Research ArticleCLIMATOLOGY

Enhanced North American carbon uptake associated with El Niño

See allHide authors and affiliations

Science Advances  05 Jun 2019:
Vol. 5, no. 6, eaaw0076
DOI: 10.1126/sciadv.aaw0076
  • Fig. 1 Variability of monthly anomalies of atmospheric CO2 mole fractions and δ13CO2 observations shows a broadly consistent response to ENSO.

    (A) The ONI. (B and C) Six-month running averages of monthly CO2 mole fraction and δ13CO2 anomalies averaged across NOAA’s long-term flask air sampling sites over North America (table S1). The number of sites included to calculate the monthly average anomalies of CO2 and δ13CO2 is 7 to 12 for 1995–2003, 16 to 19 for 2004–2007, and 25 to 30 for 2008–2015. Gray shading indicates standard errors of the calculated 6-month running average anomalies. Light yellow shading indicates El Niño periods, whereas light blue indicates La Niña periods.

  • Fig. 2 Anomalies of derived NEE over North America display distinct responses to different phases of ENSO between 2007 and 2015.

    (A) Derived monthly NEE anomalies from 18 inversions considered in this study from CT-L (gray shading), their monthly means (black thin line), and 6-month running averages (black thick line). Blue and red solid lines indicate the 6-month running averages of monthly NEE anomalies derived for boreal and temperate North America. Light yellow shading indicates El Niño, whereas light blue indicates La Niña. (B) Average monthly NEE anomalies during El Niño, neutral, and La Niña periods [which were indicated by yellow, white, and blue shadings in (A)], derived from TBMs (CASA GFEDv4.1s, CASA GFED-CMS, and SiBCASA), CarbonTracker (CT2016), and CT-L (this study). Anomalies derived from CT-L were indicated by the boxplot with red lines representing the medians, blue boxes indicating the 25th and 75th percentiles, and black dashed bars indicating the minimums and maximums of the ranges calculated from the 18 inversions.

  • Fig. 3 Climate drivers for anomalies of North American NEE and their response to ENSO.

    (A) Correlation coefficients between NEE anomalies and anomalies of air temperature, precipitation, VPD, and SM in different seasons using prior fluxes (computed from TBMs) and fluxes derived from CT-L. Strong correlations with P < 0.1 are indicated by color-filled symbols, whereas weaker correlations with P > 0.1 are shown in empty symbols. (B) Differences of average air temperature anomalies between El Niño and non–El Niño (La Niña and neutral conditions) periods in spring and summer and difference of average VPD anomalies between El Niño and non–El Niño periods averaged for all seasons. While the impact of El Niño on North American air temperature is opposite between spring and summer, its impact on North American VPD (and other hydrological variables) is relatively more constant throughout the year (fig. S14).

  • Table 1 Correlation coefficients between CO2 flux anomalies over North America and anomalies of area-weighted average air temperature, precipitation, RH, VPD, and SM over temperate North America, using prior fluxes (CASA GFEDv4.1s) and derived posterior fluxes from CT-L.

    Correlations were calculated with fluxes lagging climate variables by 1 month. The P value associated with each correlation is included in parentheses. The correlation is higher for yearly average anomalies than that for monthly anomalies as noise in the datasets are smoothed out.

    Climate variablesPrior fluxes (monthly)Posterior fluxes (monthly)Posterior fluxes (yearly)
    Air temperature−0.09 (0.34)−0.01 (0.90)0.30 (0.43)
    Precipitation−0.03 (0.78)−0.21 (0.03)−0.78 (0.01)
    RH−0.14 (0.15)−0.38 (<0.0001)−0.69 (0.04)
    VPD0.16 (0.09)0.39 (<0.0001)0.75 (0.02)
    SM−0.23 (0.02)−0.32 (0.001)−0.66 (0.05)

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/6/eaaw0076/DC1

    Fig. S1. Air sampling sites for measurements of atmospheric CO2 mole fractions and δ13CO2 under NOAA’s Global Greenhouse Gas Reference Network between 2007 and 2015.

    Fig. S2. Monthly and annual NEE used as prior fluxes in this study.

    Fig. S3. Estimates of the North American land sink for 2000 to 2015 from published studies [Butler et al. (19), Gourdji et al. (20), and Peylin et al. (18), and King et al. (17)], recent releases of global inverse models (CAMS, Jena CarboScope, CT2016, and CTE2016), and from this study (CT-L).

    Fig. S4. Multiyear average annual NEE (blue bars) and differences of NEE anomalies between El Niño and La Niña periods (red bars) between 2007 and 2015 for different ecoregions defined in fig. S1.

    Fig. S5. IAV and ENSO response of North American NEE simulated by TBMs and atmosphere inverse models.

    Fig. S6. Residuals between simulated and observed CO2 mole fractions.

    Fig. S7. Vertical profiles indicating seasonally averaged residuals (differences) between simulated and observed CO2 mole fractions.

    Fig. S8. Comparison of inversion results from OSSEs based on different model designs.

    Fig. S9. Differences between posterior and true fluxes derived from observing system simulation experiments.

    Fig. S10. Anomalies of NEE for North America derived from CarbonTracker (CT2016: black; CT2017: red).

    Fig. S11. Correlation coefficients between NEE anomalies and anomalies of air temperature, precipitation, VPD, RH, and SM for spring (March to May), summer (June to August), fall (September to November), and winter (December to February) for major biome types over North America that have larger carbon uptake (indicated in fig. S4).

    Fig. S12. Correlations between monthly NEE anomalies and monthly anomalies of hydrological variables (precipitation, relative humidity, vapor pressure deficit, and soil moisture) and the sensitivity of NEE anomalies to hydrological conditions.

    Fig. S13. Correlations between NEE anomalies and anomalies of air temperature in spring and summer, and the sensitivity of NEE anomalies to air temperature.

    Fig. S14. Difference of anomalies of air temperature, precipitation, relative humidity, vapor pressure deficit, and soil moisture between El Niño and non El Niño (neutral and La Niña) periods.

    Fig. S15. Correlations between the ONI and 3-month average anomalies of area-weighted average precipitation, RH, SM, and VPD over boreal (blue symbols) and temperate (red symbols) North America with ±10-month time lags.

    Fig. S16. Correlations between the ONI and 3-month average anomalies of area-weighted average precipitation, RH, SM, and VPD over temperate North America for every 20 years between 1950 and 2016.

    Fig. S17. Atmospheric CO2 observations between 2007 and 2015 with different colors indicating their total sensitivity [sumH, ppm (μmol m−2 s−1)−1] to North American land fluxes.

    Fig. S18. Observed CO2 mole fractions for observations used in this analysis and their associated background estimates.

    Table S1. Site information for CO2 mole fraction and δ13CO2 measurements made from NOAA flask air samples.

    Table S2. Prior NEE, error covariance parameters, and background CO2 mole fractions used in the 18 inversion ensemble members in this study.

    Table S3. Correlations between prior and posterior NEE anomalies over North America and anomalies of area-weighted average precipitation, RH, VPD, and SM over temperate North America.

    Table S4. Correlations between the ONI and 3-month average anomalies of area-weighted average air temperature, precipitation, RH, VPD, and SM over boreal and temperate North America.

  • Supplementary Materials

    This PDF file includes:

    • Fig. S1. Air sampling sites for measurements of atmospheric CO2 mole fractions and δ13CO2 under NOAA’s Global Greenhouse Gas Reference Network between 2007 and 2015.
    • Fig. S2. Monthly and annual NEE used as prior fluxes in this study.
    • Fig. S3. Estimates of the North American land sink for 2000 to 2015 from published studies Butler et al. (19), Gourdji et al. (20), and Peylin et al. (18), and King et al. (17), recent releases of global inverse models (CAMS, Jena CarboScope, CT2016, and CTE2016), and from this study (CT-L).
    • Fig. S4. Multiyear average annual NEE (blue bars) and differences of NEE anomalies between El Niño and La Niña periods (red bars) between 2007 and 2015 for different ecoregions defined in fig. S1.
    • Fig. S5. IAV and ENSO response of North American NEE simulated by TBMs and atmosphere inverse models.
    • Fig. S6. Residuals between simulated and observed CO2 mole fractions.
    • Fig. S7. Vertical profiles indicating seasonally averaged residuals (differences) between simulated and observed CO2 mole fractions.
    • Fig. S8. Comparison of inversion results from OSSEs based on different model designs.
    • Fig. S9. Differences between posterior and true fluxes derived from observing system simulation experiments.
    • Fig. S10. Anomalies of NEE for North America derived from CarbonTracker (CT2016: black; CT2017: red).
    • Fig. S11. Correlation coefficients between NEE anomalies and anomalies of air temperature, precipitation, VPD, RH, and SM for spring (March to May), summer (June to August), fall (September to November), and winter (December to February) for major biome types over North America that have larger carbon uptake (indicated in fig. S4).
    • Fig. S12. Correlations between monthly NEE anomalies and monthly anomalies of hydrological variables (precipitation, relative humidity, vapor pressure deficit, and soil moisture) and the sensitivity of NEE anomalies to hydrological conditions.
    • Fig. S13. Correlations between NEE anomalies and anomalies of air temperature in spring and summer, and the sensitivity of NEE anomalies to air temperature.
    • Fig. S14. Difference of anomalies of air temperature, precipitation, relative humidity, vapor pressure deficit, and soil moisture between El Niño and non El Niño (neutral and La Niña) periods.
    • Fig. S15. Correlations between the ONI and 3-month average anomalies of area-weighted average precipitation, RH, SM, and VPD over boreal (blue symbols) and temperate (red symbols) North America with ±10-month time lags.
    • Fig. S16. Correlations between the ONI and 3-month average anomalies of area-weighted average precipitation, RH, SM, and VPD over temperate North America for every 20 years between 1950 and 2016.
    • Fig. S17. Atmospheric CO2 observations between 2007 and 2015 with different colors indicating their total sensitivity sumH, ppm (μmol m−2 s−1)−1 to North American land fluxes.
    • Fig. S18. Observed CO2 mole fractions for observations used in this analysis and their associated background estimates.
    • Table S1. Site information for CO2 mole fraction and δ13CO2 measurements made from NOAA flask air samples.
    • Table S2. Prior NEE, error covariance parameters, and background CO2 mole fractions used in the 18 inversion ensemble members in this study.
    • Table S3. Correlations between prior and posterior NEE anomalies over North America and anomalies of area-weighted average precipitation, RH, VPD, and SM over temperate North America.
    • Table S4. Correlations between the ONI and 3-month average anomalies of area-weighted average air temperature, precipitation, RH, VPD, and SM over boreal and temperate North America.

    Download PDF

    Files in this Data Supplement:

Navigate This Article