Using Space-based Observations to Better Constrain Emissions of Precursors of Tropospheric Ozone Dylan B. A. Jones, Paul I. Palmer, Daniel J. Jacob Division of Engineering and Applied Sciences and Department of Earth and Planetary Sciences Harvard University, Cambridge, MA Kevin W. Bowman, John Worden California Institute of Technology Jet Propulsion Laboratory, Pasadena, CA Ross N. Hoffman Atmospheric and Environmental Research, Inc. Lexington, MA Randall V. Martin, Kelly V. Chance Harvard-Smithsonian Center for Astrophysics Cambridge, MA
Rise in Tropospheric Ozone over the 20 th Century Observations at mountain sites in Europe [Marenco et al., 1994] Concentrations of O 3 have increased dramatically due to human activity
Impact of Human Activity on Background O 3 NO x Hydrocarbons (RH) NO x Hydrocarbons (RH) CONTINENT 2 OCEAN O3O3 Boundary layer (0-2.5 km) Free Troposphere CONTINENT 1 OH RO 2, HO 2 RH, CO NO NO 2 h O3O3 O3O3 Global Background O 3 Direct Intercontinental Transport greenhouse gas air pollution (smog)
“Bottom-up” Emission Inventories Biofuel Fossil Fuel Biomass Burning Bottom-up emissions E = (fuel burned) x (emission factors) Must extrapolate in space and time using population and economic data Emissions are highly uncertain
How Uncertain are Bottom-up Inventories? Source Tg CH 4 /yr Uncertainty Energy Animals Rice Biomass burning Landfills Wetlands Termites Other (hydrates, etc..) Total EDGAR 1995 CH 4 inventory [Olivier, 2002]
How Uncertain are Bottom-up Inventories? Source Tg CH 4 /yr Uncertainty Energy Animals Rice Biomass burning Landfills Wetlands Termites Other (hydrates, etc..) Total EDGAR 1995 CH 4 inventory [Olivier, 2002] Range of uncertainty for 1995 estimate [IPCC, 2001] Error based on 15% uncertainty in globally averaged OH
Inverse Modeling a priori “bottom-up” estimate E a a Monitoring site measures concentration C Atmospheric “forward” model gives C = kE “top-down” estimate E Best “a posteriori” estimate: Inverse model E = k -1 C “observational” error includes terms from instrument representativeness model parameters
PlatformmultipleERS-2TerraENVISATSpace station AuraTBD SensorTOMSGOMEMOPITTMODIS/ MISR SCIAMACHYMIPASSAGE-3TESOMIMLSCALIPSOOCO Launch O3O3 NN/LLL NL CONN/LL CO 2 N/LN NOL NO 2 NN/LN HNO 3 LL CH 4 NN/LN HCHONN/LN SO 2 NN/LN BrONN/LN HCNL aerosolNNNLNN Increasing spatial resolution N = Nadir L = Limb Satellite Observations of Tropospheric Chemistry
Global Ozone Monitoring Experiment (GOME) Launched April 1995 Nadir-viewing solar backscatter instrument ( nm) Polar sun-synchronous orbit Complete global coverage in 3 days Spatial resolution 320x40 km 2, three cross-track scenes
GEOS-CHEM Global 3-D Model Driven by assimilated meteorology from the NASA Data Assimilation Office Horizontal resolution of 2º latitude x 2.5º longitude O 3 -NO x -hydrocarbon chemistry Emission Inventories: Isoprene: Global Emission Inventory Activity (GEIA) [Benkovitz et al., 1996] CO: biofuels (Yevich and Logan, 2003), fossil fuels and biomass burning (Duncan et al., 2003)
GOME HCHO Columns for July 1996 Hot spots reflect high hydrocarbon emissions from fires and biosphere [10 16 molec cm -2 ] Dominant hydrocarbon sources: CH 4 Isoprene HCHO OH CO
GEOS-CHEM Relationship Between HCHO Columns and Isoprene Emissions in N. America Palmer et al. [2003] GEOS-CHEM Model, July 1996 (25-50ºN, ºW) NW NE SE SW Isoprene emission [10 13 atomC cm -2 s -1 ] Model HCHO Column [10 16 molec cm -2 ] Cont. from other sources Isoprene lifetime < 1 hour Isoprene emissions correlated with HCHO columns
Isoprene Emissions for July 1996 by Scaling of GOME Formaldehyde Columns GOME (official EPA inventory) Use linear relationship between isoprene emissions and HCHO columns from model to map GOME HCHO column to top-down isoprene emissions [Palmer et al., 2003] (a priori inventory in GEOS-CHEM) Top-down emissions provide a better fit to in situ observations of HCHO over US [10 12 atom C cm -2 s -1 ]
The Next Generation: Tropospheric Emission Spectrometer Will be launched in 2004 Infrared, high resolution Fourier spectrometer ( m) Polar, sun-synchronous orbit Makes orbits per day and repeats its orbit every 16 days Has 2 viewing modes: a nadir view and a limb view Spatial resolution of nadir measurement is 8 x 5 km 2
Assume that we know the true sources of CO Use GEOS-CHEM 3-D model to simulate “true” pseudo-atmosphere Sample pseudo-atmosphere along orbit of TES and simulate retrievals of CO Make a priori estimate of CO sources by applying errors to the “true” source Use an optimal estimation inverse method Obtain a posteriori sources and errors; How successful are we at finding the true source and reducing the error? APPROACH: OBJECTIVE: Determine whether nadir observations of CO from TES will have enough information to reduce uncertainties in our estimates of continental sources of CO
Inversion Analysis State Vector CHEM NAFF RWFFRWBB AFBB EUFF SABB ASFF ASBB NAFF:121.3SABB: 96.5 EUFF: RWBB: 98.0 ASFF: RWFF: ASBB: 96.0 CHEM: AFBB: “True” Emissions (Tg/yr) Total = 2270 Tg/yr FF = Fossil Fuel + Biofuel All sources include contributions from oxidation of VOCs OH is specified Use a “tagged CO” method to estimate contribution from each source Use inverse method to solve for annually-averaged emissions
Modeled CO Level 13 (6 km) 15 Mar. 2001, 0 GMT North American fossil fuel CO tracer
Sample the synthetic atmosphere along the orbit of TES Assume that the TES nadir footprint of 8 x 5 km is representative of the entire GEOS-CHEM 2º x 2.5º grid box We consider data only between the equator and 60ºN Assume cloud-free conditions CO at hPa for March 15, 2001 Simulating the TES Data
Generating the Nadir Retrievals of CO The retrieved profile is: A = averaging kernels y a = a priori profile y t = true profile from GEOS-CHEM true profile retrieved a priori
= retrieved profile x = estimate of the state vector (the CO sources) x a = a priori estimate of the CO sources (generated by perturbing true state) S a = error covariance of the a priori (assume a priori error of 50%) F(x) = forward model simulation of x S = error covariance of observations = instrument error + model error + representativeness error Unlike isoprene, CO has a long lifetime (weeks – months) so transport is important need a more formal inversion analysis Minimize a maximum a posteriori cost function Inversion Methodology
The forward Model (GEOS-CHEM) must account for the vertical sensitivity of TES and the a priori information used in the retrievals The Forward Model i = Retrieval noise (< 10%, based on spectral noise of instrument) Must add noise to forward model to account for errors in GEOS-CHEM simulation of CO r = representativeness error m = model error H(x) = GEOS-CHEM model
root mean square error based on the difference of 89 pairs of 48-hr and 24-hr forecasts of CO during Feb-April 2001 Model Level 16 (8 km) CO Mixing Ratio (ppb) The “NMC method” assume that the difference between successive forecasts, which are valid at the same time, is representative of the forecast error structure Constructing the Covariance Matrix for the Model Error
Compare model with observations of CO from TRACE-P and estimate the relative error Assume mean bias in relative error is due to emission errors Remove mean bias and assume the residual relative error (RRE) is due to transport Use RRE to scale forecast error structure Characterizing the Model Error Representativeness error = 5%, based on sub-grid variability of TRACE-P data Mean forecast error RRE
Model Error (Feb-April, 2001)
Inversion Results: 4 Days of Pseudo-Data With unbiased Gaussian error statistics, TES is extremely powerful for constraining continental sources of CO
Conclusions Current space-based observations provide valuable information to help quantify emissions of O 3 precursors Observations from TES and future high resolution satellite instruments have the potential to significantly enhance our ability to constrain regional source estimates of O 3 precursors Realization of this potential rests on proper error characterization, particularly for the model transport error Will require an integrated approach – in situ (surface and aircraft) and space-based observations THE FUTURE: Multi-species chemical data assimilation Integrate observations from surface sites, aircraft, and satellites (low Earth orbit and geostationary) to construct an optimized 4-D view of the chemical state of the atmosphere
EXTRA SLIDES
Model Transport Error Model Level 16 (8 km) CO Mixing Ratio (ppb) RMS error based on the difference of 89 pairs of 48-hr and 24-hr forecasts of CO during TRACE-P RMS error b/w model and MOPITT CO column during TRACE-P x10 18 molec/cm 2 (Colette Heald)
The forward Model (GEOS-CHEM) must account for the vertical sensitivity of TES and the a priori information used in the retrievals The Forward Model G n = Retrieval noise G = gain matrix associated with A = standard deviation of spectral noise S i = G( I)G T n = Gaussian white noise with unity variance Must add noise to forward model to account for errors in GEOS-CHEM simulation of CO r = representativeness error m = model error H(x) = GEOS-CHEM model
CO Source Distribution in GEOS-CHEM Fossil Fuels Biofuels Biomass Burning Other Sources: oxidation of methane and other VOCs Main sink: reaction with atmospheric OH