Order of magnitude wall time improvement of variational methane inversions by physical parallelization: a demonstration using TM5-4DVAR

Abstract. Atmospheric inversions are used to constrain emissions of trace gases using atmospheric mole-fraction measurements. The four-dimensional variational (4DVAR) inversion approach allows optimization at a higher temporal and spatial resolution than ensemble or analytical approaches but provides limited opportunities for scalable parallelization because it is an iterative method. Multidecadal needed optimally extract information from the long measurement records long-lived like carbon dioxide methane. However, wall time – up months complicates these multidecadal inversions. physical (PP) method introduced by Chevallier (2013) addresses this problem splitting period chemical transport model into blocks running them in parallel. Here, we present new implementation PP which suitable methane accounting sink performance tested 11-year TM5-4DVAR setup that assimilates surface observations optimize grid scale. Our improves factor 5 shows excellent agreement with full serial identical configuration (global mean difference =0.06 % interannual variation correlation R=0.99; regional emission <5 R>0.94). improvement increases size period. planned be future releases Copernicus Atmosphere Monitoring Service (CAMS) reanalysis.