NASA Technical Reports Server (NTRS) 20180000781: Air Quality Modeling Using the NASA GEOS-5 Multispecies Data Assimilation System

5.1 MB·English

by  NASA Technical Reports Server (NTRS)

National Aeronautics and Space Administration
Air Quality Modeling Using the NASA GEOS-5
Multispecies Data Assimilation System
Christoph A. Keller1,2
Steven Pawson1, Krzysztof Wargan1, Brad Weir1,2
)
1NASA Global Modeling and Assimilation Office (GMAO)
2Universities Space Research Association (USRA)
AMS 98th Annual Meeting
10 January 2018
GMAO Global Modeling and Assimilation Office
gmao.gsfc.nasa.gov
R. Van Dingenen et al. / Atmospheric Environment 43 (2009) 604–618 613
Table 3 data may not adequately represent the regional-scale ozone
Regionallyaveragedmodelled-to-measuredratioofbothmetricsduringthemonths
concentrations. In fact, out of the 4 N.-Indian measurement
May–June–July, at a model height of 30m and 10m above the surface respectively.
stations, 3 are located in densely populated urban areas where
Region M7, 30m M7,10m AOT40, 30m AOT40,10m ozone levels may be suppressed by local titration, whereas the 4th
SW US 1.03 1.01 1.02 0.95 is a regional station however using a passive sampler as measure-
SE US 1.23 1.20 1.25 1.12 ment technique.
US Great Lakes 0.87 0.83 0.60 0.44
Indeed, more recent air pollution measurements in the peri-
Mediterranean 1.07 0.99 0.98 0.73
urban and rural areas around Varanasi in the Indo-Gangetic plane
Central Europe 0.97 0.95 0.37 0.25
(Agrawal et al., 2003) show that summer average ozone concen-
Japan 1.33 1.21 1.71 1.27
trations may span from 10 to 58 ppbV, depending on the location
relative to the nearby city. In contrast to this, the S.-Indian obser-
vations are obtained in peri-urban locations, and in this case the
an in-service Airbus aircraft in the framework of the MOZAIC pro- agreement with the model is much better.
gramme during the flights. From the measured MOZAIC vertical We also evaluated the model performance in reproducing
ozone profiles, we used tRh.VaenDlinogenweneetasl.t/AtamolstphietricuEndvireonmvenat4l3u(20e09)a60v4–a618ilable for the mont61h3ly accumulated AOT40 and monthly averaged M7 for those
loTacblae3tions in Central-West Africa (Lagos, Adabtaidmjaaynn,otDaodeuquaatlealy).represent the regional-sclaolecaoztoinoens where hourly ozone data are available (Europe, US and
Regionallyaveragedmodelled-to-measuredratioofbothmetricsduringthemonths concentrations. In fact, out of the 4 N.-Indian measurement
May–TJuhne–eJulye,artarmoodrelhbeiaghrtosf30omnandt1h0meabomveteheasusrfaucerreespdectivvelya. lues represent the standard Japan). Results are shown in Fig. 7 (M7) and Fig. 8 (AOT40). Note
stations, 3 are located in densely populated urban areas where
Region M7,30m M7,10m AOT40,30m AOT40,10m ozonelevelsmaybesuppressedbylocaltitration,whereasthe4th
deviation on the station monthly means. They do not include the that for these metrics, obtained during daytime only, the vertical
SWUS 1.03 1.01 1.02 0.95 isaregionalstationhoweverusingapassivesamplerasmeasure-
inSEdUiSvidual 1s.2t3ation1’.s20 stand1.2a5rd dev1i.1a2tions omnenthteicghnhiqeuer. temporal scales, gradient becomes less pronounced than for the monthly means,
USGreatLakes 0.87 0.83 0.60 0.44 Indeed, more recent air pollution measurements in the peri-
Mediterranean 1.07 0.99 0.98 0.73
nCoenrtratlEhureopeana0.9l7ytical0.9u5ncert0.a37inty. 0.25 urbanandruralareasaroundVaranasiintheIndo-Ganbgeeticcpalaunese of the better vertical mixing of the boundary layer. For M7,
Japan 1.33 1.21 1.71 1.27 (Agrawal et al., 2003) show that summer average ozone concen-
In general, the model is reproducitrnatgionsrmeayasspoannfraombl1y0tow58peplblV,dtehpeendingontthheleocaatiognreement between model and measurements is excellent for
relativetothenearbycity.Incontrasttothis,theS.-Indianobser-
monthly mean ozone concentrations invatironesgarieoonbtsainewd inhpeerrie-urbqanuloacaltiitonys-, and in thsisocuasethth-ewest and south-east US, the Mediterranean area, and central
anin-serviceAirbusaircraftintheframeworkoftheMOZAICpro- agreementwiththemodelismuchbetter.
controlled ozone monitoring programs are routinely running Europe. For the US Great Lakes region, spring time M7 is under-
gramme during the flights. From the measured MOZAIC vertical We also evaluated the model performance in reproducing
ozoneprofiles,weusedthelowestaltitudevalueavailableforthe monthlyaccumulatedAOT40andmonthlyaveragedM7forthose
(Central Europe, U.S.A., Japan). During the summer months, the predicted by 15–20 ppbV but summer months are well reproduced.
locationsinCentral-WestAfrica(Lagos,Abidjan,Douala). locations where hourly ozone data are available (Europe, US and
moTdheeelrlreordbar1s0onmthemceoasnurcedevnalutersartepiroesnenstthfeasltlanwdaridthinJapa1n).sRetsaulntsdaraershdowdnienvFiiga.7ti(Mo7n)anodfFig.8(AOFTo40r).NJaotpe an, the summer months are significantly over-predicted by
deviationonthestationmonthlymeans.Theydonotincludethe thatfor thesemetrics, obtained duringdaytimeonly, the vertical
thinedivoidbuaslsetartviona’stsitoanndasrdadnevdiatitonhseonsheigahesrotenmpaolratlsrcealnes,d isgrwadieenltlbreceompersoledssupcroenodun.cAedltshoanffoorrthe montuhlypmetaons, 20 ppbV. Modelled M7 (as is the case for M12 and the
northeanalyticaluncertainty. becauseofthebetterverticalmixingoftheboundarylayer. ForM7,
SouInthge-nEeraal,sttheAmsoidael aisnrdeproSduocuingthreeasronnablIynwdeillathwe e tfihenagdreeamenstabettiwsefeancmtooderlyandmmeoasdureemlentsisexmcelolennttfohrly mean) appears not to be very sensitive to the ozone
monthly mean ozone concentrations in regions where quality- south-westandsouth-eastUS,theMediterraneanarea,andcentral
pceorntfroollremd oazonnecem.onIitnorinNg oprrotghramesrnareInroudtiinaelyarnundningtheEutrwopeo. FoAr tfhreiUcSaGnreatrLeagkeisorengison,, stphrineg time Ms7aismundpelr-e height.
(Central Europe, U.S.A., Japan). During the summer months, the predictedby15–20ppbVbutsummermonthsarewellreproduced.
model is significantly overestimating the observed ozone levels. The picture looks similar for AOT40 (Fig. 8), but differences
modelled10mconcentrationsfallwithin1standarddeviationof ForJapan,thesummermonthsaresignificantlyover-predictedby
theobservationsandtheseasonaltrendiswellreproduced.Alsofor up to 20ppbV. Modelled M7 (as is the case for M12 and the
This is particularly of concern for S.-India seen the expected impact between model and measurements are amplified as a consequence
South-East Asia and Southern India we find a satisfactory model monthly mean) appears not to be very sensitive to the ozone
onperfocrrmoanpce.lInoNsosrethser.n ITndhiaeandrethaestwoonAfrficoanrretghiones, twheorssaempmlehoeidghet.l performance in of the cumulative nature of the metric in combination with a non-
National Aeronautics and Space Administration model is significantly overestimating the observed ozone levels. The picture looks similar for AOT40 (Fig. 8), but differences
thTheissiseparrteicguliarolynofscoinscernnofortS.c-Inlediaasreeanthperexipoecrtie.diUmpnacctertbaetiwneetniemsodeilnandthmeeasueremmeinstssairoeanmpolififedasacozneserquoenctehreshold (Tuovinen et al., 2007). In particular for Central
on crop losses. The reason for the worse model performance in ofthecumulativenatureofthemetricincombinationwithanon-
ozthoesnereegiopnsrisencoutcrlesaroarpsriormi.Unacyertaibnteiesiantnheeimmisspionoorftanzetroftharcesthooldr,(Tuaosvinewn eetlall., 2a0s07).tIhn eparticularEfour Creontprael , the difference between model and measured AOT40 is
Air quality is a global problem
ozone precursors may be an important factor, as well as the Europe, the difference between model and measured AOT40 is
reduced model resolution over Africa. But also the observational disturbingly high; other regions are performing better. Table 3
reduced model resolution over Africa. But also the observational disturbingly high; other regions are performing better. Table 3
Fig.9. Averagerelativeyieldlossfrom2metricsforthe4crops,year2000.
World Bank: ~$5 trillion Up to 50% of crop yield can be lost to
in welfare losses in 2013 ozone pollution (e.g. VanDingenen, 2009)
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov
Fig. 9. Average relative yield loss from 2 metrics for the 4 crops, year 2000.
National Aeronautics and Space Administration
GEOS-CF model produces near real-time air quality forecasts
GEOS-FP
Ozone
GEOS-Chem Nitrogen Dioxide
Carbon Monoxide
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov
National Aeronautics and Space Administration
Model has low bias compared to OMI NO observations
2
S
O
E
G
I
M
O
I
M
O
-
S
O
E
G
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov
National Aeronautics and Space Administration
Also low bias in (surface) carbon monoxide CO
]
v
b
p
p
[

O
C
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov
National Aeronautics and Space Administration
Toward an air quality modeling system in the NASA GEOS model
GEOS-Chem
composition
GEOS-5 data
forecast model
assimilation system:
à Weather
Chemical data
à Aerosols
assimilation:
O , NO , CO, …
3 2
Air Quality Modeling System
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov
National Aeronautics and Space Administration
The GEOS chemical data assimilation system
GEOS-5
Ø Based upon GEOS-ADAS (GSI)
Ø Joint assimilation of NO , CO, O
2 3
Data Assimilation System (GSI)
Ø Weakly coupled (no covariances)
MOPITT
Ø 6-hour assimilation window
OMI
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov
National Aeronautics and Space Administration
Assimilate independent NO scale factors for three layers
x
boundary layer free troposphere
stratosphere
NO scale factor
x
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov
National Aeronautics and Space Administration
NO assimilation reduces model-observation mismatch
2
JJA SON DJF
I
M
O
-
S
O
E
G
Control
I
M
O
-
S
O
E
G
Analysis
Ø Are we now overestimating NO over polluted regions?
2
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov
National Aeronautics and Space Administration
Impacts of ozone assimilation are primarily seen in stratosphere
Boulder, 2013-07-25
GMAO Global Modeling and Assimilation Office
[email protected]
gmao.gsfc.nasa.gov