Overview of current evaluations, applications and advances in global air quality models

Daven K. Henze (CU Boulder) Arlene Fiore (Columbia University) and many others…

Support from NASA HAQAST, HTAP Outline

1. Global and regional emissions estimates: past and future

2. Multi-model intercomparisons and evaluations

3. Applications to regional air quality

4. Advances in global modeling Uncertainty in U.S. air pollutant emissions and trends: 1980-2010 ACCMIP emissions: Lamarque et al., ACP, 2010 used in CMIP5 & ACCMIP BC CO

NOx SO2

Granier et al., Climatic Change, 2011 Future air pollutant emissions – the role for AQ policies

160 180 8 SO2 NOx BC 140 160 7

140 120 6

120 100 5 100 80 4 80 Million tons Million tons Million Million tons 60 3 RCP 60 RCP RCP

40 2 40 GAINS CLE GAINS CLE GAINS CLE

20 GAINS NFC 20 GAINS NFC 1 GAINS NFC GAINS MTFR GAINS MTFR GAINS MTFR 0 0 0 1990 2000 2010 2020 2030 2040 2050 1990 2000 2010 2020 2030 2040 2050 1990 2000 2010 2020 2030 2040 2050

Source: GAINS model; ECLIPSE V5 scenario c/o T. Keating AQAST10 Jan 2016 Latest global 21stC scenarios: “Shared Socioeconomic Pathways” (SSPs)

AerChemMIP is using SSP3-7.0 (“regional rivalry without policy”) as reference scenario against which pollution controls are imposed

Fig. 5 from Riahi et al., Global Environmental Change, 2017 AerChemMIP chemistry-climate simulations include pollution control policies

Table 3 from Collins et al., GMD, 2017 (based on Rao et al., 2017 Table 2) 2. Global model intercomparison activities and evaluations Chemistry-climate models are typically evaluated with ozonesonde climatologies

ACCMIP models [Young et al. ACP 2013] (ACCMIP goal is to quantify radiative forcing) Chemistry-climate models are typically evaluated with ozonesonde climatologies

ACCMIP models: Normalized biases for ACCMIP against ozonsondes by region [Young et al. ACP 2013]

(ACCMIP models also evaluated against satellite OMI/MLS tropospheric product) AeroCom

“is an international science initiative on aerosols and climate” -- coordinate simulations -- evaluate aerosol distributions with satellite and surface observations -- quantify radiative forcing and uncertainties in these estimates http://aerocom.met.no/aerocomhome.html http://aerocom.met.no/references.html Chemistry-Climate Model Initiative simulations now available

 1980-2010 “hindcasts” use observed meteorology For more info: http://blogs.reading.ac.uk/ccmi/ Table 3 from Morgenstern et al., GMD, 2017 Evaluating PM2.5 trends in global models

Observed vs. 7-model mean (“ECLIPSE” project) in PM2.5 (%/yr-1) over Europe, U.S.A.

 Models capture larger decline in U.S. surface PM2.5 in 2000s compared to full period (1989-2009)

Table 2 of Myhre et al., ACP, 2017 AerChemMIP Simulations (2018) seek to answer 4 questions:

(historical transient simulations + time slice historical to quantify effective radiative forcing)

(SSP3-7.0 vs. SSP3-7.0-clean)

(time slice historical to quantify effective radiative forcing)

(time slice with doubled natural emissions relative to Pre-industrial control) Collins et al., Analysis of air quality is an AerChemMIP goal GMD, 2017 Hemispheric Transport of Atmospheric Pollution (HTAP) Phase 2

Terry Keating, RTP, April, 2017 www.htap.org HTAP model comparison: O3 metrics

Max 6 month mean of 1 Annual mean of 8 hr daily max hr daily max (m6_MDA1) (a12_MDA8) Daven. K. Henze1, Christopher S. Malley2, Johan C.I. Kuylenstierna2, Harry W. Vallack2, Yanko Davila1, Susan C. Anenberg3, Michelle C. Turner4,5,6,7, Mike R. Ashmore1, Kengo Sudo7, Jan Eiof Jonson8

Model ranges

• Largest model diversity related to natural sources (biomass burning, biogenics) • Range over anthro source regions is smaller, e.g. NE US ~10% HTAP2 model evaluation – PM2.5

2- Comparisons with IMPROVE surface SO4 concentration

Mian Chin Huisheng Bian, Yanko Davilla, Louisa Emmons, Johannes Flemming, Daven Henze, Jan Eiof Jonson, Tom Kucsera, Marianne Lund, Bjorn Samset, Michael Schulz, Kengo Sudo, Toshihiko Takemura, Simone Tilme 3. Applications HTAP2 source attribution results – PM2.5 Mian Chin Huisheng Bian, Yanko Davilla, Louisa Emmons, Johannes Flemming, Daven Henze, Jan Eiof Jonson, Tom Kucsera, Marianne Lund, Bjorn Samset, Michael Schulz, Kengo Sudo, Toshihiko Takemura, Simone Tilme RERER surface conc. RERER total deposition HTAP2 source attribution results – O3

Response of surface O3 to -20% EAS emissions:

Huang et al., ACP, 2016 HTAP2 models as boundary conditions for CMAQ – BC diversity (Hogrefe et al., ACPD, 2017)

See paper for comparison ozonesondes HTAP2 models as boundary conditions for CMAQ– evaluation w/obs (Hogrefe et al., ACPD, 2017)

Regional vs CASTNET, monthly daytime O3 Global vs CASTNET, monthly daytime O3

Regional vs AQS, monthly daytime O3 Regional vs AQS, MDA8 HTAP2 models as boundary conditions for CMAQ– evaluation w/obs (Hogrefe et al., ACPD, 2017)

See paper for analysis of NME, R US US FS US FS FS Health EPA orgs US Cities FS States Industry US FS US US FS FS

28 NASA Health and Air Quality Applied Sciences Team (HAQAST) •Tracey Holloway (Team Lead, UW-Madison) •Bryan Duncan (NASA GSFC) •Arlene Fiore (Columbia University) •Frank Freedman (San Jose State University) •Daven Henze (University of Colorado, Boulder) •Jeremy Hess (University of Washington, Seattle) •Yang Liu (Emory University) •Jessica Neu (NASA Jet Propulsion Laboratory) •Susan O’Neill (USDA Forest Service) •Ted Russell (Georgia Tech) •Daniel Tong (George Mason University) •Jason West (UNC-Chapel Hill) •Mark Zondlo () haqast.org 4 New Tiger Teams from HAQAST

• Led by Bryan Duncan & Jason West: Demonstration of the Efficacy of Environmental Regulations in the Eastern U.S. for Health and Air Quality • Led by Arlene Fiore: Supporting the use of satellite data in State Implementation Plans (SIPs) • Led by Pat Kinney: High Resolution Particulate Matter Data for Improved Satellite-Based Assessments of Community Health • Led by Brad Pierce & Daniel Tong: Improved NEI NOx emissions using OMI Tropospheric NO2 retrievals HAQAST TT: Supporting the use of satellite data in State Implementation Plans (SIPs)

Satellite data may be used in the SIP process as: (1) Weight-of-evidence that a particular strategy is anticipated to succeed in attainment, or to show that transported pollution is confounding attainment efforts (2) Constraints on modeling included in SIPs. (3) Evidence supporting “exceptional events” demonstrations

HAQAST “SIP” TT Objective: Identify at least three different applications of satellite data that we can showcase in a user-friendly, technical guidance document that includes frequently asked questions (FAQs). HAQAST “SIPs” TT activity: How can I incorporate satellite data into regional model boundary conditions?

HAQAST member Jessica Neu (NASA JPL) is generating boundary conditions for use by SCAQMD in their SIP modeling

image credit:: https://www.nasa.gov/mission_pages/aura/main/index.html

image credit: https://www.epa.gov/cmaq/cmaq-models-0

image credit: http://acmg.seas.harvard.edu/geos/doc/man/ 4. Recent advances in global modeling Updates to global model chemistry (Mat Evan’s group) Updates to global model chemistry (Mat Evan’s group) The GFDL CM3/AM3 chemistry-climate model Donner et al., J. Climate, 2011; Golaz et al., J. Climate, 2011 Modular Ocean Model version 4 (MOM4) & GFDL-CM3 Sea Ice Model

Forcing Atmospheric Dynamics & Physics Solar Radiation Radiation, Convection (includes wet Well-mixed Greenhouse deposition of tropospheric species), Clouds, Gas Concentrations Vertical diffusion, and Gravity wave Volcanic Emissions

Atmospheric Chemistry Ozone–Depleting 86 km Substances (ODS) Chemistry of Ox, HOy, NOy, Cly, Bry, and Polar Clouds in the Stratosphere

Chemistry of gaseous species (O3, CO, NOx, hydrocarbons) and aerosols (sulfate, carbonaceous, mineral dust, Pollutant Emissions sea salt, secondary organic) (anthropogenic, ships, biomass burning, natural, & Aerosol-Cloud Dry aircraft) Interactions Deposition 0 km

Land Model version 3 Naik et al.,2013 (soil physics, canopy physics, vegetation dynamics, disturbance and land use) The GFDL CM3/AM3 chemistry-climate model Donner et al., J. Climate, 2011; Golaz et al., J. Climate, 2011 SSTs/SIC from observations or CM3 GFDL-AM3 CMIP5 Simulations cubed sphere grid ~2°x2°; 48 levels

Forcing Atmospheric Dynamics & Physics Solar Radiation Radiation, Convection (includes wet Well-mixed Greenhouse deposition of tropospheric species), Clouds, Gas Concentrations Vertical diffusion, and Gravity wave Volcanic Emissions

Atmospheric Chemistry Ozone–Depleting 86 km Substances (ODS) Chemistry of Ox, HOy, NOy, Cly, Bry, and Polar Clouds in the Stratosphere Chemistry of gaseous species (O , CO, 3 AM3 option to NOx, hydrocarbons) and aerosols (sulfate, carbonaceous, mineral dust, Pollutant Emissions nudge to “real winds” sea salt, secondary organic) (anthropogenic, ships, biomass burning, natural, & Aerosol-Cloud Dry High-res. ~0.5°x0.5° aircraft) Interactions Deposition 0 km e.g., applied to CalNex, SENEX

Land Model version 3 Naik et al.,2013 (soil physics, canopy physics, vegetation dynamics, disturbance and land use) High resolution GEOS-Chem global simulations

GCHP: Here shown at 0.5°, can go to 0.125°.

GEOS-FP met

Cube-sphere

MPI-enabled

Seb Eastam, Mike Long Global-multi-regional Two-way Coupled Modeling Based on GEOS-Chem Slides from Jintai Lin • High-res regional nested simulations ‘correct’ global model • Global and multiple regional models interact simultaneously • High computation efficiency and low model complexity

Global model : ~ 200km res. Regional models: 25-50 km res.

41 Yan Y.-Y. et al., ACP, 2014, 2016 2-way Coupling Better Simulates Tropospheric CO CO at 6.5 km altitude. 2008/07/01 – 2008/08/15 Two-way model Global model

Two-way - Global Relative difference

42 Yan Y.-Y. et al., ACP, 2014, 2016 2-way Coupling Improves CO Simulation over the Pacific 2-way Model Better Simulates Surface O3 Instrumented modeling

Data assimilation (ACP, 2017) - state updates - forecasts

Inverse modeling - emission constraints

Sensitivity/Attribution - tagging schemes - adjoint modeling

Reduced form modeling - Source/receptor parameterizations (Wild et al 2012) - response surface modeling: FASST (van Dingenen), H-CMAQ (Dong, Fu) Outstanding issues

Transport (e.g., Orbe et al., 2017)

Boundary layer mixing (e.g., Travis et al., 2017)

Long-term O3 trends (Parrish; Lin et al., 2017)

SOA, aerosol microphysics

Biomass burning

Dry Deposition (e.g., Hardacre et al., 2015; Clifton et al., 2017) Dry deposition may be an important control on daily surface ozone

Pennsylvania State University [78ºW, 41ºN, 378 m]

Observations GFDL AM3 GFDL AM3, but vd reduced by 35% in drought-stricken U.S. regions

Lin et al., 2017  requires improved mechanistic understanding of ozone dry deposition Strong year-to-year variations in ozone deposition velocities derived from observations at Harvard Forest (central MA) ) 1 - ( cm s dry deposition velocity deposition velocity dry

 Not simulated by GEOS-Chem CTM [Wesely 1989 scheme] yet EUS surface ozone strongly sensitive to dry deposition [Walker et al., 2014; val Martin et al., 2014] Clifton et al., GRL, 2017 end Uncertainty in U.S. air pollutant emissions and trends

From Xing et al., ACP, 2013 O3, surface level, BASE, daily, MDA1 (uncorrected) CHASER_re1 CHASER_t106 EMEP_rv4.5

EMEP_rv48 GEOS-Chem adjoint GEOS-Chem

OsloCTM.v2 RAQMS Premature deaths from long-term O3 exposure owing to respiratory causes using Jerrett 2009 exposure-response

Multi-model mean

Standard deviation

For updated analysis using more recent exposure-response, and LR transport, see poster 88, Copper Top II