Estimating the Trend, Seasonality, and Variability of Global CO2 Fluxes with a Bayesian Flux-Inversion Framework

• Activity Number: 394 - Spatial and Spatio-Temporal Modeling in Climate and Meteorology
• Type: Contributed
• Date/Time: Wednesday, August 10, 2022 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Michael Bertolacci* and Andrew Zammit Mangion and Noel Cressie and Andrew Schuh and Beata Bukosa and Jenny Fisher and Yi Cao
• Companies: University of Wollongong and University of Wollongong and University of Wollongong and Colorado State University and NIWA and University of Wollongong and University of Wollongong
• Keywords: spatiotemporal; carbon cycle; inversion; flux; Bayesian hierarchical model; satellite data

Abstract

The well-quantified anthropogenic emissions of carbon dioxide (CO2) are the leading cause of climate change. However, more than half of the emitted CO2 is later absorbed by the biosphere and the ocean. These natural sinks of CO2 are difficult to quantify because they arise as imbalances in the natural source and sink processes of the carbon cycle. Furthermore, climate change and land use change are expected to cause changes in the carbon cycle itself. Quantifying the natural sources and sinks (known as fluxes) of CO2 is therefore needed to better understand climate change.

We address this problem through the Wollongong Methodology for Bayesian Assimilation of Trace-gases (WOMBAT, Zammit-Mangion et al., Geosci. Model Dev., 15, 2022), a fully Bayesian hierarchical framework for estimating trace-gas fluxes using data on atmospheric mole fractions. We extend WOMBAT to estimate spatially-varying time-series decompositions of the fluxes. This separates the fluxes into trend, seasonal, and residual components. The method is applied to both satellite and in situ/flask data over a six-year period. The resulting estimates suggest that the carbon cycle is indeed changing in some regions.