Constraining the exchange of carbon dioxide over the Amazon: New insights from stable isotopes, remote sensing and inverse modeling

In this thesis we study the exchange of CO₂ between the atmosphere and biosphere, with a focus on the Amazon region. The exchange of CO₂ between atmosphere and biosphere occurs through photosynthetic uptake, which we usually refer to as gross primary production (GPP), and through respiratory release, usually referred to as terrestrial ecosystem respiration (TER). The sum of GPP and TER is the net ecosystem exchange (NEE). The focus of the thesis is on quantifying the exchange of CO₂ between biosphere and atmosphere, and the control of environmental variables on this exchange. In particular, the effect of drought on CO₂ exchange, and the post-drought recovery, are major themes of this thesis. 

Chapter 2 describes the 3-D model that we developed for Δ¹⁷O in atmospheric CO₂ (a potential tracer for GPP). We implemented this tracer in the global atmospheric transport model TM5, which is driven by ERA-Interim meteorological fields. We parameterized the stratospheric source of Δ¹⁷O in CO₂ by exploiting its observed correlation with stratospheric [N₂O]. The exchange between the atmosphere and biosphere is simulated using the terrestrial biosphere model SiBCASA. We also included the contributions of soils, oceans, biomass burning, fossil fuel combustion and the oxidation of atmospheric CO on Δ¹⁷O in CO₂. For CO₂ in the lowest 500 m of the atmosphere, we simulated a Δ¹⁷O signature of 39.6 per meg, which is ~20 per meg lower than estimates from existing box models. Our model results show good agreement with a measured stratospheric Δ¹⁷O in CO₂ profile from Sodankylä (Finland). In addition, we compared model simulations with tropospheric measurements of Δ¹⁷O in CO₂ from Göttingen (Germany) and Taipei (Taiwan), where we found some agreement but also substantial discrepancies. Finally, we showed model results for Zotino (Russia), Mauna Loa (United States), Manaus (Brazil) and South Pole, which we proposed as possible locations for future measurements of Δ¹⁷O in tropospheric CO₂ that can help to further increase our understanding of the global budget of Δ¹⁷O in atmospheric CO₂.  

In Chapter 3 we studied Δ¹⁷O at the ecosystem level, which is the domain that integrates the contributions from vegetation and soil to the atmospheric signal. We reported for the first time an observed diurnal cycle of Δ¹⁷O in CO₂, measured from air samples collected on 15-16 August 2019 at the mid-latitude pine forest Loobos (FLUXNET site NL-Loo). Most notable is an observed peak for Δ¹⁷O in CO₂ in the morning (~140 per meg, around 6:30 am) that we tentatively ascribed to the entrainment of residual air masses. Besides Δ¹⁷O in CO₂, we report observations of δ¹³C and δ¹⁸O in CO₂ for flasks collected close to the surface (at 0.5 m height, inside the canopy) and from the top of the tower (1-2 m above the canopy). To support the interpretation of observations, we simulated δ¹³C, δ¹⁸O and Δ¹⁷O in CO₂ in the atmospheric boundary layer (ABL) during daytime for Loobos using the mixed layer model MXL. Finally, we used the global atmospheric transport model TM5 to (1) quantify the contribution of different sources that affect Δ¹⁷O in CO₂ at Loobos using a `tracer tagging' method; and (2) extend our analysis of the diurnal cycle to the global scale. Based on these simulations, we expect the largest diurnal cycles for Δ¹⁷O in CO₂ in tropical ecosystems, and we propose to follow-up on this study by sampling air from well-equipped tropical sites, such as the ATTO tower in the Amazon

In Chapter 4, we study the response of the Amazon forest to the 2015/2016 El Niño, using sun-induced fluorescence (SIF) a proxy for GPP. We developed a new remotely-sensed SIF product with improved signal-to-noise in the tropics, and found that SIF was strongly suppressed over areas with anomalously high temperatures and decreased levels of water in the soil. SIF went below its climatological range starting from the end of the 2015 dry season (October) and returned to normal levels by February 2016 when atmospheric conditions returned to normal, but well before the end of anomalously low precipitation which persisted through June 2016. Impacts were not uniform across the Amazon basin, with the eastern part experiencing much larger (10-15%) SIF reductions than the western part of the basin (2-5%). We estimated that the integrated loss of GPP relative to eight previous years is 0.34-0.48 PgC in the 3-month period Oct-Nov-Dec 2015. 

Finally, in Chapter 5 we consider the response of the Amazon to droughts in the period 2010-2017. We used CO₂ profiles collected by aircraft over the Amazon to quantify the net ecosystem exchange for drought and post-drought years using the CarbonTracker South America (CTSAM) inverse modelling system. In addition, we performed inversions using independent CO₂ column observations from OCO-2 to quantify CO₂ exchange for the years 2015-2017. We estimate that the total Amazon CO₂ emissions for the years 2010 and 2016 are 0.3-0.5 PgC and 0.0-0.3 PgC larger than the subsequent years 2011 and 2017, respectively. Furthermore, we used near-infrared reflectance of terrestrial vegetation (NIRv), from MODIS and MAIAC, to diagnose the direct and delayed response of GPP to the 2015/2016 drought. We find a substantial reduction of NIRv in the 2016 dry season, when precipitation and temperature have returned to normal values, while soil moisture is still anomalously low, suggesting a persistent impact of the preceding 2015/2016 El Niño drought.  

This work contributed to a better understanding of the exchange of CO₂ between the biosphere and atmosphere. We have investigated variations of Δ¹⁷O in atmospheric CO₂ and paved the way for new measurements campaigns and the actual application of Δ¹⁷O in CO₂ as a tracer for GPP. We have demonstrated that SIF and NIRv are powerful methods to quantify reductions of GPP during drought and track its post-drought recovery. Finally, our analysis of atmospheric CO₂ inversions showed how the interannual variability of the net exchange of CO₂ over the Amazon is controlled by variations in environmental drivers. Although important challenges remain, the work described in this thesis contributes to better understanding of carbon dioxide exchange over the Amazon.