Regional Carbon Sources and Sinks Derived from Over a Decade of GOSAT and OCO-2 CO2 Measurements

Abstract:

Satellite CO₂ measurements provide a dense, global constraint on regional carbon sources and sinks, particularly in the tropics and south where in situ measurements are sparse. CO₂ fluxes from the tropical land biosphere appear to dominate the interannual variability of global atmospheric CO₂– understanding the key processes involved is crucial for predicting future CO₂ levels and climate. While passive satellite instruments have difficulty seeing through the clouds prevalent over tropical forests, full-column measurements taken over the surrounding oceans can provide an adequate constraint there in top-down inversions.

Satellite CO₂ measurements are susceptible to systematic errors that corrupt flux estimates derived from them. Because most validation data are located over or around the continents, satellite data over oceans are particularly affected. However, the sheer volume of the ocean glint data, as well as its inherently low noisiness, make it critical that we understand and remove these biases. To help do so, we compare CO₂ fluxes from inversions of two new datasets: 11+ years of version 9 ACOS GOSAT data and 6 years of v10 OCO-2 data. The new (v10) release of OCO-2 data has fixed a negative ocean glint mode bias, yielding a more reasonable global land/ocean flux partition. The v10 OCO-2 results differ markedly from the v9 OCO-2 and GOSAT results in the aftermath of the 2015/16 El Niño: the land-based release of CO₂ during the warm phase finishes by mid-2016 as the following La Niña commences, followed by an ocean outgassing peaking in mid- to late-2016, in contrast to a longer-lasting land release in the v9 results with no clear ocean signal. All are in agreement that the tropical land regions are responsible for the release of CO₂ seen globally during the 2015/16 El Niño; however, the GOSAT record shows that this was not much larger in magnitude than similar tropical releases in 2010 and 2012/13, calling into question the driving processes involved. The v10 OCO-2 data yields a more plausible land/ocean partition in the extra-tropical NH than v9, though like v9 it yields too much CO₂ uptake in the tropical oceans. Overall, though, it appears that the systematic errors have been reduced enough that satellite-derived regional to global carbon budgets should now be considered alongside traditional in situ-based ones.