background o3 scientific assessment workshop denver, march … · 2017. 5. 10. · background o 3...
TRANSCRIPT
Background O3 Scientific Assessment WorkshopDenver, March 28-29, 2017
Dan Jaffe, University of Washington
Photo of LA smog from citiesspeak.org
Acknowledgements• American Petroleum Institute
– Cathe Kalisz
• Western States Air Resources Council (WESTAR)– Tom Moore– Mary Uhl– Bob Lebens
BOSA Core Team• Prof. Dan Jaffe (University of Washington)• Dr. Owen Cooper (University of Colorado /NOAA ESRL)• Prof. Arlene Fiore (Columbia University)• Dr. Barron Henderson (EPA OAQPS)• Prof. Daven Henze (University of Colorado)• Dr. Andrew Langford (NOAA-ESRL)• Dr. Meiyun Lin (Princeton University / NOAA GFDL)• Mr. Tom Moore (WESTAR-WRAP)• Dr. Gail Tonnesen (EPA Region 8)• Prof. Ted Russell (Georgia Institute of Technology)
The Ozone Serenity Prayer
Grant me the courage to control the ozone I can control.
The serenity to accept the ozone I can’t control.
The wisdom to know the difference.
Goals for this assessment• Summarize key spatial and temporal patterns of baseline O3. • Review published work on USBO for the continental U.S. and
summarize consistent and robust patterns. • Identify discrepancies between estimates of USBO and, if
possible, the causes for these discrepancies. • Examine different approaches used to get USBO and
evaluate strengths/weaknesses of these approaches.• Examine evidence for NCOS and their role in daily and
seasonal O3 concentrations.• Review methods to quantify NCOS and evaluate
strengths/weaknesses of each approach.• Develop a set of recommendations for future research in this
area.• Emphasis on new research since 2011.
Goals for the workshop• Provide a forum for broader input to the assessment from all
knowledgeable experts;• Provide a forum for discussion of key uncertainties;• Identify new methods and tools that the core group might not
be aware of;• Identify specific scientific publications that we may have
missed;• Make specific recommendations for future research that the
committee should consider.
• Your input will be most useful to the extent it:• Relates to the goals of this assessment;• Provides a specific reference or publication that supports your
input;
Assessment Process and Timeline• October 2016: DJ contacted by API to lead assessment;• Fall 2016: Team organized/invited;• Jan 2017: First team conference call on scope;• Wtr 2017: Continued discussions on scope, bibliography,
definitions;• March 2017: Draft document released on goals of the
assessment, bibliography, early recommendations, etc.• March 2017: Denver Workshop to get broad community
input;• Spring-Summer 2017: Continued discussions by core team;• August 1 (goal): Submission of critical review paper to EST
or other journal.
What will we cite in this assessment? How will we utilize grey lit?
• Strong preference for published papers from peer-reviewed scientific literature;
• Publically available EPA documents can be cited if peer-reviewed;
• Papers submitted or in review, may be considered if results are new or of very high significance;
• Grey literature (eg modeling studies for State ImplentationPlans) can be cited and, if cited, will be permanently archived with a doi in a e-repository through the University of Washington library system.
• We can’t cite everything. Emphasis will be on work that demonstrates key aspects or research on background O3.
Definitions• U.S. background O3 (USBO): O3 formed from all natural sources plus
anthropogenic sources in countries outside the U.S. , plus impacts from CH4.
• North American Background (NAB) is defined as O3 formed from all natural sources plus anthropogenic sources in countries outside North America, plus impacts from CH4.
• USB and NAB must be determined using chemical transport models or source apportionment.
• Baseline O3: O3 measured at relatively remote sites that have little or no recent influence from US domestic emissions.
• Non-Controllable O3 sources (NCOS): These are sources of O3, or its precursors, that could not be reasonably controlled by domestic legislation. Examples of NCOS are intrusions of stratospheric air or emissions from wildfires.
• All of the terms above can be expressed as seasonal, monthly or daily means, maximum daily 8-hour averages (MDA8) or using other statistical metrics.
Why does background O3 matter?• Background O3 represents the endpoint in health and
ecosystem risk analysis or cost-benefit analyses as emissions are reduced;
• Knowledge on background O3 is needed to set reasonable and achievable air quality standards;
• Knowledge on background O3 and non-controllable sources is needed to determine when exceedances might be beyond local control.
• A model's ability to capture baseline ozone indicates that we are capturing the key processes and increases our confidence in assessing reduction strategies for O3 as well as other air pollutants.
Why does background O3 matter?
History• Discovery of Intercontinental transport• NAS/NRC Assessment (NRC 2009).• UT-Austin Panel (McDonald-Buller et al.
2011).
Discovery of intercontinental transport of air pollutants
Daniel A. Jaffe, Theodore Anderson, Dave Covert, Robert Kotchenruther, Barbara Trost, Jen Danielson, William Simpson, Terje Berntsen, Sigrun Karlsdottir, Donald Blake, Joyce Harris, Greg Carmichael and Itsushi Uno.
Terje Berntsen, Sigrun Karlsdottir and Daniel Jaffe
Daniel J. Jacob, Jennifer A.Logan and Prahant P.Murti
Cited 395 times.
Cited 115 times.
Cited 266 times.
UW Research in the PNW on O3, PM and Hg
Beechcraft Duchess
Mt. Bachelor, ORCheeka Peak, WA
NCAR/NSF C130
NASA TRACE-P 2001
NOAA ITCT 2002
NASA INTEX-B 2006
O3 at Mt. Bachelor in GEOS-CHEM Global Model
Zhang et al 2008, 2009
Black is observationsRed is model
Blue is Asian contribution
Purple is change due to 2000-2006 rising Asian emissions.
Other work done by NOAA, NASA, etc
Message = Polluted air can mix with stratospheric air. Hard to untangle sources.
NAS/NRC Study (2009)
Charles Kolb, Aerodyne Research [Chair]Tami Bond, Univ Illinois. Urbana [emission inventories, PM]Mae Gustin, Univ of Nevada - Reno [mercury]Gregory Carmichael, Univ of Iowa [atm chemistry and modeling]Kristie Ebi, IPCC TSU II [pollution impacts]David Edwards, NCAR [remote sensing]Henry Fuelberg, Florida State Univ [meteorology]Jiming Hao, Tsinghua Univ [pollution control, Asian perspective]Daniel Jacob, Harvard Univ [ozone, chemical-transport modeling]Daniel Jaffe, Univ Washington-Bothell [ozone, PM, mercury]Sonia Kreidenweis,Colorado State Univ [PM chemistry, obs]Katharine Law, CNRS (France) [atm.chem, European perspective]Michael Prather, UC Irvine [atm chemistry, radiative forcing]Staci Simonich, Oregon State Univ [POPs]Mark Thiemens, UC San Diego [isotopic analysis methods]Laurie Geller, Study Director
NAS panel on Global Sources of Local Pollution
Baseline concentrations of tropospheric O3 have risen above pre-industrial levels by 40 - 100%.
‘Imported’ pollution mixes to the surface and contributes to increased concentrations of O3 over populated regions, with detrimental impacts on human health and ecosystems.
U.S. NAAQS O3 violations are caused primarily by domestic emissions, but concentrations are augmented by the changing baseline and episodic non-domestic sources. The influence of imported O3 will become more acute if US NAAQS continue to get tighter and background concentrations increase due to increasing emissions in developing countries.
NAS (2009) selected findings on Ozone
Changing O3 NAAQS:1979: 120 ppb -1 hr1997: 85 ppb - 8 hr2008: 75 ppb - 8 hr2015: 70 ppb - 8 hr
• Strengthen the observation network to include more vertical profiles, mtn top and multi-species observations;
• More rigorous analysis of existing long-term observations, to better understand trends in background ozone;
• Coupled observational/modeling studies, especially focused on quantifying the exchange between the boundary layer and free troposphere;
• Calibrate, test, and improve models and develop new methods, such as adjoint and data assimilation.
• Integration of satellite observations;• Development of an integrated source attribution system.
Selected NAS recommendations for Ozone
HTAP models with CASTNET monthly mean MDA8CAMCHEM ECHAM5 EMEP FRSGCUCI GEMAQ-EC GEMAQ-v1p0 GISS-PUCCINI GMI-v02f INCA-vSSz LLNL-IMPACT MOZECH OsloCTM2 TM5-JRCOBSMulti-model meanGEOS-Chem_v07GEOS-Chem_v45 MOZARTGFDL
20
30
40
50
60
70
80
J F M A M J J A S O N D
Mountain West
15
25
35
45
55
65
75
85
95
105
J F M A M J J A S O N D
Southeast
Ensemble mean gives best fit with monthly mean obs;Large positive bias for SE US
in summer. Small negative bias for
Western US in springRelationship to domestic
and/or foreign influence?
Reidmiller, Fiore et al 2009 ACP
UT-Austin Panel: March 30-April 1, 2011
Black: Obs MDA8
Red: GEOS-ChemMDA8Blue is USBO MDA8
Gothic, CO- 2.9 km aslStrat intrusion
UT Panel: Policy and Research Findings(McDonald-Buller 2011)
Research recommendations:• Multi-model comparisons;• Better integration of modeling and baseline observations
(obs at “Relatively remove sites”);• Observations on commercial aircraft in the U.S.Policy recommendations: Clarity on definition of background O3; Need for improved knowledge on BGO including spatial and
inter-annual variability and significance w.r.t. standards; Better communication between federal and state agencies
on BGO.
What do observations say about current “baseline” O3 ?
Max daily 8 hour avg O3 at Mt. Bachelor for 2012-2014 Latest 3-year design value =79 ppbv
What are causes of high O3 in the Western US?
MDA8 O3 at the Mt. Bachelor ObservatoryCurrent U.S. O3 standard of 70 ppbv.
Causes of high O3 days at MBO: Strat intrustions, Asian pollution and wildfires
Ambrose et al 2011
Baylon et al 2014
MDA8 vs DOY
Daily MDA8 (3-year average)with 10-day running smooth
Current Design Values
Design Values for 2013-2015 (3-year average of annual 4th highest)
Design Values vs Altitude
Design Values for 2013-2015 (3-year average of annual 4th highest)
Mt. Bachelor
Design Values vs Altitude
Design Values for 2013-2015 (3-year average of annual 4th highest)
Uncertainty in Background O3
• USB and NAB O3 can not be determined directly-requires a model;
• Need appropriate observations to evaluate the model;
• How do we evaluate uncertainty in the answers?
2007 April-Oct mean MDA8 USB from CMAQ modeling
• Is this the answer?• What do other models
show?• What about other years?• What about frequency of
high USB events?• What do model-obs bias
and correlations tell us about uncertainty?
• What do different chemical species tell us?
• How can baseline obshelp evaluate modeled USB?
U.S. EPA. Policy Assessment for the Review of the O3 National Ambient Air Quality Standards, 2014.
Some thoughts on what’s new since 2011
• Models are getting a lot better at capturing means, but all models need better evaluation with baseline and boundary observations;
• Multi-model comparisons are very useful to constrain USBO;• Days with O3 from non-controllable sources are not un-common
and we need better, simpler and faster tools to id and quantify;• In many cases, models can capture some high USBO days, especially
for stratospheric intrusions;• Models still have very limited capability to re-produce O3 from
wildfires correctly, need additional tools to do this;• We still need an integrated system for air quality management in
the US.
Questions on Goals, Process or Definitions?
WildfiresWolverine Fire, Lake Chelan, WashingtonAug 3, 2015
Area Burned for US Wildfires (NIFC)
The last decade has seen a significant increase in the area burned. ~60% of area burned is in W.U.S;~20% of area burned is in Alaska.
Review O3 production from >100 published studiesJaffe and Wigder 2012
Boreal/ Temperate:
Plume Age Mean ∆O3/∆CO (ppbv/ppbv) (# plumes)
Range of ∆O3/∆CO
≤ 1-2 days 0.018 (n=55) -0.032-0.34
2-5 days 0.15 (n=39) -0.07-0.66
≥ 5 days 0.22 (n=29) -0.42-0.93
Jaffe, D.A. and Wigder, N.L., Ozone production from wildfires: A critical review. Atmos. Envir., doi:10.1016/j.atmosenv.2011.11.063, 2012.
1. Generally more O3 production downwind.2. Large variability within all categories.
Wildfires can make O3 very quickly
Mt. Bachelor observations of the Pole Creek Fire on three successive days. O3 production of 20-50 ppbv in 6 hours. Many other examples of fast O3 production in literature.O3 production is highly variable!
O3
COAerosol scattering
Primary emissions in a wildfire plume
CO2Primary aerosols (largely Organic compounds)Volatile Organic Compounds (VOCs)Oxygenated-VOCs (eg CH2OH; CH3COCH3, CH3CHO, etcCO, NOx (NO+NO2), NH3, HONO, etc
100s of VOCs have been identified in wildfire smoke (Akagi et al 2011)
Emissions depend on combustion efficiency
CO2Primary aerosols (largely Organic compounds)Volatile Organic Compounds (VOCs = gas phase)Oxygenated-VOCs (eg CH2OH; CH3COCH3, CH3CHO, etcCO, NOx (NO+NO2), NH3, HONO, etc
Smoldering Flaming(white smoke) (black smoke)
Wolverine Fire, Lake Chelan, WashingtonAug 3, 2015
Emissions depend on combustion conditions
CO2Primary aerosols (largely Organic compounds)Volatile Organic Compounds (VOCs = gas phase)Oxygenated-VOCs (eg CH2OH; CH3COCH3, CH3CHO, etcCO, NOx (NO+NO2), NH3, HONO, etc
Smoldering FlamingMore VOCs ↔ Lower VOCsLess Black carbon ↔ More Black CarbonLess NOx ↔ More NOxMore NH3 ↔ Less NH3More primary PM ↔ Less primary PM
Per kg fuel consumed
VOCs/NOx molar ratio ≈ 20-150Oxy-VOCs/total VOCs ≈ 0.5
PAN/NOy in 6 fire plumes at MBO
• PAN is about 50% of the NOyin fire plumes at MBO (Briggs et al 2016)
• Typical urban plumes 12-15% (Roberts 2008)
WRF model over-predictions of O3Baker et al 2016
Sensitivity runs (in blue) used reduced photolysis rates, but this had little impact on over-prediction.
Key challenges to model O3 in wildfire plumesHighly variable emissions from fire to fire;Highly variable plume injection heights; Importance of sub-grid scale processing; Large emissions of Oxygenated-VOCs, which are
not well represented in most models; Impacts on photolysis rates due to smoke; Impacts due to heterogenous chemistry.
Jaffe et al 2013; Zhang et al 2014; Lu et al 2016Baker et al 2016
Key challenges to model O3 in wildfire plumes:These are solvable problems in the future
Highly variable emissions from fire to fire; Satellite observations (e.g. FRP, CO, PM, NOx) may be able
to provide emissions on each fire. Highly variable plume injection heights; Satellite observations using Lidar may be able to provide
plume heights on each fire. Importance of sub-grid scale processing; Embedded plume within gridded models. Large emissions of Oxygenated-VOCs, which are not well
represented in most models; Need to update model chemistry to reflect the reality of
fire emissions.
Impacts on photolysis rates due to smoke; Impacts due to heterogenous chemistry.
Wildfires and O3: Implications Regional models may over-estimate O3 production close
to fire, under-estimate downwind.Global models may underestimate O3 production overall
due to non-linearity of production.Mixing of smoke into NOx rich urban areas can further
enhance O3 production thru both enhanced VOCs plus NOx (Pfister et al 2008; Singh et al 2010);
NOx → PAN → → → → → → → → → PAN → NOx →O3
Colorado MDA8 O3 Impacted by Wildfires: June 16, 2016
Very difficult to identify, quantify and prove an exceptional event.Max 24 hr PM in the core area = 15 ug/m3; Highest 1-hr PM = 25 ug/m3.
Overhead HMS smoke shown in Grey
Fires
Statistical Approach: Generalized Additive Model Examines the relationship between the observed mixing ratios
(maximum daily 8-hour average (MDA8) and meteorological factors. GAMs can incorporate numerical, ordinal or categorical variables. Possible factors to include are temp, WS, WD, RH, solar flux, etc. Use “mgcv” package in “R” software. Outliers (high residuals) represent an additional O3 source and are
candidates for further investigation. g(O3i) = f1 (tempi) + f2 (windi) + f3 (DOYi) … + residuali
Where f1, f2, etc are “link” functions which are obtained from spline fits to the observations. The “i” refers to each daily observation.
Jaffe et al 2004; 2013; Camalier et al 2007; CARB 2011; EPA 2015; Sun 2015; Gong et al 2017
Aug 20, 2015 with 24 hr avg PM2.5
Graphics from AIRNOW-Tech site.PM2.5 with HMS smoke product.
Aug 20, 2015 with 24 hr avg PM2.5
PM is highest near the fires and decreases with distance.
<12 ug/m3
<35 ug/m3
<55 ug/m3
<150 ug/m3
Aug 20, 2015-MDA8 O3
The observations show that O3increases away from the fire.Highest in urban centers.
<55 ppbv56-71 ppbv72-85 ppbv
GAM residuals for Aug. 20, 2015 = O3 due to the wildfire
-1
+13 +11
+6 +7+16 +14
Wildfire O3Summary• HMS smoke product shows good utility to
identify overhead smoke plumes in near real time.
• Statistical model shows good utility to estimate O3 production due to wildfires;
• Combining surface PM with HMS smoke allows for positive fire identification. GAM residual can then quantify influence on MDA8 O3;
• See our poster for more details.
Extra Slides
Generalized Additive Model results (for MDA8)
Spokane(WA)
Fort Collins
(CO)
Salt lake(UT)
Boise(ID)
R2 0.60 0.55 0.54 0.52
Denver(CO)
Yell.(WY)
Provo(UT)
Houston(TX)
R2 0.56 0.41 0.50 0.77Variables used: Daily TMAX, Daily TAVG, DPAVG, RHAVG, TDELTA, 8hr_vectorWD, 8hrWS, YEAR, DOW, DOY, Trajectory quadrant, Trajectory distance, U700mb, V700mb.
GAM result vs Observed MDA8 for Denver
Distribution of surface PM2.5 on days w/wo HMS smoke
SLC-Hawthorne siteBlue No smoke daysRed: HMS Smoke days
Kaulfus, Nair, Jaffe and Christopher, in-prep.
0 10 20 30 40µg/m3
Model residual for Salt Lake City for 104 days in 2008-2015 with enhanced PM and overhead smoke (HMS)
This shows that not all days with smoke make O3, but smoke days are statistically higher than non-smoke days. Maximum influence on MDA8 ~25 ppb.
NOAA Hazard Mapping System (HMS) Fire and Smoke Product (FSP)
Based on visible imagery from 7 NOAA and NASA satellites. This gives near continuous coverage of North America;
Human analysts identify fire locations and draw contours of smoke location (low, medium, high);
KMZ and other formats available in near real time.
But no information on plume height or whether smoke extends to the surface.
HMS Smoke and fire locations for Aug 27, 2015 with surface O3 conc.
Graphics from EPA Airnowtech site
Transport of Asian pollution to Mt. Bachelor, April 25, 2004
POPs were also enhanced during this period including HCH and particulate PAHs. (Jaffe et al 2005; Weiss-Penzias et al 2006; Primbs et al 2008; NAS 2009)