1. Atmospheric Measurement of Regional Methane Emissions Kenneth J. Davis 1, Thomas Lauvaux 1 and Colm Sweeney 2 1 The Pennsylvania State University, 2 NOAA ESRL/U. Colorado Current Developments and Impacts of Natural Gas in Transportation Transportation Research Board 93 rd Annual Meeting Washington, D.C., 12 January, 2014 P14-7201

2. Outline Global context Need for “regional” emissions measurements Overview of atmospheric methods –Local, regional, global Regional emissions estimates –Aircraft based –Tower based –“top-down” vs. “bottom-up” studies –Role of satellite sensors Conclusions

3. Global context Human activity is driving the greenhouse gas (GHG) content of the atmosphere far beyond anything seen for at least 600,000 years.

4. IPCC, Third Assessment Report, 2001 Last 2000 years Ambient and ice core observations: Ultimate “top down” GHG assessment

5. Global context Human activity is driving the greenhouse gas (GHG) content of the atmosphere far beyond anything seen for at least 600,000 years. Large reductions in GHG emissions will be needed to stabilize climate, even at 2K global warming (current <1K). –What level of mobile source GHG emissions is tolerable?

6. B1 scenario limits global warming to ~2K by 2100 IPCC Assessment Report 4, 2007

7. B1 scenario requires large emissions reductions – peak only 25% above 2000 levels

8. Need for “regional” measurements of GHG emissions Emissions mitigation (e.g. reduction of emissions by adoption of new natural gas – based technologies) will happen at “regional” scales. Validation of emissions mitigation (e.g. current debate over methane leakage rates) will require independent measurements Atmospheric GHG measurements have the potential to provide such independent emissions estimates. “Regional” = counties to continents

9. Atmospheric methods: A very brief summary

10. Summary of Atmospheric Emissions Measurement Methods: Gaps between Chamber/Turbulent Flux and Inversion Methods 6 GAP Chamber flux Eddy covariance or Plume dispersion Airborne flux Global year month hour day Time Scale Spatial Scale (1m) 2 = 10 -4 ha (1000km) 2 = 10 8 ha (100km) 2 = 10 6 ha (10km) 2 = 10 4 ha (1km) 2 = 10 2 ha R earth Regional Atmospheric Inversions Bridging the gap between atmospheric inversions and turbulent flux measurements is my research expertise.

11. “Handshakes” needed Combine regional estimates to match global budget Combine chamber and/or turbulent flux estimates to match regional budget (e.g. CH 4 bottom up / top down comparison) Must consider not just total emissions, but also emissions source (e.g. agriculture, gas production, wetlands, gas seeps)

12. Methodological clarification Turbulent flux measurement –Eddy covariance or plume dispersion –Uses observations of the dispersion of a gas within the Atmospheric Boundary Layer (ABL) to infer source. –Typically used at 1km 2 domain or smaller Regional atmospheric budget/inversion –Uses changes in gas concentration over space and time in a well mixed ABL to infer sources. –More methodologically challenging –Suitable for “counties to continents”

13. Regional atmospheric budgets: a part of the needed toolbox of methods Aircraft budgets –Excellent spatial coverage –Limited temporal coverage Tower (or satellite) based atmospheric inversions –Excellent temporal coverage –Spatial coverage (domain, resolution) limited by density of long-term measurement network

14. Regional atmospheric measurements of CH 4 (and CO 2 ) sources and sinks

15. Regions of NOAA aircraft emissions estimates for natural gas production Uinta, UT Karion et al. 2013 (8.9%) Denver Julesburg, CO Petron et al. 2012 (4%) Petron et al. submitted Marcellus, PA/NY Ongoing work Barnett, TX Karion et al. in prep Haynesville, LA/TX Peischl et al. in prep Fayetteville, OK Peischl et al. in prep % values are estimated leakage rates as a fraction of production

16. Penn State regional tower-based measurement campaigns Midcontinent intensive, 2007-2009. Richardson et al (2012) Miles et al (2012) Lauvaux et al (2012a,b) Schuh et al (2013) Lauvaux and Davis, 2013 Diaz et al, in review INFLUX, 2010-201? Pubs in prep Gulf coast intensive(?), 2014-2016. Funds pending. N. American tower GHG network circa 2008 N. Marcellus. 2014-2016 Deployment stage

17. Aircraft Mass Balance Method Perpendicular wind speed Wind emissions Wind Background CH 4 Downwind CH 4 CH 4 flux Molar CH 4 enhancement in PBL References: White et al., 1976; Ryerson et al., 2001; Mays et al., 2009 mixing height (PBL)

18. June 1, 2011 Flight path Cambaliza et al, in reviewINFLUX, Purdue/Shepson group

19. Vertical Profiles of Potential Temperature and H 2 O (~ 1:00 to 1:30 p.m. EDT) 6 June, 2012 Vertical structure of the atmospheric boundary layer (ABL) Turbulent ABL ABL top, entrainment fluxes Stably stratified “free troposphere” Heat, water fluxes INFLUX, Purdue/Shepson group

20. Vertical Profiles of CO 2 and CH 4 (~ 1:00 to 1:30 p.m. EDT), 6 June, 2012 INFLUX, Purdue/Shepson group

21. 22,000 moles s -1 203 moles s -1 June 1, 2011 Results Cambaliza et al, in review INFLUX, Purdue/Shepson group

22. Methane Flux Matrix (June 1, 2011) City CH 4 Flux: 79.9 mol s -1 Landfill Flux: 38.3 mol s -1 Cambaliza et al, in prep INFLUX, Purdue/Shepson group

23. CH 4 Emission Flux from Indianapolis and contributions from Southside Landfill (SSLF) and Wastewater Treatment Plant (WWTP) INFLUX, Purdue/Shepson group

24. Utah, 2012 Distance perpendicular to wind (km) CH 4 (ppb) downwind upwind Karion et al. 2013 HRDL NOAA/Sweeney group

25. Uncertainty ParameterMean Value Variability (1  ) Relative Uncertainty wind speed (V)5.2 m/s1.2 m/s24% wind direction55.2°10.1° Vcos  3.8 m/s0.7 m/s24%  X CH4 56.3 ppb5.6 ppb10% BL depth1700 m125 m7% CH 4 Flux56 tonnes/hr15 tonnes/hr28% 8.9% of production is leaking in Uintah Basin Karion et al. 2013 NOAA/Sweeney group

26. Trace gases for attribution Ethane, propane, butane – associated with natural gas sources, but not agriculture or wetland sources 13CH4 – different sources have different isotopic ratios CO – associated with combustion NOAA/Sweeney group

27. Conclusions: Airborne budgets Powerful regional "snapshots" of total emissions - moderate levels of uncertainty. CH 4 source types can often be disaggregated with trace gases or source location data. Estimates to date suggest bottom-up methods underestimate total CH4 emissions. Why? Not clear at this point. Temporal variability difficult to capture. But if you sample a large enough area, perhaps statistics are in your favor. Little info on spatial distribution of emissions within the "box."

28. Tower-based “atmospheric inversion” CO 2 (and CH 4 ) source and sink estimates

29. Penn State regional tower-based measurement campaigns Midcontinent intensive, 2007-2009. Richardson et al (2012) Miles et al (2012) Lauvaux et al (2012a,b) Schuh et al (2013) Lauvaux and Davis, 2013 Diaz et al, in review INFLUX, 2010-201? Pubs in prep Gulf coast intensive(?), 2014-2016. Funds pending. N. American tower GHG network circa 2008 N. Marcellus. 2014-2016 Deployment stage

30. INFLUX objectives Develop improved methods for determination of urban area-wide, and spatially and temporally-resolved (e.g. monthly, 1 km 2 resolution) fluxes of greenhouse gases, specifically, CO 2 and CH 4. Determine and minimize the uncertainty in the emissions estimate methods.

31. Observational system 12 surface towers measuring CO 2 mixing ratios, 5 with CH 4, and 5 with CO. (Penn State) 4 eddy-flux towers from natural to dense urban landscapes. (Penn State) 5 automated flask samplers. (NOAA/CU) Periodic aircraft flights (~monthly) with CO2, CH4, and flask samples. (Purdue / NOAA) Periodic automobile surveys of CO2 and CH4. (Purdue) Doppler lidar. (NOAA/CU) TCCON-FTS for 4 months (Sept-Dec 2012). (NASA Ames)

32. INFLUX ground-based instrumentation Picarro, CRDS sensors; NOAA automated flask samplers; Communications towers ~100m AGL

33. Atmospheric inversions 101 Take a first guess at emissions Transport these through the atmosphere using an atmospheric model (reanalysis) Compute CH 4 at measurement points Compare modeled and observed CH 4 Adjust first guess of emissions to minimize the difference between observed and modeled CH 4.

34. Comparison of [CO2] at INFLUX sites 201120122013 Afternoon daily [CO2]

35. Afternoon [CO2] with 21-day smoothing Site 03 (downtown): high [CO2] Site 01 (background): low [CO2] Seasonal and synoptic cycles are evident Comparison of [CO2] at INFLUX sites

36. Site 09 measures 0.3 ppm larger than Site 01 Site 03 (downtown site) measures larger [CO2] by 3 ppm Spatial Structure of Urban CO2 Average [CO2] above background site East of city Downtown Afternoon daily values, 1 Jan – 1 April 2013 Eastern edge of city

37. Vulcan and Hestia Emission Inventories / Models Vulcan – hourly, 10km resolution for USA Hestia: high resolution emission data for the residential, commercial and industrial sectors, in addition to the transportation and electricity production sectors. See: Kevin Gurney/ http://hestia.project.asu.edu/ 250m res - Indy.

38. Combined sector temporally and spatially resolved Hestia emissions

39. Backward model results using footprints and Hestia 2002 fluxes Agreement in terms of the ordering of the sites Observations are 25% higher than modeled values, on average Average [CO2] above background site Spatial structure: Model-data comparison Miles et al, in prep

41. Lauvaux et al, in prep

42. CH4 Enhancement (Site 02 – Site 01) as a Function of Wind Direction April – November 2011 (Afternoon hours only)

43. Urban enhancement (Site 02 – Site 01): 100+ m AGL tower: CH4 7 ppb Green arrows point to the sources of enhanced CH4 measured at Site 02, compared to Site 01 Large source to the southeast of Site 02, as well as to the west (urban center) Maximum enhancements: ~ 10 ppb CH4

44. Conclusions: Atm inversions Capture total emissions, like airborne mass balance. strength and weakness. Can be used to quantify temporal variability in emissions over years Sources can be disaggregated via 1) trace gases or 2) prior knowledge of location and time of emissions. Can provide spatially resolve emissions - given sufficient atmospheric data density. “Footprints" are still relatively small and dependent on wind direction. Uncertainty assessment is complex, but MCI results show inversion uncertainties equal to those of agricultural inventories (Schuh et al., 2013).

45. Conclusions: Atm inversions INFLUX and N. Marcellus experiments are attempting to bring together (relatively) rigorous top-down and bottom-up assessments, and include both airborne and tower-based atmospheric inversions.

46. Overall summary Comparisons across methods / scales are important to gather understanding of the true net impacts on the global budget. Constructing a rigorous comparison is challenging. Aircraft are great for covering space, and for simple methods/uncertainty assessment, but are poor for temporal sampling. Towers are great for long-term monitoring and temporal variability, but the methodology is complex. In both cases, high measurement accuracy is required and sensors are costly. The effort is important - reducing GHG emissions is critical – we need to achieve (and verify) large reductions in C emissions.