Performance evaluation of isoprene in ozone modeling of Houston Mark Estes, Clint Harper, Jim Smith, Weining Zhao, and Dick Karp Texas Commission on Environmental.
Post on 30-Dec-2015
243 Views
Preview:
Transcript
Performance evaluation of isoprene in ozone modeling of
HoustonMark Estes, Clint Harper, Jim Smith, Weining
Zhao, and Dick Karp
Texas Commission on Environmental Quality
Presentation for the CMAS Conference, October 2008
mestes@tceq.state.tx.us
Acknowledgements
• TCEQ Air Modeling Team: Doug Boyer, Pete Breitenbach, Bright Dornblaser, Barry Exum, Marvin Jones, Chris Kite, Jim MacKay, Jocelyn Mellberg, Ron Thomas, Zarena Post, Steve Davis.
• TCEQ Monitoring Operations
Questions of interest
• At the isoprene monitors used to evaluate performance, what is the long-term behavior? How does the model behave during similar time periods?
• How much geographic variation is observed in these patterns, and does the modeled variation match observed variation?
Houston isoprene 1997-2007Monthly median concentrations
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
pp
bv
Channelview
Danciger
Lynchburg Ferry
Methods: Biogenics modeling
• Emissions model: GloBEIS v3.1 biogenic emissions model (Yarwood et al., Guenther et al.) at 2km maximum resolution.
• Land cover data: University of Texas Center for Space Research (UT-CSR) land cover data (Feldman et al., 2007), 30 meter native resolution.
• Vegetation data: Houston Green Urban Forest Survey (Smith et al., 2005)
• Met data: interpolated temperature data for local networks, and GOES-derived photosynthetically active solar radiation data (Byun et al., 2005)
Methods: Photochemical grid modeling
• CAMx v4.51, run at 4km and flexi-nested to 2km. • MM5 v3.7.3, with ETA PBL scheme, UH GOES-
derived sea surface temperatures, UT-CSR land cover data, NOAH LSM, 4km maximum resolution, analysis nudging on outer grids, obs nudging with profiler data in 4km grid, TKE Kv scheme.
• Carbon Bond 05 chemical mechanism (Luecken et al., 2008).
• TCEQ emissions inventory, version bcYYMMM.reg8_pscfv2
Episodes of interest
• May 19 – June 3, 2005• June 17 – June 30, 2005• July 26 – August 8, 2005• May 31 – June 15, 2006• August 15 – September 15, 2006 (TexAQS II
field study intensive)• September 16 – October 11, 2006 (TexAQS II
field study intensive)• Total number of days of interest: 96
Example for Aug 16, 2006:Total isoprene emissions, 1924 tons/day for whole domain; 610 tons/day in Houston nonattainment area.
Channelview diurnal isoprene variation by month, for 2005-2006 auto-GC data
0.01
0.1
1
10
1 2 3 4 5 6 7 8 9 10 11 12
pp
bv
25th
median
75th
90th
95th
Suburban/ex-urban monitoring site
Channelview modeled diurnal profile of isoprene, 2005-2006 episodes (96 days)
0
0.5
1
1.5
2
2.5
3
0 6 12 18
Hour
pp
bv
mod25
modmd
mod75
mod90
mod95
Channelview measured diurnal profile of isoprene, May-Oct 2005-2006
0
0.5
1
1.5
2
2.5
3
0 6 12 18
Hour
pp
bv
msd25
msdmd
msd75
msd90
msd95
Danciger modeled diurnal profile of isoprene, 2005-2006 episodes
0
0.5
1
1.5
2
2.5
3
0 6 12 18
Hour
pp
bv
msd25
msdmd
msd75
msd90
msd95
Danciger measured diurnal profile of isoprene, May-Oct 2005-2006
0
0.5
1
1.5
2
2.5
3
0 6 12 18
Hour
pp
bv
msd25
msdmd
msd75
msd90
msd95
Southeast Texas ISOP emissions, 2005-2006 episodes
0
500
1000
1500
2000
250019
-May
26-M
ay
2-Ju
n
9-Ju
n
16-J
un
23-J
un
30-J
un
7-Ju
l
14-J
ul
21-J
ul
28-J
ul
4-A
ug
11-A
ug
18-A
ug
25-A
ug
1-S
ep
8-S
ep
15-S
ep
22-S
ep
29-S
ep
6-O
ct
ton
s/d
ay
Monthly median isoprene, measured at Houston area auto-gcs, 2005-2006
0
0.2
0.4
0.6
0.8
1
1.2
1.4
5 6 7 8 9 10Month
pp
bv
CCHS
CHVW
CLIN
DANC
DRPK
HRM3
LKJK
LYNF
MIPK
MUST
TXCT
WALV
Finding
• Temporal patterns: Modeled isoprene matches the diurnal and seasonal patterns of the measurements, but doesn’t always match the magnitude.
• Spatial patterns: Modeled isoprene appears to be correlated with the measured spatial patterns, but doesn’t always match the magnitude.
Geographic analyses
• Isoprene performance varies by site. Is the geographic distribution of trees correct?
Are the trees in the right places?
• Calculate the difference between the elevations estimated by the Shuttle Radar Tomography Mission (Feb 11-22, 2000) and the elevation of the ground surface using the National Elevation Database (USGS). The difference can represent the height of the tree canopy.
• Calculate the Normalized Difference Vegetation Index (a vegetative greeness index) for a Landsat image of approximately the same age (1999). This tells where the vegetation is located.
• Identify all areas with both high NDVI and height of 4 to 80 meters.• Calculate the number of “tree pixels” within each 4km grid cell.• Contrast the locations of these areas to the areas identified by the
UT-CSR landcover data as forested.• Plot isoprene emissions per grid cell vs. number of tree pixels per
grid cell.
Isoprene emissions vs satellite-identified treesExample for June 24, 2005
0
50
100
150
200
250
0 5000 10000 15000 20000
Number of 30m pixels identified as primarily trees per 4km grid cell
Em
iss
ion
s (
kg
pe
r 4
km
gri
d c
ell
)
Future work
• Further comparisons between modeled isoprene and TexAQS II observations (aircraft data, RHB ship data, Moody Tower data)
• C. Warneke analysis comparing PTRMS data aboard NOAA P3 aircraft to the biogenic emissions models GloBEIS, MEGAN, and the latest version of BEIS.
• Hyperspectral satellite data analysis to distinguish tree species? Aerial photography to assist in species identification?
top related