six-nadir orbit operational scale average. Taking advantage of the enhanced spatial detail, we derive a chemical lifetime of 1.9 hours for an NO2 plume advected over an uninhabited desert downwind of Dubai, UAE, where the spatial gradient in column NO2 depends more on chemical processes than on emissions.

 a) MODIS RGB image, b) OMI super-zoom mode slant column NO2 (SCDNO2), c) operational-scale SCDNO2,, and d) a six-orbit operational-scale average SCDNO2 over the Rihand Resovoir in India. Power plants located around the Rihand Reservoir include Singrauli and Vindhyachal (PP1 – 4200 MW), Anpara (PP2 –1600 MW), Rihand (PP3 –1000 MW pre-2006), and Obra (PP4 – 1600 MW; Valin et al., 2011).



Using Satellite Observations to Investigate VOC and OH Concentrations

Day of week patterns in NO2 have been reported around the world and used to characterize daily patterns in emissions. However, changes in NO2 with day of week also reflect changes in the chemical removal rate of NO2. We used the WRF-Chem model to simulate the response of column NO2 to decreases in weekend NOx emissions and compare the simulations with spatial variations in the day  of week pattern of NO2 as observed by OMI in the Los Angeles Basin. We find that in the model, the absolute reduction in NO2 and its spatial variation depend on emission reductions and changes in its chemical removal via feedback on OH and RO2. While the standard WRF-Chem model predicts weekday column NO2 that is in qualitatively good agreement with observation, it is not able to accurately capture the magnitude or spatial pattern of weekend decreases in column NO2. By increasing emission of volatile organic compounds in the 2005 EPA National Emission Inventory (NEI)  by a factor of two, the simulated response of column NO2 to decreased NOx emissions is in much better agreement with observations than the standard WRF-Chem scenario. Taken together, these results indicate that the observations and model provide constraints not only on NO2 emissions, but also on VOC and HOx production rates.


OMI-observed average JJA 2005–2008 weekday column NO2 (top left) and the percent decrease of column NO2 observed by OMI in the corresponding weekend average (bottom right). WRF-Chem simulation with standard (top middle, EVOC=1x NEI2005) and 2x standard VOC emissions (top right, EVOC=2x), over Southern  California for June 10–24, 2006 of “weekday” column NO2 and the simulated decrease in simulated “weekend” column NO2. Anthropogenic mobile NOx emissions are decreased by 37.5% for weekend simulations while emissions of all other species are held constant. Values where decreases are less than 37.5% are not shown.