Quantifying the Impact of Atmospheric Transport Uncertainty on CO2 Surface Flux Estimates - PubMed (original) (raw)

. 2019 Apr;33(4):484-500.

doi: 10.1029/2018GB006086. Epub 2019 Apr 16.

Andrew R Jacobson 2, Sourish Basu 2, Brad Weir 3, David Baker 1, Kevin Bowman 4, Frédéric Chevallier 5, Sean Crowell 6, Kenneth J Davis 7, Feng Deng 8, Scott Denning 9, Liang Feng 10 11, Dylan Jones 8, Junjie Liu 4, Paul I Palmer 10 11

Affiliations

Quantifying the Impact of Atmospheric Transport Uncertainty on CO2 Surface Flux Estimates

Andrew E Schuh et al. Global Biogeochem Cycles. 2019 Apr.

Abstract

We show that transport differences between two commonly used global chemical transport models, GEOS-Chem and TM5, lead to systematic space-time differences in modeled distributions of carbon dioxide and sulfur hexafluoride. The distribution of differences suggests inconsistencies between the transport simulated by the models, most likely due to the representation of vertical motion. We further demonstrate that these transport differences result in systematic differences in surface CO2 flux estimated by a collection of global atmospheric inverse models using TM5 and GEOS-Chem and constrained by in situ and satellite observations. While the impact on inferred surface fluxes is most easily illustrated in the magnitude of the seasonal cycle of surface CO2 exchange, it is the annual carbon budgets that are particularly relevant for carbon cycle science and policy. We show that inverse model flux estimates for large zonal bands can have systematic biases of up to 1.7 PgC/year due to large-scale transport uncertainty. These uncertainties will propagate directly into analysis of the annual meridional CO2 flux gradient between the tropics and northern midlatitudes, a key metric for understanding the location, and more importantly the processes, responsible for the annual global carbon sink. The research suggests that variability among transport models remains the largest source of uncertainty across global flux inversion systems and highlights the importance both of using model ensembles and of using independent constraints to evaluate simulated transport.

Keywords: GEOS‐Chem; OCO‐2; TM5; atmosphere; carbon; inversions.

PubMed Disclaimer

Figures

Figure 1

Figure 1

Plot (a) shows simulated CO2 in micromoles per mole from GEOS‐Chem (left column) and TM5 (right column), averaged zonally and over the month of January 2001 (top panels) and August 2001 (bottom panels) as a function of latitude and altitude from the simulations described in section 2.2. Differences are shown in plot (b). Both models use a terrain‐following sigma vertical coordinate. For simplicity, the

y

axis shows the 47 GEOS‐Chem model levels with approximate pressure levels at sea level.

Figure 2

Figure 2

GEOS‐Chem minus TM5 difference in simulated fossil fuel CO2 in micromoles per mole, averaged zonally and over the month of March 2008 as a function of latitude and pressure. Both simulations used CT2016 fossil fluxes. The vertical axis of the plot (b) represents equally spaced pressure levels with the label corresponding to the equivalent sea surface level grid for the sigma terrain‐following coordinates. Panel (a) shows the GEOS‐Chem minus TM5 difference in pressure‐weighted average XCO2 as a function of latitude (solid black line). This is equivalent to the vertical average of the field shown in panel (b), with the exception that panel (b) is area weighted from a value of one near the equator to zero at the poles. Panel (c) shows the meridional‐averaged difference as a function of pressure level. The dashed lines in panels (a) and (c) show the approximate effect of computing XCO2 using an estimated Orbiting Carbon Observatory‐2 (OCO‐2) averaging kernel as a function of solar zenith angle and latitude. The blue line in top panel (a) shows the pressure‐weighted average of the bottom five model levels. The OCO‐2 averaging kernel (AK) is only defined where OCO‐2 observations exist; thus, we estimated the AK as a function of latitude and solar zenith angle in order to be able to apply to arbitrary atmospheric columns.

Figure 3

Figure 3

Similar to Figure 2 but for the sum of all CT2016 nonfossil fuel CO2 tracers (land biosphere, fires, and ocean) for March 2008 and September 2008. Top panel: Black line is full column XCO2, black dotted line is with addition of OCO‐2 averaging kernel, and blue line is the column CO2 average for the bottom five model levels.

Figure 4

Figure 4

Zonal‐mean GEOS‐Chem minus TM5 differences in simulated XCO2 in micromoles per mole. These Hovmoller plots show the dominant latitude and time differences of simulated XCO2 between the two models. (a) Summed differences in the land biosphere, fire, and ocean tracers and excludes the fossil fuel tracer. (b) Differences in the fossil tracer alone. (c) The total signal by summing all CO2 tracers. The three plots on the left side of the figure show the average difference over the 4 years as a function of latitude, area weighted from a value of 1 near the equator to 0 near the poles. The figure is the same for all three rows with green being the biological, fires, and ocean signal, red being the fossil signal, and black being the total signal.

Figure 5

Figure 5

The 9‐year average model‐minus‐observed residuals of SF6 at the marine boundary layer sites listed in Table S1, arranged by latitude. Results are from two different versions of TM5 with ERA‐i meteorology (gold, an earlier version subject to a fault in convective transport; red, a version correcting that fault) and GEOS‐Chem using MERRA2 (blue).

Figure 6

Figure 6

Flux inversion monthly fluxes from the Orbiting Carbon Observatory‐model intercomparison project partitioned by transport model and latitude band. Thin blue lines represent GEOS‐Chem models, and thin red lines are TM5 models. Thick blue and red lines represent the ensemble mean for GEOS‐Chem and TM5 models, respectively. Upper panels are for fluxes integrated across 45°N to the north pole, and lower panels represent fluxes integrated from the equator to 45°N. Left panels show results for inversions assimilating traditional in situ CO2 measurements and right panels for inversions assimilating Orbiting Carbon Observatory‐2 land nadir retrievals of XCO2.

Figure 7

Figure 7

Annual flux average (2015–2016) from Orbiting Carbon Observatory‐2 model intercomparison project suite. IS refers to inversions constrained with traditional in situ observations, and LN refers to inversions constrained with XCO2 retrievals from Orbiting Carbon Observatory‐2 in its land nadir observing mode. The box and whiskers plot shows a box which roughly approximates the first and third quartiles of the data and whiskers which extend to the most extreme data point which is no more than 1.5 times the length of the box away from the box.

Figure 8

Figure 8

The effect of monthly average climatological transport bias on a simple inversion of XCO2. The black line is the monthly four‐model mean GEOS‐Chem minus three‐model mean TM5 flux difference from the LN experiment of the Orbiting Carbon Observatory‐2 model intercomparison project. The red line is the result of inverting a smoothed version, by month and 5° latitude band, of the XCO2 difference plotted in Figure 4, using the Schuh GEOS‐Chem based inversion framework. “Inversion adjustment” results are smoothed with 3‐month boxcar average. Results are in Teragrams carbon per day.

References

    1. Ballantyne, A. P. , Alden, C. B. , Miller, J. B. , Tans, P. P. , & White, J. W. C. (2012). Increase in observed net carbon dioxide uptake by land and oceans during the past 50 years. Nature, 488(7409), 70–72. 10.1038/nature11299 -DOI -PubMed
    1. Barnes, E. A. , Parazoo, N. , Orbe, C. , & Denning, A. S. (2016). Isentropic transport and the seasonal cycle amplitude of CO2 . Journal of Geophysical Research: Atmospheres, 121, 8106–8124. 10.1002/2016JD025109 -DOI
    1. Basu, S. , Baker, D. F. , Chevallier, F. , Patra, P. K. , Liu, J. , & Miller, J. B. (2017). The impact of transport model differences on CO2 surface flux estimates from OCO‐2 retrievals of column average CO2 . Atmospheric Chemistry and Physics Discussions, 18, 7189–7215.
    1. Bey, I. , Jacob, D. J. , Yantosca, R. M. , Logan, J. A. , Field, B. D. , Fiore, A. M. , Li, Q. , Liu, H. Y. , Mickley, L. J. , & Schultz, M. G. (2001). Global modeling of tropospheric chemistry with assimilated meteorology: Model description and evaluation. Journal of Geophysical Research, 106(D19), 23,073–23,095.
    1. Bosilovich, M. G. (2015). Technical report series on global modeling and data assimilation, volume 43 MERRA‐2: Initial evaluation of the climate. NASA‐GMAO. Retrieved from https://gmao.gsfc.nasa.gov/pubs/docs/Bosilovich803.pdf

LinkOut - more resources