A Lagrangian Particle Dispersion Model Approach for Evaluating - PowerPoint PPT Presentation

A Lagrangian Particle Dispersion Model Approach for Evaluating CarbonTracker A. Andrews 1 , A. Hirsch 2 , A. Michalak 3 , C. Sweeney 2 , S. Wofsy 4 , J. Eluszkiewicz 5 , T. Nehrkorn 5 , A. Jacobson 2 , K. Masarie 1 , W. Peters 2,6 , K. Croasdale 7 , P. Tans 1 1 NOAA Earth System Research Laboratory, 325 Broadway, Boulder, CO 80305 2 Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, CO 80309 3 Department of Civil and Environmental Engineering, and Department of Atmospheric, Oceanic and Space Sciences, University of Michigan, Ann Arbor, MI 48109 4 Department of Earth and Planetary Sciences, Harvard University, Cambridge, MA 02138 5 Atmospheric and Environmental Research, Inc., Lexington, MA 02421 6 Wageningen University and Research Center, Wageningen, 6708 PB, the Netherlands 7 Department of Environmental Sciences, University of Illinois, Springfield, IL 62703

CarbonTracker Overview Optimization step is Ensemble Kalman Filter http://carbontracker.noaa.gov http://carbontracker.noaa.gov

Mechanics Optimized FIRES= GFED2 Specified http://carbontracker.noaa.gov http://carbontracker.noaa.gov

Mechanics Optimized FIRES= GFED2 Specified http://carbontracker.noaa.gov http://carbontracker.noaa.gov

Sampling Footprints Lagrangian Particle Dispersion Modeling Stochastic Time Inverted Lagrangian Transport Model Sensitivity = 0 Sensitivity = 0 Sensitivity > 0 Sensitivity > 0 PBL height PBL height Altitude (m) -24 hours -48 hours -72 hours -96 hours Time (hours) -120 hours t N 1 ∑∑ f = + meas BC CO CO SENS * FLUX 2 2 N = = p 0 t 0 CarbonTracker Footprint Initial Condition [ppm per unit flux] (CarbonTracker)

Sampling Footprints Lagrangian Particle Dispersion Modeling Stochastic Time Inverted Lagrangian Transport Model Sensitivity = 0 Sensitivity = 0 Sensitivity > 0 Sensitivity > 0 PBL height PBL height Altitude (m) -24 hours -48 hours -72 hours -96 hours Time (hours) -120 hours t N 1 ∑∑ f = + meas BC CO CO SENS * FLUX 2 2 N = = p 0 t 0 CarbonTracker Footprint Initial Condition [ppm per unit flux] (CarbonTracker)

Normalized Footprint: Linear Color Scale Composite: MAY-JULY 2004 LEF, 19GMT Note: Footprints are time-resolved, but shown as 10-day composite

Normalized Footprint: Log10 Color Scale Composite: MAY-JULY 2004 LEF, 19GMT

0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7 -0.9 Cumulative Surface Sensitivity

0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7 -1.0

STILT BOUNDARY LAYER RESIDENCE TIME 0 -2 < Model Mixed Layer Height -4 Days -6 -8 <2km <1km -10 Fraction of Particles

LEF TOWER: PARK FALLS, WI 396 magl OBS CT

OBS CT

OBS CT STILT-WRF-CT CT ~4KM AC OBS >3.5 KM STILT-BKGRD

MLO COBRA 2004 (Harvard)

AC OBS >3.5 KM STILT-BKGRD CT ~4KM STILT OBS CT

, ppm OBSERVED CO 2 STILT R 2 =0.89 CT R 2 =0.87 MODELED CO 2 , ppm

AMT: JUNE-JULY 2004 19:00 GMT (TOWER) OBS CT STILT CT ~4KM AC OBS >3.5 KM STILT-BKGRD

AMT: JUNE-JULY 2004 19:00 GMT (TOWER) OBS CT STILT + COBRA 2004 CT ~4KM (HARVARD) AC OBS >3.5 KM STILT-BKGRD

CT R 2 =0.56 (0.67) STILT R 2 =0.29 (0.65) OBS CT STILT CT ~4KM AC OBS >3.5 KM STILT-BKGRD

AMT: JUNE-JULY 2004 19:00 GMT (TOWER) AMT: JUNE-JULY 2004 19:00 GMT (TOWER) JULY 18, 19:00 GMT OBS OBS CT CT STILT STILT CT ~4KM CT ~4KM AC OBS >3.5 KM AC OBS >3.5 KM STILT-BKGRD STILT-BKGRD

AMT: JUNE-JULY 2004 19:00 GMT (TOWER) AMT: JUNE-JULY 2004 19:00 GMT (TOWER) JULY 19, 19:00 GMT OBS OBS CT CT STILT STILT CT ~4KM CT ~4KM AC OBS >3.5 KM AC OBS >3.5 KM STILT-BKGRD STILT-BKGRD

AMT: JUNE-JULY 2004 19:00 GMT (TOWER) AMT: JUNE-JULY 2004 19:00 GMT (TOWER) JUly 25, 19:00 GMT OBS OBS CT CT STILT STILT CT ~4KM CT ~4KM AC OBS >3.5 KM AC OBS >3.5 KM STILT-BKGRD STILT-BKGRD

COBRA-MAINE: HARVARD UNIVERSITY/NSF & NOAA 7 MAY – 16 AUGUST 2004 33 FLIGHTS Wyoming King Air NSF paid for equipment & establishing AMT tower site under this project

COBRA: 2004-07-30 AMT OVERFLIGHT + +

COBRA: 2004-07-30 300m agl +

COBRA: 2004-07-30 3200m agl + Free troposphere has much smaller surface sensitivity than boundary layer � strongly driven by initial condition.

July/Aug 2004 4000 OBS Altitude, m above ground level STILT-WRF-CT CT interp 3000 2000 May/Jun 2004 OBS 1000 STILT-WRF-CT CT interp 0 360 365 370 375 380 385 Carbon Dioxide, ppm

July/Aug 2004 4000 OBS Altitude, m above ground level STILT-WRF-CT CT interp 3000 More/Earlier drawdown needed in Spring? 2000 May/Jun 2004 OBS 1000 STILT-WRF-CT CT interp 0 360 365 370 375 380 385 Carbon Dioxide, ppm

July/Aug 2004 4000 OBS Altitude, m above ground level STILT-WRF-CT CT interp 3000 2000 May/Jun 2004 OBS 1000 STILT-WRF-CT CT interp 0 360 365 370 375 380 385 •Differences are due to model transport. •Keep in mind that fluxes have been optimized Carbon Dioxide, ppm using CT. •Aircraft data were not optimized.

Summary Points •Lagrangian models can provide insight into the mechanics of CarbonTracker •Diagnose patterns and residuals •Footprint information can inform decisions about how to weight different types of data •Generally good agreement between STILT-WRF-CT and CarbonTracker is encouraging—STILT tends to predict lower CO 2 near the surface •“Campaign” data are valuable independent datasets for CarbonTracker evaluation: e.g, COBRA-2003, COBRA-2004, TEXAS AQS 2006, ARCPAC- 2008, START-08, HIPPO •Footprints are generic and can be used to interpret mixing ratio measurements of other species (halocarbons, COS, isotopes, etc.) •Other LPDMS are in use around ESRL: FLEXPART, CSU LPDM, HYSPLIT •We are working toward building an archive of footprints that can be linked to the GMD database

AC OBS >3.5 KM STILT-BKGRD CT ~4KM STILT OBS CT

AC OBS >3.5 KM STILT-BKGRD CT ~4KM STILT OBS CT

AC OBS >3.5 KM STILT-BKGRD CT ~4KM STILT OBS CT

STILT R 2 =0.65 (0.85) CT R 2 =0.64 (0.87)