MESOSCALE MODELING FOR

1
MESOSCALE MODELING FOR AIR QUALITY
FORECASTING by Robert D. Bornstein DEPT. OF
METEOROLOGY SAN JOSE STATE UNIVERSITY SAN JOSE,
CA USA pblmodel_at_hotmail.com Prepared for CLEAN
AIR CONGRESS LONDON, UK 2004
2

• ACKNOWLEDGEMENTS
• CO WORKERS
• T. Ghidey, LBNL, SJSU
• H. Taha, ALTOSTRATUS, SJSU
• D. Boucouvala and R. Balmori, SJSU
• J Ching and S. Dupont, EPS/RTP
• S. Burian, Univ of Utah
• S. Stetson, SWS, Inc.
• D. Byan, Univ of Houston
• FUNDING AGENCIES
• USAID
• State of Texas
• CARB
• LBNL
• DHS

3
OUTLINE

• IMPORTANT MODELING-FACTORS
• CURRENT MM5 CONFIGURATION
• CASE STUDIES
• SFBA WINTER STORMS (LOZEJ 2000)
• ATLANTA THUNDERSTORMS (CRAIG 2002)
• LOS ANGELES 03 (Atmos Environ 2003a,b)
• SFBA O3 (this paper)
• HOUSTON UHI AND O3 (this paper)
• Mid-East O3 (other papers here)
• NYC ER (just starting)
• CONCLUSIONS

4
KEY IDEA IDEAL MESO ATM MODEL CAPTURES ALL BC
FORCINGS IN CORRECT ORDER

• O3 EPISODES OCCUR ON A GIVEN DAY
• NOT B/C TOPO, EMISSIONS, OR SFC MESO-FORCING
  (EXCEPT FOR FOG) CHANGES
• BUT DUE TO CHANGES IN UPPER-LEVEL SYNOPTIC WX
  PATTERNS, WHICH COME FROM AN EXTERNAL MODEL
  WHICH ALTER MESO SFC-FORCINGS (I.E., TOPO,
  LAND/SEA, URBAN) VIA MESO-TEMP AND THUS WIND
• MUST THUS EVALUATE ABOVE FACTORS
• SYN WX Patterns (upper pressure and hence winds)
• TOPO (via grid spacing) channeling
• MESO SFC-Temp (and hence winds)

5
(No Transcript)
6
Urbanization Techniques

• Urbanize surface, SBL, PBL (momentum,
• thermo, TKE) Eqs
• Allows prediction within UCL
• From veg-canopy model (Yamada 1982)
• Veg param replaced with urban (GIS/RS) ones
• Brown and Williams 1998
• Masson 2000
• Sievers 2001
• Martilli et al. 2001 (in TVM)
• Dupont et al. 2003 (in MM5)

7
? From Masson (2000)
8
(No Transcript)
9
Current MM5 Configuration

• EPA urbanized uMM5xx with up to five domains
• Down to a 1 km grid
• Up to 190 x 190 grid points in inner domain
• GIS/RS surface conditions
• Five-layer soil-model, Gayno-Seaman PBL TKE
• Simulations up to 8 days
• Up to 96 CPUs on SJSU 106 CPU cluster
• 1 CPU is 15 to 1, but 96 CPUs are 0.25 to 1
• Up to 48 sigma levels
• up to 100 mb
• first half-sigma level down to 10 m
• GDAS or Eta IC and BC
• Analysis nudging only V and T above PBL in outer
  domains
• Obs nudging only in inner domains

10

• NORMAL SYNOPTIC WIND CONVENTION
• FULL BARB 1O M/S
• FLAG 5O M/S
• CURRENT MESO WIND CONVENTION
• FULL BARB 1 M/S
• FLAG 5 M/S

11
New GIS/RS inputs for uMM5 as f (x, y, z)

• land use (38 categories)
• roughness elements
• anthropogenic heat as f (t)
• vegetation and building heights
• paved surface fractions
• drag-force coefficients for buildings
  vegetation
• building height-to-width, wall-plan,
  impervious-
• area ratios
• building frontal, building plan, and rooftop
  area-
• densities
• wall and roof e, c?, a, etc.
• vegetation canopies, root zones, stomatal
  resistances

12
SFBA CCOS O3 CASE STUDY

• Movement of inland NWS 700 hPa H causes O3
  episodes
• Pre-episode over Nevada
• Episode moves SW to SJV intensifies
• Post-episode dissipates

13

• Warm-core NWS upper-H projects down to become
• a sfc inverted thermal-L
• Pre-episode it is over Nevada
• Episode days it moves over SJV intensifies
• Post-episode it weakens

14
MM5 Domain-1 700 hPa WINDS

• NWS charts give only approx pressure-center
  locations and cannot give flow-details in SFBA
• Thus need uMM5xx

15

(B/F episodes) MM5 correct offshore GC H Syn H
in SE Nev boundary-L but S-flow over SFBA vs.
SW NWS
L
H
H
16
SAC episode day Syn H to Utah with max (now
NE-SW) bulge (vs. NWS-H over SJV) L now S in
Cal (vs. NWS E over CA/Az border) correct SW
flow over SFBA to Sac
H
H
L
17
(No Transcript)
18
MM5 DOMAIN-3 700 hPa WINDS

• 30 July Offshore-H formation causes LIV episode,
  but not SFBA to Sac transport
• 31 July Coastal-H directs SFBA flow to Sac
  (pro-ducing SAC episode), but Fresno eddy blocks
  SFBA flow into SJV
• 1 Aug Eddy moves to N and becomes Sac eddy
• SFBA flow into Sac is again blocked
• SFBA flow into SJV is now allowed (producing SJV
  episode)

19
Pre-episode uniform S-flow
20
SJV episode Fresno eddy moved N H moves inland
(both better defined than in D-1) flow around
eddy blocks SFBA flow to SAC, but forces it S
into SJV
L
H
21
DOMAIN-3 SFC TRANSPORT TO LIV

• Daytime-confluence of three-flows E of LIV (on Mt
  peak) causes episode
• Flow from N from Carquinez Straits
• Flow from W thru GGG
• Upslope flow from E-side of hills E of LIV
• For LIV episode need
• Strong confluence
• Low speeds

22
Sfc obs at 0700 PST 31 July (LIV episode
morning) Note confluent flow into LIV
23
Episode afternoon (1400 PDT) W flow thru GGG
strong con into E-Liv
24
DOMAIN 3 SFC TRANSPORT TO SAC

• TRANSPORT FROM SFBA BLOCKED BY OPPOSING FLOW ON
  DAYS BEFORE AND AFTER SAC EPISODE
• TRANSPORT FROM SFBA NOT BLOCKED ON DAY OF SAC
  EPISODE

25
LIV-episode late-afternoon (1800 PDT) flow to
SAC from SFBA blocked
26
SAC-episode late-afternoon (1800 PDT) flow to
SAC from SFBA not blocked
27
(No Transcript)
28
CONCLUSIONS

• SMALL CHANGES IN 700 MB-H AND SFC-L POSITIONS
  CAUSED SEQUENTIAL DAILY OZONE MAX IN LIV, SAC,
  AND SJV
• NWS CHARTS COULD NOT DISCERN DETAILS OF CHANGES,
  BUT MM5 COULD
• MM5
• MATCHED NWS PATTERNS AND MESO-OBS REASONABLY WELL
  
• PROVIDED ADDITIONAL DETAILS
• ADDITIONAL SIMULATIONS SHOULD CORRECT REMAINING
  DIFFERENCES
• Lower deep soil BC ? lower nighttime temp
• Increase soil mositure ? less heat-flux into atm
  ? lower daytime temp

29
HOUSTON O3 CASE STUDY

• Movement of cold-core coastal L causes 03
  episodes
• Pre-episode it forms NE of Houston
• Episode days it moves off-shore of Houston
• and intensifies
• Post-episode it moves SW of Houston

30
Domain 4 (3 PM) Low off of Houston on O3 day
(Aug 25)
31
12 km MM5 4 PM Coastal cold-core L
32
Domain 4 weak sea breeze meets inland NE-flow

4th (MM5) 5th (uMM5) Domains Non-episode day
at 2 PM
Domain 5 Finer details of sea breeze and NE-flow
33
1 km uMM5 11 PM nocturnal 2.5 K UHI
34

1 km uMM5 3 PM daytime 3 K UHI
35
1 km uMM5 end of daytime ?UHI 8 PM 21 Aug

• Upper L MM5
• Upper R uMM5
• Lower L uMM5-MM5
• uMM5? 1.5 K warmer
• Blob is LU/LC error

36
1km uMM5 End of night ?UHI 9 AM 22 Aug

• Upper L MM5
• Upper R uMM5
• Lower L uMM5-MM5
• uMM5? 1.5 K cooler

37
Explanation of UHI changes due to Thermal-Inertia
(TI) differences

• Wet rural-soil TIgt urban TI gt dry rural-soil TI
• Urban area surrounded by wet-soil thus has
• Daytime UHI (as urban area warms faster than
  soil)
• Nighttime UCI (as urban area cools faster than
  soil)
• Reverse true with dry rural soil
• Current results thus consistent with wet rural
  soil (as expected) around Houston, as uMM5
  produced daytime warming nighttime cooling over
  urban Houston

38
Ongoing and Planned efforts

• Houston simulations with 96 CPUs ? speedup factor
  of 60
• Use Stetson Houston GIS/RS urban sfc parameters
  to test
• de- and re-forestation plans?
• UHI changes ? O3 SIP simulations ?
• emission reduction credits
• Real-time NYC emergency-response system
• SIT ocean model ? SST (x, y, t)
• uMM5xx
• CFD street canyon model
• Street canyon dispersion model

39
Incorporate Stetsons high- resolution zo data
40
Houston Sensitivity Study UHI effects of re-
and deforestation
Planned simulation Prescriptive increases in
urban vegetation fraction (urban-max increase)
Current simulation Base-case vegetation
fraction (urban-min)
41
Conclusions

• SMALL CHANGES IN SFC COASTAL-L POSITION CAUSED
  HOUSTON OZONE MAX, AS LOW WAS S OF CITY AND THE
  NE FLOW ON ITS N-SIDE CAUSED CONFLUENT FLOW WITH
  SEA BREEZE
• NWS CHARTS COULD NOT DISCERN DETAILS OF CHANGES,
  BUT MM5 COULD
• MM5
• MATCHED NWS PATTERNS AND MESO-OBS REASONABLY WELL
  
• PROVIDED ADDITIONAL DETAILS

42
UAO SUMMARY

• follows US and international urban dispersion
  studies (OKC, DAPPLE, SLC, etc.) and extends them
  to deep urban-canyons.
• combines science and model development with
  long-term real-time support of NYC EM.
• will encompass obs, real-time data collection
  assimi-lation, and model development on different
  scales.
• pilot study has shown feasibility of long-term
  urban canyon flux obs, rooftop SODAR, and
  wireless tech.
• 2004 MSG intensive tracer study begins series
  that focuses on rapid vertical-dispersion over
  tall buildings.
• development of accurate rapid-response models
  with high-resolution data-ingest is long-term
  goal.

43
UAO
REGIONAL MESONET CENTRAL TESTBED
INTENSIVE STUDIES Continuous facility
Permanent communications backbone Integrated
into NYC Office of Emergency Management (OEM)
44
(No Transcript)
45
MESOSCALE-NETWORK FOR URBAN-MODELING WITH
DATA-ASSIMILATION
The Challenge Assimilate existing and new
observations into a predictive urban mesoscale
model and provide high-resolution (100 m),
one-hour forecasts over the urban region.
46
(No Transcript)
47
(No Transcript)
48
(No Transcript)
49
(No Transcript)
50
(No Transcript)
51
(No Transcript)
52
(No Transcript)
53
QUIC Simulation with dd 215 eg
wind vectors at 5 m height
54
(No Transcript)