The coronavirus outbreak in 2020 had a devastating impact on human life, albeit a positive effect for the environment, reducing primary atmospheric constituents and improving air quality. Here we present, for the first time, inverse modelling estimates of ammonia emissions during the European lockdowns of 2020 based on satellite observations. Ammonia has a strong seasonal cycle; it mainly originates from agriculture, which was influenced insignificantly by the lockdowns, as practically agricultural activity never ceased. The key result is a -0.7% decrease in emissions in the first half of 2020 compared to the same period in 2016–2019 attributed to restrictions related to the global pandemic or an abrupt -9.8% decrease due to reductions in the traffic-related precursors of atmospheric acids, with which ammonia reacts to form secondary aerosols. When comparing emissions before, during and after lockdowns, the typical seasonal trends of ammonia prevail. However, when reductions in the precursors of atmospheric acids are considered, a delay of 11% was found in the evolution of the emissions. Thus, changes in atmospheric conditions such as those of the ammonia’s reactant precursor species induce extra bias in top-down calculations and, hence, emissions should be interpreted carefully. Despite the small drop in emissions, satellite levels of ammonia increased. On one hand, this was due to the reduction of atmospheric acids that caused binding and thus removing less ammonia; on the other, the reduction of traffic-related emissions in Europe increased the oxidative capacity of the atmosphere resulting in nitrate abatement that favored accumulation of free ammonia.

Update March 2023:

- 4deg_avgEENV.tar.gz file was added containing the inversion results using the avgEENV dataset as a priori information. This prior creates a better fit of the posterior modelled concentrations to ground-based independent observations of NH3 over Europe in the first half of 2020.

- FLEXPART v10.4 sensitivities for 6 model levels using chemistry loss after taking into account changes in emissions of NOx and SO2 in the calculations (see also associated paper): AVE_FOOTPRINT_L1_chem.nc, AVE_FOOTPRINT_L2_chem.nc, AVE_FOOTPRINT_L3_chem.nc, AVE_FOOTPRINT_L4_chem.nc, AVE_FOOTPRINT_L5_chem.nc, AVE_FOOTPRINT_L6_chem.nc

- FLEXPART v10.4 sensitivities for 6 model levels using chemistry loss after taking into account changes in emissions of NOx and SO2 in the calculations (see also associated paper): AVE_FOOTPRINT_L1_nochem.nc, AVE_FOOTPRINT_L2_nochem.nc, AVE_FOOTPRINT_L3_nochem.nc, AVE_FOOTPRINT_L4_nochem.nc, AVE_FOOTPRINT_L5_nochem.nc, AVE_FOOTPRINT_L6_nochem.nc

- Different surrounding estimates of posterior emissions covering 2°, 3° and 4° from each grid-cell: 2deg.tar.gz, 3deg.tar.gz, 4deg.tar.gz

- Performance calculations for the prior and posterior emissions against the dependent observations (CrIS): Correction_R4.tar.gz

- Forward simulations with FLEXPART v10.4 using prior and posterior emissions that were used for validation against independent (observations that were excluded from the inversion) observations from the EMEP network: FWD_pri.tar.gz, FWD_pos.tar.gz