EMEP Task Force on Measurements and Modelling TFMM

EMEP Task Force on Measurements and Modelling (TFMM) RETROSPECTIVE AIR QUALITY SIMULATION OVER NORTHERN ITALY C. Silibello(1), S. Finardi(1), G. Calori(1), P. Radice(1), G. Brusasca(1) (1)ARIANET Srl, Via Gilino 9, 20128 Milano, Italy 17 th Task Force on Measurement and Modelling Meeting: Utrecht - May 18 th to May 20 th, 2016 TNO, Princetonlaan 6, 3584 CB Utrech

Rationale Traditional air pollution epidemiology studies being conducted in large cities are based on ambient monitoring data from an outdoor monitoring station (major assumption: these ambient measurements are representative of the air quality over a broad area). Air quality modelling is a preferred approach to improve spatial/temporal resolution of air pollutant concentrations to be used by exposure models. Purpose The Maggiore Policlinico Hospital, Milan (Epidemiology Unit) estimated PM 10 levels (2000 -2009) using satellite data and asked to ARIANET to reconstruct air quality conditions in Lombardy Region and Milano urban area for the year 1990 to support epidemiological investigations concerning lung cancer.

The IDEA Mean PM 10 (1990) ? Mean PM 10 (2000 -2009) Trend analysis ’ 91 ’ 92 … ’ 99

OUTLINE q air quality trends over Milano urban area q 1990 Air quality modelling: ü meteorological fields (WRF); ü emissions (National Inventory + MACCity data); ü concentration fields (FARM) q Major findings

Trend analysis MSCE trend tool version 1 SIGNAL = NOISE + SEASONALITY + TREND variations not described by the two major components. subject to strong seasonal variations decrease/increase during the considered period Markus Wallasch (UBA, DE) , Dos & don'ts in trends analysis, 15 th annual 1 - INEMAR EMEP TFMM meeting Bologna, Italy, 8 - 10 April 2014

Milano – Juvara SPM & PM 10 SPM 1985 -2007 co n me vers tha ion ne to PM 10 SO 2 ind de ustr loc y ali sa ti on NO 2 catalytic converter introduction (1991) O 3 Diesel vehicles Increment (2002)

Milano – Juvara Seasonal variation SPM & PM 10 SO 2 NO 2 O 3

Milano - Overall average (1981 -2004) NO 2 Replacement of older vehicles with less polluting vehicles Economic crisis drastic diminution of automotive market Tebaldi et al. (2005) RAPPORTO SULLA QUALITA’ DELL’ARIA DI MILANO E PROVINCIA UNRAE – L’auto 2013. Sintesi statistica.

Milano - Overall average SO 2 In 1989, the Mayor of Milano issued an urgent decree prescribing the reconversion of domestic heating systems to methane or diesel with a lower content of sulphur (from 3% to 1% of sulphur) Tebaldi et al. (2005) RAPPORTO SULLA QUALITA’ DELL’ARIA DI MILANO E PROVINCIA

Air quality modelling Target domain (3 km x 3 km) 1 - INEMAR 9 km x 9 km

Atmospheric system Meteorological Emissions treatment sub-system ERA-Interim analyses Geo data Chemical-trasport ISPRA MACCity Space, time, speciation WRF Emission Manager SURFPRO sub-system 1990 Emission scenario 1990 meteo year BCs FARM Conc & dep. fields Post-processing sub-system Excel Air quality data Visualiser NCO, CDO

STEM 80’s Comprehensive regional model Regional applications (Asia, N & Latin America, Europe. . ) Chemical mechanisms 90’s Flexible Air quality Regional Model (FARM) a long history…now available at: https: //hpc-forge. cineca. it/ Numerical aspects Aerosols FARM Biomass burnings Regional vs. global chemistry Chemical forecast 00’s Data assimilation . . . System integration, multi-domain Chemical mechanisms, online photolysis rates calculation (TUV) 1 st Quale. Aria AQFS development Data assimilation 10 EU scale applications 13 Full parallel version 1 st “Open. Source” release

Boundary conditions from SILAM model SILAM 1990 concentration fields computed by SILAM (part of 35 -years-long reanalysis 1980 -2014). 0. 5° x 0. 5° Target domain (3 km x 3 km) 9 km x 9 km

Emission inventories Italian national inventory province-based Surrounding areas: MACCity (0. 5°) … disaggregated on Lombardy municipalities using representative indicator (proxies) Proxies have to be: • Consistent: proximity/coincidence with indicator used to estimate provincial • Uniform: available throughout the country • Continuous: to replicate consistently the methodology for future years

… point sources 1990 ISPRA database did not distinguish point and diffuse sources! After a complex information collection activity, about 90 industrial plants have been identified, with a total of 159 chimneys. Following sources have been identified: power plants, refineries, steel mills and waste treatment plants. Total NOX emissions at municipal level NOX emissions at a municipal level distinguished in point and diffuse sources

Emission input on model grids Example of composite emissions on N Italy (9 km res. ) and Lombardia (3 km res. ) simulation grids NO, near-ground, 8 AM of January working day

Significant overestimation of SO 2 concentrations Ø In 1989, the Mayor of Milano issued an urgent decree prescribing the reconversion of domestic heating systems to methane or diesel with a lower content of sulphur (from 3% to 1% of sulphur); Ø The national inventory probably not took it into account leading to a significant overestimation of SO 2 emissions; Ø Reducing by a factor 3 the Milan Province SO 2 emissions from domestic heating, coherently with the decree, a better agreement with observation was obtained. Original Inventory (O. I. ) Reduced Inventory (R. I. )

Air Quality network ARPA Lombardia EMEP - JRC MILANO SPM ARPA Lombardia # Station 503 Milano - Zavattari 502 Milano - Juvara 501 Milano - Marche 528 Milano - Verziere 504 Sesto San Giovanni 529 Cinisello Balsamo 525 Corsico 531 Limito 510 Inzago 508 Cassano d`Adda 610 Groppello d`Adda 509 Rivolta d` Adda 609 Casirate d'Adda 530 Agrate Brianza 507 Villasanta 505 Legnano 527 Castano Primo 532 Cuggiono 517 Turbigo 515 Pero 514 Rho 516 Robecchetto 582 Bergamo - via S. Giorgio 587 Casnigo 649 Brescia – Broletto 656 Gambara 657 Lonato del Garda 660 Ospitaletto 662 Vobarno

SO 2 EMEP - JRC

NO 2 EMEP - JRC

O 3

PM 10 Measurements SPM*0. 8 -> PM 10

PM 10 EMEP - JRC

EMEP – JRC Inorganic ions

DJF Seasonality MAM PM 10 JJA SON

Emissions and concentrations trends (Lombardy) PM 10 yearly averaged concentrations (2005 and 2010 from MINNI: www. minni. org) 2005 mg m-3 1990 PM 10 em. 1995 2010 2000 PM 10 2005 2010 ton y-1 1990

Data fusion (Lombardy) PM 10 Average concentrations (year) Raw model + obs > 50% availability The integration of dispersion models and observations (Data fusion techniques) permits to obtain more realistic air quality maps and to better estimate the exposure of population

Data fusion (1990) vs AOD (2000 -2009) PM 10 Average concentrations Mean PM 10 concentration in each 3 x 3 km grid during 1990 predicted by the “Data fusion” methods Mean PM 10 concentration in each 10 x 10 km grid during the entire modeling period (2000 -2009) predicted by the AOD models

Major findings q Good reproduction of concentration levels at various monitoring sites within Milan urban area q Good reproduction of inorganic ions at EMEPJRC site q Modelled concentration trends coherent with observations ü 1990 a. PM 10 concentration fields will be further processed for following epidemiological studies

After Querol presentation …. Ozone nocturnal peak

Acknowledgments This work is supported by the Epidemiology Unit, Department of Preventive Medicine, IRCCS Ca’ Granda Foundation, Maggiore Policlinico Hospital, Milan, Italy Special thanks to: ü Mikhail Sofiev and co-workers for providing 1990 concentration fields computed by SILAM ü ECCAD project and ISPRA for providing emission datasets ü Edoardo Peroni (ARPA-Lombardia) and Chemical Coordinating Centre (CCC) of EMEP for providing air quality data ü ENEA for providing MINNI Project 2005 and 2010 maps Thank you for your attention !!!

EMEP – JRC SPM PM 2. 5

WRF AWR V 3. 5. 1 Downscaling of global meteorological analyses to the local scale Spectral Nudging: to selectively nudge the coarser scales only (u, v, theta and geop. height) Applied to the outer domains to maintain mesoscale reanalyses fields features without suppressing high resolution meteorological variability Grid spacing: 36, 12, 4 km Vertical grid: 41 levs (up to 50 h. Pa) BCs: ERA-Interim analyses (0. 75°x 0. 75°) on pressure levels Spin-up: 10 days Land Cover: CORINE 2006 1 - INEMAR

Temperature at 850 h. Pa Comparison with ERA-Interim fields to check the conservation of the analyses fields at continental scale (monthly average fields on pressure levels) January 1990 March 1990 ERA-Interim WRF

Juvara street (Milan) Temperature 1990

NO 2 and O 3 averaged concentrations NO 2 Year O 3 JJA

Grid resolution NO 2 Average concentrations (DJF) SILAM 0. 5° x 0. 5° FARM 9 km x 9 km FARM 3 km x 3 km