Assimilation of Satellite Ozone Measurements during the 1999 Southern Oxidants Study:

1
Assimilation of Satellite Ozone Measurements
during the 1999 Southern Oxidants Study Impact
on Continental US Regional Air Quality
Predictions R. Bradley Pierce1, Todd Schaack2,
Jassim A. Al-Saadi1, Ivanka Stajner3, Hiroo
Hayashi3, Steven Pawson3, Martin D. Mueller3, Don
Johnson2, Jack Fishman1, Jim Szykman4
1NASA/LaRC, 2UW/SSEC, 3NASA/GMAO, 4EPA/OAQPS
2
NASA Air Quality Applications
Goal Improved capability to Air Quality managers
to assess, plan implement sound-science,
emissions control strategies, policy, air
quality forecasts.
The primary partners for the Air Quality
Management program are the US Environmental
Protection Agency (EPA) and the National Oceanic
and Atmospheric Administration (NOAA).
From NASA Earth Science Enterprise Applications
Plan
3
(No Transcript)
4
Constituent Assimilation in GMAO
Stratospheric Ozone mature system, ability to
assimilate different types of data from multiple
platforms (PI Ivanka Stajner) Aerosol evolving
system, focusing initially on MODIS data with
GOCART aerosol modules (PI Arlindo da
Silva) Carbon species CO starting (AIRS, TES)
CO2 under development (PI Steven Pawson)
Foci Research-based assimilation efforts are
the main focus (e.g., data impacts confronting
models) New development of ozone modules for
GEOS-5 Involvement in NASA (and other) field
missions (NRT products) pre-AVE (Jan 2004)
INTEX (summer 2004),
5
Regional Air Quality Modeling System (RAQMS)
A NASA Langley/UW-Madison Cooperative Research
Effort
Public Impact
RAQMS Ozone Assimilation/Prediction February 27,
2001
Regional Prediction
Global Assimilation
Scientific Understanding
NASA Satellite Products
RAQMS Pierce et al., 2003 is a nested global-
to regional-scale meteorological and chemical
modeling system for assimilating and predicting
the chemical state of the atmosphere (air
quality).
RAQMS includes online chemistry from the NASA
LaRC unified (troposphere/stratosphere) chemical
mechanism driven by the UW-Hybrid (global
isentropic/sigma coordinates) and UWNMS (regional
Non-Hydrostatic) dynamical cores.
6
15
1 NOAA National Environmental Satellite Data and
Information Service 2 NOAA Office of Oceanic and
Atmospheric Research 3 NOAA National Center for
Environmental Prediction
7
4
8
RAQMS Assimilation Procedure
5
Occultation Measurement
Five Day Stratospheric Trajectories
P (mb)
RAQMS First Guess T6hr
RAQMS First Guess T12hr
RAQMS First Guess T18hr
Troposphere
O3
Assimilation Cycle
Time
Trajectory mapped solar occultation measurements
constrain stratospheric O3 assimilation. Mass
weighted column ozone analysis increment
(TOMS-background) provides constraint on
tropospheric column.
9
Data Sets used for SOS99 evaluation

• Daily ozone profiles from Global Ozone Monitoring
  Experiment (GOME) Neural Network Ozone Retrieval
  System (NNORSY) Muller et al., 2003

GOME NNORSY absolute and relative error profile

• Global WMO ozone sondes

GOME is onboard the European Remote Sensing
Satellite (ERS-2) PI, John Burrows Institute for
Environmental Physics, University of Bremen,
Germany
10
GOME NNORSY Stratospheric Comparison June-July,
1999 Zonal Means
FvDAS and RAQMS show similar difference patterns
in the upper (above 10mb) stratosphere. FvDAS
shows additional differences in the Northern
Hemisphere lower stratosphere.
11
GOME NNORSY Tropospheric Comparison June-July,
1999 Zonal Means
Both RAQMS and FvDAS show high biases of 10-15
ppbv relative to the tropical GOME NNORSY
retrievals. FvDAS shows an additional low bias of
10-15 ppbv relative to the mid-latitude GOME
NNORSY retrievals.
12
Tropospheric Ozone Column (TOC)
6
TOMS/SBUV TOC is derived using the residual
techique Fishman and Balock, 1999 GOME NNORSY
and FvDAS Northern Hemisphere TOC are low
relative to TOMS/SBUV. RAQMS is lower than
TOMS/SBUV over Northern Hemisphere industrial
areas (Europe, Eastern US, E. Asia)
13
Comparison with Sonde derived TOC (Huntsville, AL)
Both RAQMS and FvDAS capture daily variations
in Tropospheric O3 Column at Huntsville, AL. The
single GOME NNORSY data point significantly
underestimates the observed tropospheric
column.
Huntsville is not included in GOME NNORSY
training set
14
June-July, 1999 Sonde profile comparison
(Latgt30oN)
RAQMSG assimilation shows mean biases of lt10
except in the mid troposphere (400mb). RAQMSG RMS
errors reach 50 at 400mb and the surface.
FvDAS assimilation shows mean positive biases
of 50 in the lower stratosphere (100mb) and 15
negative biases in the free troposphere. RMS
errors are similar to RAQMSG with an additional
peak in the lower stratosphere (100mb).
15
Impact of MIPAS limb measurements on FvDAS
Assimimation1
(SBUVMIPAS) - SBUV assimilation December 2002
Adding MIPAS data results in large differences
(shaded) in the Tropics, below the ozone peak,
and in the polar night during December 2002.
MIPAS (Michelson Interferometer for Passive
Atmospheric Sounding) is a Fourier transform limb
viewing spectrometer measuring in the near to mid
infrared onboard the ESA ENVISAT.
1 Wargan et al., sub. to Q. J. R. Meteorol.
Soc., 2004
Contours assimilated ozone in December 2002,
using operational SBUV data
16
FvDAS Dec 2002 assimilation vs Sondes (Latgt45oN)
FvDAS December 2002 Assimilation vs Sondes
N59
Adding MIPAS 45oN, improves the agreement with
WMO sonde measurements poleward of 40oN. Such
MIPAS impacts would lead to improvements in the
SOS99 middle stratospheric FvDAS biases shown
earlier.
17
Regional AQ Forecast BC impact Study
Two regional (CONUS) 80km AQ forecasts for the
period from June 11- July 18, 1999 were conducted
using the nested component of RAQMS (RAQMSN) to
evaluate the impact of large-scale boundary
conditions (BC) during SOS99. The ASSIM-BC
forecast uses 3D chemical BC from the RAQMSG
assimilation. The CBC forecast uses CONUS area
averaged BC (on constant pressure surfaces) from
the RAQMSG assimilation. Both forecasts are
constrained with NOAA EDAS meteorological fields.
Tropospheric Ozone Column (TOC)
18
Surface O3 from EPA Network June 11 - July 18,
1999
8hr average gt85 ppbv corresponds to an AQI
unhealthy for sensitive groups
A regional high ozone event occurred over the
Midwest and Northeastern US during the SOS99 time
period (June and July, 1999). The RAQMSN 80km
CONUS grid is also shown by (). The interior
domain used for the BC evaluation is indicated by
( )
19
Comparison of ASSIM-BC Forecast with EPA Surface
Network June 11 - July 18, 1999
Correlations are generally high (gt0.5) over
Eastern US except for the extreme NE, Southern
Florida and along the Appalachian Mountains.
Largest RMS forecast errors are in urban areas of
California, NE due to inability of 80km forecast
to capture local variations in precursors.
Biases are generally within (/-) 10-15 ppbv
except along the Gulf Coast and the Central
Valley
20
Example of ASSIM-BC and CBC forecasts vs EPA
Surface Timeseries Cleveland, OH
The RAQMSN ASSIM-BC forecast for Cleveland, OH
captures the observed decline in O3 near the 5th
and 30th days of the forecast period better than
the CBC-BC forecast resulting in slight increases
in the correlation with surface measurements.
21
PDFs of RAQMSN vs EPA Statistics Interior sites
June 11- July 18, 1999
All local times
Mid-day (Zenithlt60)
ASSIM-BC forecast shows..
systematic increases1 in mid-day correlations and
decreases2 in mid-day RMS errors
but only small changes in overall correlations
and RMS errors,
only small changes in mid-day biases
EPA-RAQMS
EPA-RAQMS
but systematic increases3 in overall mean biases
...relative to the CBC forecast.
1 gt 25 increase in mid-day correlations for 27
of EPA sites. gt25 decrease in mid-day
correlations for 9 of EPA sites. 2 gt 5
decrease in mid-day rms errors for 22 of the EPA
sites. gt5 increase in mid-day rms errors for 12
of the EPA sites. 3 gt 5 decrease in overall
mean biases for 20 of the EPA sites. gt5
increase in overall mean biases for 44 of the
EPA sites.
22
Future directions...
23
Intercontinental Transport and Chemical
Transformation, 2004 Lagrangian Mission
(ITCT-Lagrangian-2K4)
NOAA New England Air Quality Study (NEAQS)
mission
EU Intercontinental Transport of Ozone and
Precursors (ITOP) North Atlantic Study
Coordinators David Parrish NOAA Aeronomy
LabKathy Law IPSL Service Aéronomie Goals To
investigate intercontinental transport of manmade
pollution and determine the chemical
transformation that occurs during this transport.
RAQMS will provide daily on-line global chemical
forecasts, initialized with assimilated ozone
distributions, to the NASA INTEX-NA science team
for mission flight planning.
NASA Intercontinental Chemical Transport
Experiment- North America (INTEX-NA) mission
24
Conclusions

• Assimilation of trajectory mapped solar
  occultation measurements reduces biases in the
  lower stratosphere/upper troposphere relative to
  assimilation of SBUV measurements.
• RAQMSN AQ forecast correlations and RMS errors
  are systematically improved relative to mid-day
  EPA surface O3 measurements during SOS99 when 3D
  large-scale BC are used to constrain the
  forecast.
• Knowledge gained from studies such as these can
  provide valuable guidance for the development of
  an operational chemical DAS and AQ forecasting
  system.
• The concurrent summer 2004 NEAQS and INTEX-NA
  missions provide an ideal opportunity for
  NASA/NOAA collaborative studies of the impact of
  large-scale BC on AQ forecasts.

Contact Information R. Bradley Pierce,
NASA/LaRC r.b.pierce_at_larc.nasa.gov
25
Extra Figures
26
RAQMS unified (strat/trop) chemistry
(55 species/families explicitly transported, 86
calculated, PCE assumptions for fast species)
37) C2H6 (ethane, 2C) 38) ALD2 (acetaldehydehighe
r group, 2C) 39) ETHOOH (ethyl hydrogen peroxide,
2C) 40) PAN (2C) 41) PAR (paraffin carbon bond
group, 1C) 42) ONIT (organic nitrate group,
1C) 43) AONE (acetone, 3C) 44) ROOH (C3hydrogen
peroxides group, 1C) 45) MGLY (methylglyoxal,
3C) 46) ETH (ethene, 2C) 47) XOLET (terminal
olefin carbon group, 2C) 48) XOLEI (internal
olefin carbon group, 2C) 49) XISOP (isoprene,
5C) 50) XISOPRD (isoprene oxidation product-long
lived, 5C) 51) PROP_PAR (propane paraffin,
1C) 52) CH3OH (methanol) 53) XMVK (methyl vinyl
ketone, 4C) 54) XMACR (methacrolein, 4C) 55)
XMPAN (peroxymethacryloyl nitrate, 4C)
StratosphereCH4CO oxidation
NMHC Chemistry
Chemical families OxO(1D)O(3P)O3NO2HNO32(NO3
)3(N2O5)HNO4PANMPAN NOyNNONO2NO32(N2O5)H
NO3HNO4BrNO3ClNO3PANONITMPAN ClyHClClONO2
ClO2(Cl2O2)OClOClO22(Cl2)BrClHOClCl BryHBr
BrONO2BrOBrClHOBrBr
27
RAQMS NMHC Treatment

• Explicit treatment of C2H6 (ethane), C2H4
  (ethene) and CH3OH (methanol) oxidation Sander
  et al., 2003. C3H8 (propane) is handled
  semi-explicitly.
• C4 and larger alkanes and C3 and larger alkenes
  are lumped via a carbon-bond approach Zaveri and
  Peters, 1999 which accounts for long-lived
  species and their intermediates based on the
  Carbon Bond Mechanism IV Gery et al., 1989.
• Isoprene is modeled after the Carter 4-product
  mechanism as modified for RADM2.

10-day diurnal equilibrium runs with/without NMHC
conducted as part of the NASA Global Modeling
Initiative (GMI) unified chemistry development.
GMI Harvard mechanism Bey et al., 2001 with
Gear solver for 80 species (all transported in
GMI)
LaRC Run Versions

• Standard
• Revised 1 Remove NO3 peroxy radical rxns
• Revised 2 Revised 1 ...

• Peroxide oxidation branching matched to GMI
• Organic nitrate production matched to GMI
• RO2 NO branching matched to GMI

28
NASA Earth Observing System (EOS) Aura Satellite

• Ozone Monitoring Instrument (OMI)
• hyperspectral (740 wavelength bands)imaging solar
  backscatter (visible and ultraviolet) radiometer
• Large swath large provides global coverage in 14
  orbits (1 day) at 13 x 24 km
• Tropospheric Emission Spectrometer (TES)
• high-resolution infrared-imaging Fourier
  transform spectrometer
• capability to make both limb and nadir
  observations.

• High Resolution Dynamics Limb Sounder (HIRDLS)
• limb scans in the vertical at multiple azimuth
  angles,
• measures infrared emissions in 21 channels
  ranging from 6.12 mm to 17.76 mm.
• Microwave Limb Sounder (MLS)
• 5 microwave channels (118, 190, 240, 640 GHz,
  and 2.5THz
• Ability to see through upper tropospheric clouds

Each of the Auras instrument team's Algorithm
Theoretical Basis Documents (ATBD) are now
available after having gone through peer review.
They can be found at http//eospso.gsfc.nasa.gov/
atbd/pg1.html
29
The GMAO ozone DAS is running in near-real-time
as a part of the GMAO's first and late look
assimilation system configurations in support of
the EOS Terra satellite. The GMAO ozone DAS uses
the Finite-Volume Data Assimilation System
(fvDAS) dynamical core and Physical-space
Statistical Analysis System (PSAS1) assimilation.
Operational DAO's ozone assimilation is currently
using NOAA SBUV/2 data. 1Cohn, S., A. da Silva,
J. Guo, M. Sienkiewicz, and D. Lamich. Assessing
the effects of data selection with DAO PSAS. Mon.
Wea. Rev., 1262913-26, 1998.
30
GOME NNORSY Sonde profile comparison (Latgt30oN)
P (mb)
Percent
O3 (ppbv)
GOME overestimates ozone in the planetary
boundary layer. Note 1999 WMO sonde profiles
and HALOE solar occultation measurements were
used in NNORSY training set.
31
TOC Probability Density Functions1
1Comparison of distributions removes time and
space coincidence criteria between the sonde and
assimilation/retrieval. Inferred RMS errors are
based on Monte Carlo estimates of random error
associated with KS significance.
32
Comparison with EPA surface measurements Huntsvill
e, AL
Online chemistry in the RAQMS assimilation
significantly improves representation of peak
amplitudes and diurnal variability in surface
ozone.
33
Example of RAQMSG ASSIM and RAQMSN ASSIM-BC Fx vs
EPA Surface Timeseries Huntsville, AL
The 80km RAQMSN ASSIM-BC forecast for Huntsville,
AL captures the observed decline in O3 near the
30th day of the forecast period better than the
RAQMSG 2x2.5o ASSIM but underestimates mid-day
peaks between days 5-10 resulting in a slight
decrease in the correlation with surface
measurements.