Analysis of the March-April 2020 - COVID situation in

Michael Schulz , Augustin Mortier, Svetlana Tsyro, Anna Benedictow, Hilde Fagerli and CAMS71 partners INERIS, NILU, TNO MET published as part of CAMS71/50 COVID episode report, July 6, 2020 see CAMS website

Atmosphere Monitoring Tools and methods

Atmosphere Monitoring EMEP model (~14 km, ECMWF meteorology): ❏ CAMS-50 operational setup - two runs: ● Reference EMEP-CAMS ● Emission reductions for EU+EEA countries (Barcelona Computing Center): Road transportation (54%), Industry (16%), Aviation (94%) EMEP-CAMS-covid ❏ Source-receptor runs for major cities in Europe (CAMS-71) EMEP-SR-Ref2 ❏ NO2 and PM10 ❏ Period: 1 March - 30 April 2020 ❏ Urban EMEP bias removed (-28% PM10 and -63% NO2) Observations: air quality observations reported to the EEA AQ e-reporting database (excluding traffic sites) Hypothesis and testing Atmosphere Monitoring ● Transport model captures concentration variation due to “meteorology” ● Emission reduction model run should show reduced NO2 and PM10 ● Observations should show lower concentrations after lockdown, unless meteorology variations impose different concentration evolution ● Observations to model bias should shift after lockdown

● Comparison in city areas where observations were available for report ● Daily observations and model output is averaged in 60x60km city area in period before and after country lockdown Observed and best guess modelled before and after lockdown Atmosphere Monitoring

Model run WITH emission reductions used after lockdown Analysis March-April Model versus Observation / Source Receptor split

Atmosphere Monitoring NO2 Barcelona Lockdown => Analysis March-April Model versus Observation / Source Receptor split

Atmosphere Lockdown => Monitoring NO2 London Analysis March-April Model versus Observation / Source Receptor split

Atmosphere Lockdown => Monitoring PM10 Barcelona Analysis March-April Model versus Observation / Source Receptor split

Atmosphere Lockdown => Monitoring PM10 London Analysis March-April Model versus Observation / Source Receptor split

Atmosphere Lockdown => Monitoring PM10 Oslo Analysis March-April Model versus Observation / Source Receptor split

Atmosphere Lockdown => Monitoring PM10 Sofia

Aralkum dust episode Observed and best guess modelled before and after lockdown Atmosphere Monitoring

Model run WITH emission reductions used after lockdown NO2 PM10 Concentration change (%) Concentration change (%) Lockdown Cities grouped City wrt to before lockdown start Local wrt to before lockdown start Local contribution (%), Date*) contribu- major country Model Model Observed Model tion (%) Observed Model contributions Reference Reference 19% Barcelona 16.03 -49 -51 -16 42 -23 1 16 City areas with visible PM10 EPS, IAT, FAR, DUE pollution decrease due to 27% Madrid 16.03 -59 -61 -1 24 -30 -19 -4 lockdown emission reductions Natural dust

13% City areas with PM increase (or London 24.03 -14 4 13 14 72 51 69 10 DEU, FRA, NLD, BEL, POL slight decrease) after lockdown due 19% to dominating effect of emission Paris 17.03 -33 -29 22 24 38 32 61 advections from anthropogenic rest FRA, DEU,BEL, GBR 22% sources (e.g. agricultural, Milan 10.03 _ -62 21 39 -13 3 38 industrial) over COVID related rest ITA 11% reductions Berlin 22.03 -27 -47 -28 19 26 -12 -5 rest DEU, POL; salt 21% City areas with only small decrease Oslo 12.03 -27 -55 -47 46 84 -12 -9 road dust (in observations) of PM10 levels due to considerable effect of road dust due to studded 24% Stockholm 15.03 -29 -43 33 56 4 -23 -18 tyres, road dusting and salting road dust (in observations) 17% Sofia 15.03 -27 -51 15 80 37 23 29 Aralkum desert dust City areas with mixed episode**) decrease/increase of the average 16% Budapest 28.03 -22 -42 21 39 2 -25 -20 Aralkum desert dust PM10 concentrations as the effect episode**) of lockdown emission reductions 30% was disturbed by a large natural Warsaw 25.03 0 -50 31 71 24 -31 -27 Model does not reproduce dust episode on 26-29 March observed episodes Summary Atmosphere Monitoring A preliminary analysis of the impact of emission drop caused by reduced activity due to COVID-19 lockdown on air pollution in European cities was performed within CAMS71 framework: EMEP model simulations with/without emission reductions & EEA AQ observations Compared to Reference run, the model run with Covid emission reductions

simulates larger decrease of NO2 and smaller increase in PM10 during lockdown period

NO2 concentrations decreased during the lockdown period compared to pre-COVID situation by 29% according to observations and by 46% according to the model (@measurement sites averaged for 10 cities)

The overall effect of the lockdown on PM10 is less obvious: both concentration decreases and increases were observed and modelled SR analysis, used for interpretation of these results, identified significant differences in air pollution patterns and dominating PM sources among the cities investigated: “other” sources and long range transport have masked the COVID reductions