Technical Comments

Response to Comment on “Contrasting carbon cycle responses of the tropical continents to the 2015–2016 El Niño”

See allHide authors and affiliations

Science  30 Nov 2018:
Vol. 362, Issue 6418, eaat1211
DOI: 10.1126/science.aat1211


Chevallier showed a column CO2 (Embedded Image) anomaly of ±0.5 parts per million forced by a uniform net biosphere exchange (NBE) anomaly of 2.5 gigatonnes of carbon over the tropical continents within a year, so he claimed that the inferred NBE uncertainties should be larger than presented in Liu et al. We show that a much concentrated NBE anomaly led to much larger Embedded Image perturbations.

Chevallier has asserted that the uncertainties presented in Liu et al. (1) are too small, sufficiently so that the scientific conclusions are less robust than reported. Chevallier (2) showed that uniformly redistributing a net biosphere exchange (NBE) anomaly of 2.5 gigatonnes of carbon (Gt C) (1) over the tropical continents throughout a year could result in concentration signals of ±0.5 parts per million (ppm). He then speculated that systematic errors in retrievals and transport models may be similar in magnitude, and therefore argued that the uncertainties should be higher.

However, the uniformly distributed flux anomaly proposed by Chevallier is very different from the actual distribution of the biosphere flux anomaly during the 2015–2016 El Niño event. In fact, the NBE anomaly was highly concentrated in time and space, and so led to much larger column enhancements than Chevallier’s simulation. Chevallier simulated a source of 0.27 g C m−2 day−1 to the atmosphere over the tropical continents, balanced by a regular sink of 0.01 g C m−2 day−1 elsewhere. However, the actual anomalous CO2 release over the tropics primarily occurred over small areas that experienced extreme climate conditions [figure 4 of (1)] during the period September 2015–March 2016 [figure S1 of (1)]. For example, the flux anomaly was more than 2.0 g C m−2 day−1, nearly 10 times Chevallier’s value, over southwest Kalimantan Island, Indonesia in October and November, when the peak biomass burning occurred.

We carried out an experiment similar to that of Chevallier, using the inferred, spatiotemporally varying real NBE anomaly from the 2015–2016 El Niño as a boundary condition over all three tropical continents. We then sampled the Embedded Image perturbations using the sensitivity and locations of OCO-2 observations. Figure 1A shows the mean Embedded Image perturbations averaged between September 2015 and March 2016, corresponding to the peak of the 2015–2016 El Niño. Figure 1B shows the percentage of the number of simulated Embedded Image perturbations larger than 1.0 ppm at each 4° × 5° grid (the resolution of the transport model), and Fig. 1C is a histogram distribution of Embedded Image perturbations.

Fig. 1

Embedded Image perturbations due to net biosphere flux anomaly during the peak of 2015–2016 El Niño. (A) Mean Embedded Image perturbations over the period September 2015–March 2016. Embedded Image was sampled with OCO-2 sensitivity and locations of OCO-2 observations. (B) Percentage of Embedded Image perturbations larger than 1 ppm at each 4° × 5° grid. (C) Histogram of Embedded Image perturbations.

Figure 1A shows that the Embedded Image anomaly was typically much larger than 0.5 ppm during the peak of the 2015–2016 El Niño. About 90% and 70% of samples have enhancements larger than 0.5 ppm and 1.0 ppm, respectively (Fig. 1C), well above the threshold for OCO-2 XCO2 systematic error. More than 60% of the samples at each 4° × 5° grid have perturbations larger than 1.0 ppm (Fig. 1B). This is in stark contrast to Chevallier’s claim that “still only 1.5% of the sounding perturbations exceed 0.5 ppm in that case.” Aside from using different transport models, the Embedded Image differences between the two experiments are due to the spatiotemporal distributions of the flux anomaly: concentrated versus uniformly distributed fluxes.

The perturbed Embedded Image signal over each of the three tropical continents arises primarily from flux anomalies in the respective continent [figure S1 of (1)]. Consistent with uncertainty quantification (1) and fundamental source-receptor relationships (3), this analysis supports that our inversion system can discriminate between the anomalies in the three tropical continents, enabling the main conclusion in (1) that diverse biogeochemical processes are responsible for the large NBE anomaly over three tropical continents, which was also supported by two additional types of observations: carbon monoxide and solar-induced chlorophyll fluorescence used to constrain biomass burning and plant primary productivity, respectively.

The 2015–2016 El Niño was an extreme climate event, generating intense but localized climate and NBE anomalies (1, 4) that were exploited by the inversion system in (1). Systematic errors in Embedded Image retrievals are one of the many factors contributing to the overall estimated net biosphere flux uncertainty. Quantification and mitigation of these errors is an active area of research and is likely to lead to improved data products and analyses of subtler phenomena, such as downwind enhancements from megacity fossil fuel emissions. Substantial advances have been demonstrated by the OCO-2 and GOSAT communities, and we expect continued progress toward the capability of quantifying fluxes by the emerging international constellation of satellites with a much higher confidence level than now. We agree with Chevallier that this advancement must remain a priority (5).

References and Notes

Acknowledgments: This research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA. This work was in part supported by NASA Carbon Monitoring System program grant 14-CMS14-0054 and by NASA Orbiting Carbon Observatory Science team program grants 14-OCO2_14-0007 and 11-OCO211-0024. K.R.G. was supported by NSF CAREER award 0846358.
View Abstract

Navigate This Article