Improvement in modeled ozone source contributions via
sequential assimilation of Aura satellite measurements into global
and regional models Min Huang ([email protected]), K. W.
Bowman (JPL), G. R. Carmichael (U Iowa), M. Lee (JPL), D. K. Henze
(CU-Boulder), T. Chai (NOAA/ARL) A. J. Weinheimer (NCAR), R. C.
Cohen (UC Berkeley) Acknowledgements: *NASA funding (Aura, ARCTAS,
AQAST) *ARCTAS science teams *The TES team at JPL *NASA Ames and U
Iowa supercomputers AQAST 6 meeting | Houston, TX | Jan 16, 2014
2013. All rights reserved.
Slide 2
Ozone variability in the Western US is affected by various
sources: We seek a means to reduce the uncertainty in the estimates
of their contributions 1. Background 2. Objectives/Methods 3. Assim
TES 4. Assim OMI 5. Conclusions Why challenging to model ozone
& US vs. non-US source contributions? -Anthropogenic emissions
trends (extra-regional going up, US dropping) not well represented
in the inventories -Impacts of some natural sources (e.g., biomass
burning) can be episodically strong and their emissions can be
highly uncertain -Uncertainties in model transport, chemistry,
deposition, etc Liang et al., 2004 trans-boundary pollution
transport local emissions
Slide 3
This study: Improve modeled total ozone and its partitioning by
sequentially assimilating Aura measurements into global and
regional chemical transport models *WRF v3.5 meteorology *CARB
emissions (all sectors, daily varying, Mar 2013) adjoint v34, 2x2.5
TES L2 ozone profiles OMI NO 2 tropospheric columns (KNMI) 1.
Background 2. Objectives/Methods 3. Assim TES 4. Assim OMI 5.
Conclusions 3D Var 4D Var top/lateral boundary conditions Surface
monitors Sondes Aircraft (DC-8) STEM 12 km By integrating Aura
measurements into a multi-scale modeling system, we aim to improve
the estimated impacts on ozone in California from: 1)
trans-boundary pollutants; and 2) US ozone precursor (i.e., NOx)s
emissions
Slide 4
Extra-regional pollutants were mixed with local urban pollution
and strong fire emissions. Observed strong variability in free
troposphere; high ozone and expanded exceedances areas near the
surface. Can observations improve model-estimated ozone source
contributions? The case we study: Northern California-Central
Valley during June 15-30, 2008 (ARCTAS-CARB field campaign)
Trinidad Head sondes Period-mean daily max 8h average ozone Singh
et al., 2012 1. Background 2. Objectives/Methods 3. Assim TES 4.
Assim OMI 5. Conclusions Ozone along DC-8 (
Improvement of ~6-10 ppb for the entire period Improvement in
boundary conditions: GEOS-Chem w/ constraints from TES ozone: cross
validation with ozonesondes at Trinidad Head, CA 1. Background 2.
Objectives/Methods 3. Assim TES 4. Assim OMI 5. Conclusions After-
before, 16 days (w/ TES samples) After assimilation, Jun 22-25
episode Airmasses at ~1.5-3 km offshore can impact inland surface
ozone later (Huang et al., 2010) Improvement up to >~20 ppb
during strong transport event
Slide 6
Impact of boundary conditions (GEOS-Chem before/after
assimilation) on STEM ozone: Change in absolute mixing ratios near
the surface 1. Background 2. Objectives/Methods 3. Assim TES 4.
Assim OMI 5. Conclusions Period-mean near-surface (
1. Background 2. Objectives/Methods 3. Assim TES 4. Assim OMI
5. Conclusions Mean: -0.16 Mean: -0.08 Mean: 0.04 Mean: 0.10 Impact
of boundary conditions (GEOS-Chem before/after assimilation) on
STEM ozone: Fractional bias (unitless): 2x(model-obs)/(model+obs)
Period-mean surface (AQS & CASTNET) daily-max 8h average ozone
Period-mean near-surface (< 2 km a.g.l.) ozone along DC-8
flights Negative biases in northern Sacramento Valley dropped by
>0.2 Positive biases in the Bay area increased by >0.2 Before
After
Slide 8
STEM NO 2 : Model-Obs along DC-8 flights in ppb Impact of
assimilating OMI NO 2 columns in STEM (before/after assimilation)
on near-surface (< 2 km a.g.l.) NO 2 along DC-8 flights 1.
Background 2. Objectives/Methods 3. Assim TES 4. Assim OMI 5.
Conclusions Assimilation of OMI reduced NO 2 mixing ratios along
DC-8 by ~10%, and the error by ~8%, mostly occurred in the valley
and the Bay area 10 9 kg/s/m 2 Before After Assimilation of OMI
reduced NOx emissions by ~1.7%: -(5-20)% urban areas; >+80% at
some fire locations
Slide 9
1. Background 2. Objectives/Methods 3. Assim TES 4. Assim OMI
5. Conclusions Impact of assimilating OMI NO 2 columns on STEM
ozone: Change in absolute mixing ratios near the surface
Contributions from US emissions on modeled ozone decreased by up to
~5 ppb after the assimilation. This drop in ozone compensated the
increases resulted from assimilating TES in GEOS-Chem - We
repartitioned the ozone contributions from local and non-local
sources. Period-mean near-surface (
1. Background 2. Objectives/Methods 3. Assim TES 4. Assim OMI
5. Conclusions Conclusions & Future Work Future work:
-additional constraints (VOCs, CO, etc) to distinguish the natural
and anthropogenic sources -improving assimilation settings
-extended studied period *We demonstrated a prototype multi-scale
assimilation system that integrates satellite observations across
multiple chemical species to assess the role of non-local and local
sources of ozone. *This system updates the estimates of
contributions from local and non- local sources to ozone, and the
local NOx sector emissions. 1) Assimilation of TES and OMI in two
steps enhanced transported background ozone by ~4 ppb; reduced the
US emission contributions by ~2 ppb 2) Assimilation of OMI
repartitioned US NOx emissions reduced anthropogenic emissions by
5-20%, increased biomass burning emissions by >80%.
Slide 12
Based on monthly-median TES profiles (filtered by quality flag
and c-curve) in region within lat lon (approx): 26-37N | 94-106W
Dots: Sep data Pres: hPa Sep data TES-observed tropospheric ozone
trend/variability in Texas in 2005-2013 What drove the strong
inter-annual variability? any relavance to: - climate - change in
source contributions - TES sampling strategies and retrieval
algorithm
Slide 13
Observations No assim Assim TES Assim TES&OMI Backup
Statistically, assimilation of TES&OMI case resulted in similar
total ozone as the no-assimilation case, while improved the high
ozone scenarios/areas and exceedances