Bayesian inversion for reconciling uncertainties in global mass balances



This paper presents a Bayesian method for reconciling uncertainties in individual sources and sinks in global mass balance models and applies it to global cycle of methane. Ranges of values derived from in-situ flux measurements are used to define prior probability distributions for the individual sources and sinks. Atmospheric concentrations of carbon isotopes (14C, 13C and 12C) and ice core measurements (12C) provide additional information regarding the sources and sinks. Bayes Monte Carlo simulation is used to derive a posterior range for sources that combines data from field measurements, atmospheric observations and ice core data. It is shown that careful interpretation and analysis of available data can result in better resolution of source uncertainties. Emissions of methane from rice paddies and wetlands may be smaller than assessed in the past.