Methane and Nitrous Oxide in North America: Using an LPDM to Constrain Emissions Eric Kort Non-CO2 Workshop October 23, 2008
Approach Atmospheric Atmospheric Measurements Measurements Use receptor oriented framework (STILT) to link measurements with emissions Use receptor oriented framework (STILT) to link measurements with emissions Forward Model Concentrations, Optimizations, Inversions Forward Model Concentrations, Optimizations, Inversions STILT developed by John Lin and Christoph Gerbig
Trajectories Receptor Points- Locations in space-time where measurements are made Receptor Points- Locations in space-time where measurements are made Release an ensemble of ‘particles’, which travel backwards in time, stochastically sampling the turbulence Release an ensemble of ‘particles’, which travel backwards in time, stochastically sampling the turbulence Driving wind fields Driving wind fields are of crucial are of crucial importance: importance: i.e. mass conservation i.e. mass conservation
Trajectories Meteorological Driver: WRF v2.2 Meteorological Driver: WRF v2.2 Use time-averaged mass fluxes- ensures good mass conservation Use time-averaged mass fluxes- ensures good mass conservation Uses analysis nudging to increase realism Uses analysis nudging to increase realism Turbulence included Turbulence included Release 500 particles backwards 10 days in time from each receptor Release 500 particles backwards 10 days in time from each receptor Case study used 100 particles 6 days back Case study used 100 particles 6 days back Using STILT, comparison w/ FLEXPART underway…, preliminary results encouraging, wind fields dominate answer, choice of LPDM does not strongly bias footprints
Go to Thomas Slides
Trajectories to Footprint
Footprint Critical Item which links measurements to emissions (Unit: ppb/flux) Critical Item which links measurements to emissions (Unit: ppb/flux) With this calculated can: With this calculated can: Interface w/ prior emissions field- bottom-up model concentrations Interface w/ prior emissions field- bottom-up model concentrations Follow this with simple scalar optimization Follow this with simple scalar optimization Or Perform a Bayesian optimization Or Perform a Bayesian optimization Or go straight to a Geostatistical Inversion Or go straight to a Geostatistical Inversion
Footprint * Prior Emission Field Result = Enhancement of gas at measurement point due to source *
Prior Emissions Fields Methane Methane Anthropogenic- EDGAR32FT2000 Anthropogenic- EDGAR32FT2000 Biogenic- Jed Kaplan wetland inventory Biogenic- Jed Kaplan wetland inventory Nitrous Oxide Nitrous Oxide Anthropogenic- EDGAR32FT2000 Anthropogenic- EDGAR32FT2000 Anthropogenic & Biogenic- GEIA Anthropogenic & Biogenic- GEIA FIRES
Boundary Condition To do even a Geostatistical Inversion, need ‘background’ values from where particles are 10 days back in time To do even a Geostatistical Inversion, need ‘background’ values from where particles are 10 days back in time Crucial to have good values here, as any error here directly propagates into any emissions analysis Crucial to have good values here, as any error here directly propagates into any emissions analysis Biases in particular are of large concern Biases in particular are of large concern
Boundary Condition Data-derived: Globalview type product (MBL, time/lat) Data-derived: Globalview type product (MBL, time/lat) Add vertical shape?? Add vertical shape?? Model-output: Forward model runs Model-output: Forward model runs Atmospheric Inversion output- carbontracker methane Atmospheric Inversion output- carbontracker methane
Boundary Condition Insights 2 Crucial Points Latitude Dependence Latitude Dependence Vertical gradient over ocean (for ch4) is negligible in comparison Vertical gradient over ocean (for ch4) is negligible in comparison Seasonal Variation Seasonal Variation This must be correct, in order to prevent seasonal biases This must be correct, in order to prevent seasonal biases Must check with measurement points in free troposphere with minimal surface influence– aircraft measurements are crucial Must check with measurement points in free troposphere with minimal surface influence– aircraft measurements are crucial
Bottom-Up Model Values Enhancement + Boundary Value = Modeled Mixing measurement point Enhancement + Boundary Value = Modeled Mixing measurement point Facilitates direct comparison, and optimization of emissions Facilitates direct comparison, and optimization of emissions
Case Study- COBRA-NA 2003 ~300 flasks NOAA/Boulder, UND Citation II, 23 May to 28 June 2003 ~300 flasks NOAA/Boulder, UND Citation II, 23 May to 28 June 2003
Measurements- Footprint
Results- Methane Slope: ± 0.13 Scaling Factor: 1.08 ± 0.15 Note: Prior Emissions Field EDGAR32FT 2000 & JK wetland
Results- Nitrous Oxide Slope: ± Scaling Factor: 2.62 ± 0.50 Note: Prior Emissions Field EDGAR32FT 2000, similar results using GEIA
But... Limitations in coverage Limitations in coverage Only a snapshot in time (May- June of 2003) Only a snapshot in time (May- June of 2003) Seasonality in agricultural Nitrous Oxide emissions is likely at play. Seasonality in agricultural Nitrous Oxide emissions is likely at play. Want to do with measurements over multiple years, get full seasonality picture. Want to do with measurements over multiple years, get full seasonality picture.
Concept Here Goal: Incorporate all measurements of CH4 and N2O over North America for Goal: Incorporate all measurements of CH4 and N2O over North America for Start: NOAA network, aircraft and tower flask samples, for 1 calendar year, CH4: under way Start: NOAA network, aircraft and tower flask samples, for 1 calendar year, CH4: under way Gives an initial framework from which to expand from Gives an initial framework from which to expand from Natural path is to start with same simple approach used previously Natural path is to start with same simple approach used previously
Combined Footprint, Aircraft Flasks, September 2006
Midday Footprint, LEF, Spring 04
Midday Footprint, AMT, Spring 04
Midday Footprint, WKT, Spring 04
Intensive Aircraft Campaigns & Continuous measurements Incorparation of Intensive Aircraft Campaigns and continuous measurments at towers can strongly supplement the flask measurement framework Incorparation of Intensive Aircraft Campaigns and continuous measurments at towers can strongly supplement the flask measurement framework Pre-HIPPO flight, from Rodrigo Jimenez
LEF mjj, Model Predictions LEF mjj, Model Predictions Model runs at 19 GMT Data Boundary, Note: -large day to day variation -dominance of anthropogenic emissions
LEF 2004 Model runs at 19 GMT Note Different slopes w/ different boundaries, indicating different seasonality in boundaries
Texas- model systematically too low
Maine: Model systematically too high
Acknowledgements Harvard Harvard Bruce Daube, Elaine Gottlieb, Steve Wofsy Bruce Daube, Elaine Gottlieb, Steve Wofsy AER AER Janusz Eluszkiewicz & Thomas Nehrkorn Janusz Eluszkiewicz & Thomas Nehrkorn MPI- Jena MPI- Jena Christoph Gerbig & Stefan Korner Christoph Gerbig & Stefan Korner Netherlands & Switzerland Netherlands & Switzerland Sander Houweling & Jed Kaplan Sander Houweling & Jed Kaplan NOAA & NCAR NOAA & NCAR Arlyn Andrews, Adam Hirsch, John B. Miller, Brit Stephens, Colm Sweeney, Lori Bruhwiler, Ed Dlugokencky, Pieter Tans Arlyn Andrews, Adam Hirsch, John B. Miller, Brit Stephens, Colm Sweeney, Lori Bruhwiler, Ed Dlugokencky, Pieter Tans U Michagan U Michagan Anna Michalak Anna Michalak University of Waterloo University of Waterloo John Lin John Lin