The Detectability of Extreme CO ₂ Flux Events Using Atmospheric Observations

Abstract

Climate events like droughts and floods perturb the carbon cycle and lead to extremes in CO ₂ fluxes (i.e., net biosphere exchange or NBE). These extreme NBE events are important because they contribute to inter‐annual variability in the global carbon cycle and may change in the future as climate changes. We develop synthetic experiments to evaluate how well atmospheric CO ₂ observations can detect and quantify extreme NBE events across the globe. We focus on observations from the Orbiting Carbon Observatory 2 (OCO‐2) and in situ atmospheric CO ₂ observations, and we focus on NBE extremes that are in at least the fifth percentile globally, as simulated by a commonly‐used terrestrial biosphere model. We find that OCO‐2 data have enabled much better detection of extreme events globally compared to in situ data alone, and we can accurately identify up to 10%–30% of all model grid boxes that contain NBE extremes, depending on the region. However, our inverse modeling simulations tend to underestimate the magnitude of these extremes. We have greater success detecting extreme NBE events associated with anomalously hot and/or dry climate conditions compared to cold and/or wet climate conditions, and our success rate is also several times higher in global regions with high OCO‐2 data density compared to regions with low data density. The results also vary widely with the inverse modeling setup. Overall, these results emphasize the salient role of data availability in detecting NBE extremes and highlight the potential of future planned missions like CO2M for improving event detection.