Sensitivity analysis of methane emissions derived from SCIAMACHY observations through inverse modelling

Satellite observations of trace gases in the atmosphere offer a promising method for global verification of emissions and improvement of global emission inventories. Here, an inverse modelling approach based on fourdimensional variational (4D-var) data assimilation is presented and applied to synthetic measurements of atmospheric methane. In this approach, emissions and initial concentrations are optimised simultaneously, thus allowing inversions to be carried out on time scales of weeks to months, short compared with the lifetime of methane. Observing System Simulation Experiments (OSSEs) have been performed to demonstrate the feasibility of the method and to investigate the utility of SCIAMACHY observations for methane source estimation. The impact of a number of parameters on the error in the methane emission field retrieved has been analysed. These parameters include the measurement error, the error introduced by the presence of clouds, and the spatial resolution of the emission field. It is shown that 4D-var is an efficient method to deal with large amounts of satellite data and to retrieve emissions at high resolution. Some important conclusions regarding the SCIAMACHY measurements can be drawn. (i) The observations at their estimated precision of 1.5 to 2% will contribute considerably to uncertainty reduction in monthly, subcontinental (∼500 km) methane source strengths. (ii) Systematic measurement errors well below 1% have a dramatic impact on the quality of the derived emission fields. Hence, every effort should be made to identify and remove such systematic errors. (iii) It is essential to take partly cloudy pixels into account in order to achieve sufficient spatial coverage. (iv) The uncertainty in measured cloud parameters may at some point become the limiting factor for methane emission retrieval, rather than the uncertainty in measured methane itself. Correspondence to: J. F. Meirink ([email protected])