WRF_Protocol_May29_2013v2

Report
Meteorological Modeling Protocol for the
Three States Air Quality Study (3SAQS)
Ralph Morris and Bart Brashers
ENVIRON International Corporation
Zac Adelman and Aijun Xiu
University of North Carolina at Chapel Hill
Template
3SAQS Workshop -- CSU Fort Collins, CO
May 28, 2013
Purpose
• To perform prognostic meteorological modeling to
support the development of Photochemical Grid Model
(PGM) inputs
– PGM modeling will analyze the air quality and air quality
related values (AQRVs, which include visibility and sulfur and
nitrogen deposition) at key receptor areas across the western
U.S.
• Goal is for 3SAQS PGM modeling to start in fall 2013
• To achieve this schedule requires the meteorological
modeling process to start now
2
3SAQS Meteorological Modeling Leverages off
of Several Recent Studies
• Denver RAQC and CDPHE MM5 and WRF sensitivity
modeling (ENVIRON & Alpine, Jun 2011)
– Develop improved PGM modeling procedures for Denver area
– Bridge between MM5 and WRF
• WDEQ WRF modeling (Alpine & ENVIRON, Sep 2011)
– 2008 annual 36/12 km and winter episode fine-scale
• WestJumpAQMS 2008 36/12/4 km WRF modeling
(ENVIRON & Alpine, Feb 2012)
– Use as template for 3SAQS meteorological modeling
– Update to more recent model options and procedures
3
3SAQS Meteorological Modeling Steps
• Prepare Draft Modeling Protocol – Done
– (ENVIRON and UNC, May 2013)
• 3SAQS Cooperators Comment on Protocol
• Finalize Modeling Protocol (Jun 2013)
• Meteorological Modeling (Jun-Jul/Aug 2013)
• Meteorological Model Application/Evaluation
Report (Aug/Sep 2013)
• PGM Meteorological Inputs (Sep 2013)
4
3SAQS Meteorological Modeling Protocol
• Model Selection: WRF ARW
 Performing better than MM5 that is no longer supported
• Episode Selection: 2011
– Corresponds to NEI year and not atypical AQ/Met
• Domain Definition: 36/12/4 km
– 36 km CONUS Domain (RPO, WestJumpAQMS, etc.)
– 12 km WESTUS Domain (WestJumpAQMS)
– 4 km Domain – Need Feedback from 3SAQS:
 Option 1: WestJumpAQMS 12 km Inter-Mountain West
Processing Domain (IMWD)
 Option 2: Smaller 4 km domain tailored for 3SAQS
5
WestJumpAQMS 36/12/4 WRF Domain
Modeling Domains
6
Alternative 3SAQS 36/12/4 km WRF Domain
7
36/12/4 km WRF Domain Definitions
• RPO Lambert Conformal Projection
– Center @ (40°, -97°) & true latitudes of 33° and 45°
• 36 km CONUS Domain: ~165 x ~130
• 12 km WESTUS Domain: ~255 x ~255
• 4 km Domain Options:
– WestJumpAQMS IMWD: ~325 x ~525
– Alternative: ~250 x ~290 (~half the size of the IMWD)
8
3SAQS 4 km WRF Domain Definition?
• WJ WRF modeling w/ 4 km IMWD took 25,000
processing hours each 5-day segment using 8 cores
– Elapsed time of ~100 days using 80 cores
– Can 4 km domain focus on CO, UT and WY?
9
WRF Meteorological Model
WRF
Layer
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
Sigma
0.0000
0.0270
0.0600
0.1000
0.1500
0.2000
0.2500
0.3000
0.3500
0.4000
0.4500
0.5000
0.5500
0.6000
0.6400
0.6800
0.7200
0.7600
0.8000
0.8400
0.8700
0.8900
0.9100
0.9300
0.9400
0.9500
0.9600
0.9700
0.9800
0.9850
0.9880
0.9910
0.9930
0.9950
0.9970
0.9985
1.0000
Pressure
(mb)
50.00
75.65
107.00
145.00
192.50
240.00
287.50
335.00
382.50
430.00
477.50
525.00
572.50
620.00
658.00
696.00
734.00
772.00
810.00
848.00
876.50
895.50
914.50
933.50
943.00
952.50
962.00
971.50
981.00
985.75
988.60
991.45
993.35
995.25
997.15
998.58
1000
Height
(m)
19260
17205
15355
13630
11930
10541
9360
8328
7408
6576
5816
5115
4463
3854
3393
2954
2533
2130
1742
1369
1098
921
747
577
492
409
326
243
162
121
97
72
56
40
24
12
0
Thickness
(m)
2055
1850
1725
1701
1389
1181
1032
920
832
760
701
652
609
461
440
421
403
388
373
271
177
174
171
84
84
83
82
82
41
24
24
16
16
16
12
12
WRF Vertical Domain
• 37 vertical layers
• Surface to 50 mb (~19 km
•
•
•
above msl)
Layers 1 & 2 = 12 m thick each
Max layer thickness = 2,000 m
For PGM layer collapsing to 25
vertical layers?
– Used in WJ, Denver, Allegheny Cty
– Collapse 2 WRF layers in lowest 3
layers = 24, 32 and 40
– Max layer thickness = 3,900 m
10
3SAQS WRF Modeling Methodology
• WRF Version 3.5 released August 18, 2013
• Model 2011 plus 15 day spin-up in Dec 2010
• Run in 5-day run segments from cold start
– First 12 hours used to spin-up WRF
• First guess fields and lateral BCs from either:
– ERA-Interim Reanalysis from ECMWF; or
– North American Model (NAM) archives
 Summer/winter sensitivity tests
• USGS 24-category land use data
11
Four Dimensional Data Assimilation (FDDA)
• 3-D Analysis Nudging on 36 and 12 km domains
– Winds, Temperature and Mixing Ratio
– No Temperature and Mixing Ratio in the PBL
• Surface Observation Nudging in 4 km domain?
– Meteorological Assimilation Data Ingest System
(MADIS) observations
 Winds only
 NCDC surface observations
 NDBC overwater observations
12
WRF Physics Options
WRF Treatment
Microphysics
Longwave Radiation
Option Selected
Thompson scheme
RRTMG
Shortwave Radiation
RRTMG
Land Surface Model (LSM)
NOAH
Planetary Boundary Layer (PBL) scheme
YSU
Cumulus parameterization
Kain-Fritsch in the 36 km and 12 km
domains, with KF trigger option 2 or
3. None in the 4 km domain.
Nudging applied to winds,
temperature and moisture in the 36
km and 12 km domains
Nudging applied to surface wind only
in the 4 km domain
Analysis nudging
Observation Nudging
Initialization Dataset
ERA-Interim Daily reanalysis product
or 12 km NAM
Notes
New with WRF 3.1.
Rapid Radiative Transfer
Model for GCMs includes
random cloud overlap and
improved efficiency over
RRTM.
Same as above, but for
shortwave radiation.
Four-layer scheme with
vegetation and sub-grid tiling.
Yonsie University (Korea)
Asymmetric Convective
Model with non-local upward
mixing and local downward
mixing.
4 km can explicitly simulate
cumulus convection so
parameterization not needed.
Temperature and moisture
nudged above PBL only.
Surface temperature and
moisture observation nudging
can introduce instabilities.
WestJumpAQMS used 12 km
North American Model (NAM)
13
WRF Model Performance Evaluation
• Quantitative Evaluation
–
–
–
–
Surface Temperature, Mixing Ratio and Winds
Use METSTAT and AMET evaluation tools
MADIS surface observation database
36, 12 and 4 km domains, subregions, individual states
 WestJumpAQMS went down to individual monitor
• Qualitative Evaluation
– Comparison of spatial patterns of precipitation with
analysis fields based on observations (PRISM/CPC)
– Comparison of spatial distribution of clouds with
visual satellite observations
14
Model Performance Benchmarks
• Met Model Performance Benchmarks originally derived
after analysis of “good” MM5 performance to support air
quality modeling (Emery et al., 2001)
– Primarily ozone studies under simple (flat) terrain and simple
meteorological conditions (e.g., stationary high pressure)
 Sometimes sea breezes involved (e.g., Houston and Los Angeles)
• As part of WRAP, Kemball-Cook (2005) proposed
alternative benchmarks to account for more complex
conditions in the inter-mountain west and Alaska
– McNally (2009) also proposed alternative benchmarks for
complex terrain conditions
15
Model Performance Benchmarks
Simple (Emery et al., 2001) and Complex (Kemball-Cook et al.,
2005) Meteorological Model Performance Benchmarks
Parameter
Temperature Bias
Temperature Error
Mixing Ratio Bias
Mixing Ratio Error
Wind Speed Bias
Wind Speed RMSE
Wind Direction Bias
Wind Direction Error
Simple
Complex
≤ ±0.5 K
≤ ±2.0 K
≤ 2.0 K
≤ 3.5 K
≤ ±1.0 g/kg
NA
≤ 2.0 g/kg
NA
≤ ±0.5 m/s
≤ ±1.5 m/s
≤ 2.0 m/s
≤ 2.5 m/s
≤ ±10 degrees
NA
≤ 30 degrees ≤ 55 degrees
16
Soccer Plot Example Quantitative Evaluation
• Temperature Bias vs.
Error
– 36 km CONUS, 12 km
WESTUS & 4 km IMWD
– Compared against Simple
and Complex Benchmarks
– From WestJumpAQMS
17
Example Qualitative Precipitation Evaluation
Jan (top) and Jul (bottom ) CPC analysis fields (left) vs. WRF (right)




18
3SAQS WRF Modeling Next Steps
• Decide on 4 km domain
• Comments on Draft Modeling Protocol
– By Monday June 10, 2013
• Start setting up WRF for 2011 modeling
• Limited sensitivity tests
– ERA ECMWF vs. NAM IC/BC
– Other?
• Need to initiate now to keep on schedule
19

similar documents