Global lightning NO_x production estimated by an assimilation of multiple satellite data sets

Abstract. The global source of lightning-produced NO_x (LNO_x) is estimated by assimilating observations of NO₂, O₃, HNO₃, and CO measured by multiple satellite measurements into a chemical transport model. Included are observations from the Ozone Monitoring Instrument (OMI), Microwave Limb Sounder (MLS), Tropospheric Emission Spectrometer (TES), and Measurements of Pollution in the Troposphere (MOPITT) instruments. The assimilation of multiple chemical data sets with different vertical sensitivity profiles provides comprehensive constraints on the global LNO_x source while improving the representations of the entire chemical system affecting atmospheric NO_x, including surface emissions and inflows from the stratosphere. The annual global LNO_x source amount and NO production efficiency are estimated at 6.3 Tg N yr⁻¹ and 310 mol NO flash⁻¹, respectively. Sensitivity studies with perturbed satellite data sets, model and data assimilation settings lead to an error estimate of about 1.4 Tg N yr⁻¹ on this global LNO_x source. These estimates are significantly different from those estimated from a parameter inversion that optimizes only the LNO_x source from NO₂ observations alone, which may lead to an overestimate of the source adjustment. The total LNO_x source is predominantly corrected by the assimilation of OMI NO₂ observations, while TES and MLS observations add important constraints on the vertical source profile. The results indicate that the widely used lightning parameterization based on the C-shape assumption underestimates the source in the upper troposphere and overestimates the peak source height by up to about 1 km over land and the tropical western Pacific. Adjustments are larger over ocean than over land, suggesting that the cloud height dependence is too weak over the ocean in the Price and Rind (1992) approach. The significantly improved agreement between the analyzed ozone fields and independent observations gives confidence in the performance of the LNO_x source estimation.