Downscaling the NOAA CarbonTracker Inversion for North America

Author: G. Petron, , Thomas Nehrkorn and John M. Henderson
Date: 
December 7, 2011
Type: 
Poster presentation
Venue: 
AGU Fall Meeting 2011
Citation: 

Gabrielle Petron; Arlyn E. Andrews; Huilin Chen; Michael E. Trudeau; Janusz Eluszkiewicz; Thomas Nehrkorn; John Henderson; Colm Sweeney; Anna Karion; Ken Masarie; Lori Bruhwiler; John B. Miller; Benjamin R. Miller; Wouter Peters; Sharon M. Gourdji; Kimberly L. Mueller; Anna M. Michalak; Pieter P. Tans (2011) Downscaling the NOAA CarbonTracker Inversion for North America. AGU Fall Meeting 2011.
 

We are developing a regional extension of the NOAA CarbonTracker CO2 data-assimilation system for a limited domain covering North America. The regional assimilation will use pre-computed and species-independent atmospheric sampling footprints from a Lagrangian Particle Dispersion Model. Each footprint relates an observed trace gas concentration to upwind fluxes. Once a footprint library has been computed, it can be used repeatedly to quickly test different inversion strategies and, importantly, for inversions using multiple species data (e.g., anthropogenic tracers such as radiocarbon and carbon monoxide and biological tracers such as carbonyl sulfide and stable isotopes of CO2).

The current global CarbonTracker (CT) assimilation framework has some important limitations. For example, the assimilation adjusts scaling factors for different vegetation classes within large regions. This means, for example, that all crops within temperate North America are scaled together. There is currently no distinction between crops such as corn and sorghum, which utilize the C4 photosynthesis pathway and C3 crops like soybeans, wheat, cotton, etc. The optimization scales only the net CO2 flux, rather than adjusting photosynthesis and respiration fluxes separately, which limits the flexibility of the inversion and sometimes results in unrealistic diurnal cycles of CO2 flux. The time-series of residuals (CT – observed) for continental sites in North America reveals a persistent excess of CO2 during summer. This summertime positive bias is also apparent in the comparison of CT posterior CO2 with aircraft data and with data from Pacific marine boundary layer sites, suggesting that some of the problem may originate outside of North America.

For the regional inversion, we will use footprints from the Stochastic Time-Inverted Lagrangian Transport Model driven by meteorological fields from a customized high-resolution simulation with the Weather Research Forecast (WRF) model. We will use empirically corrected boundary conditions in order to minimize sensitivity to inaccurate fluxes or transport outside of our domain. We plan to test a variety of inversion strategies that effectively exploit CO2 and isotopic data from the relatively dense North American sampling network for 2007-2010.