A three-dimensional synthesis inversion of the molecular hydrogen cycle: Sources and sinks budget and implications for the soil uptake

Our understanding of the global budget of atmospheric hydrogen (H-2) contains large uncertainties. An atmospheric Bayesian inversion of H-2 sources and sinks is presented for the period 1991-2004, based on a two networks of flask measurement stations. The types of fluxes and the spatial scales potentially resolvable by the inversion are first estimated from an analysis of the correlations of errors between the different processes and regions emitting or absorbing H-2. Then, the estimated budget of H-2 and its uncertainties is presented and discussed, for five groups of fluxes and three groups of large regions, in terms of mean fluxes, seasonal and interannual variations, and long-term trends. One main focus of the study is the improvement of the estimate of H-2 soil uptake, which is the largest sink of H-2. Various sensitivity tests are performed defining an ensemble of more than 20 inversions. We show that inferring a robust estimate of the H-2 soil uptake requires to prescribe the prior magnitude of some other sources and sinks with a small uncertainty. Doing so an estimate of the H-2 soil uptake of -62 +/- 3 Tg y(-1) is inferred for the period 1991-2004 (the uncertainty is the residual error after inversion). The inferred soil H-2 sink presents a negative long-term trend that is qualitatively consistent with a bottom-up process-based model.