3SAQS_TechCommMtg_2011_WRF_MPE_28July2014

advertisement
Three-State Air Quality Study (3SAQS)
Three-State Data Warehouse (3SDW)
2011 WRF Modeling
Model Performance Evaluation
University of North Carolina (UNC-IE)
Cooperative Institute for Research in the Atmosphere (CIRA)
ENVIRON International Corporation (ENVIRON)
July 28, 2014
3SAQS WRF MPE Summary
• Objective: Develop best possible regional meteorology
database for annual 2011 air quality modeling
• Nested 36/12/4-km WRF-ARW 3.5.1 with configuration
optimized for the intermountain West
• Single configuration across the entire year
– no special wintertime version at this time
• Standard and extended model performance evaluation
metrics focused on the 4 km three-state domain
• MPE conclusion: 3SAQS 2011 WRF application exhibited
reasonably good model performance that was as good
or better than other recent prognostic model
applications used in air quality planning
2
3SAQS Draft WRF 2011 MPE Report
• Draft dated July 2014
Executive Summary
1.0 Introduction
2.0 Methodology
3.0 WRF Surface Model
Performance Evaluation
Results
4.0 Precipitation Evaluation
5.0 Evaluation for Winter
High Ozone Periods
3SAQS Tech Committee Call June 19, 2014
• Summarized WRF surface and winter ozone
model performance during June 19, 2014
3SAQS Technical Committee webinar
• So will only highlight some of these results
here with conclusions
• In this presentation focus on new precipitation
comparisons and overall conclusions
4
WRF Domain and Configuration
• Standard (40N-97W)
Lambert-Conformal
Conic projection
• CONUS 36km (165x129)
• WestJumpAQMS 12km
(256x253)
• 3SAQS 4km (301x361)
• 37 layers to 50 mb (12m
surface layer)
Configuration Tests
• First-order simulations on all three domains for
January 1-7 and July 1-7, 2011 to test different
configurations:
– NAM vs. ECMWF Initial and boundary conditions (ICBC)
– 2006 NCLD vs. USGS land-use land cover (LULC) data
– Noah-YSU vs. PX-ACM2 land surface/boundary layer
schemes
– Obs nudging coefficient and network sensitivities
• Ultimately settled on the same configuration as
WestJumpAQMS 2008 WRF
6
Evaluation Approach
• AMET used to evaluate surface temperatures, winds,
and mixing ratios against MADIS observations
• Wintertime evaluation of winds, temperature,
humidity and vertical temperature (inversions)
• PRISM precipitation evaluations
• Compared results to performance benchmarks (bias
and error) for Western U.S. regional met modeling
• Performance in CO, UT, WY, and 4-km 3S domain
– average of all monitors within each area
• Additional performance for snow cover, STE, planned
in future
7
WRF Monthly Performance Summaries
2-m Temperature
CO
WY
WRF Monthly Performance Summaries
Wind Speed
CO
WY
WRF Monthly Performance Summaries
Wind Direction
CO
WY
10
WRF Monthly Performance Summaries
Mixing Ratio
CO
WY
Example Summary MPE Plot
Jan
Temperature in 4 km 3SAQS Domain
Jul
12
2011 WRF MPE Surface Met
• Almost always achieved Complex Performance
Benchmark and Frequently achieved Simple
Benchmarks
• Performance as good or better than past
WRF/MM5 applications in Rocky Mountains
• No systematic bias or problem performance seen.
• 3SAQS 4 and 12 km MPE products can be viewed
at:
http://views.cira.colostate.edu/tsdw/Tools/ImageBr
owser.aspx?pathid=TsdwBase11aMetPlotsRoot
13
3SAQS 2011 WRF MPE Winter Ozone
• 2011 WRF simulation included several high
winter ozone events in Upper Green River
Basin (URGRB) in southwest Wyoming during
the Upper Green River Ozone Study (UGWOS)
• 13 ozone exceedance days (76 ppb or higher)
occurred during February and March 2011
– 123 ppb on Mar 2, 2011
– 121 ppb on Mar 3, 2011
– 121 ppb on Mar 12, 2011
14
UGRWOS
Monitoring
Sites
2011 Winter Ozone WRF Evaluation
• Use UGWOS special
study and routine NOAA
ds3505 surface met obs
in SWWY
• Soccer plots for Feb and
Mar and SWWY domain
• Vertical temperature
profiles
• 3SAQS 2011 WRF run
not configured for
winter ozone conditions
WRF Wind MPE UGRWOS & ds3505
Mar 1, 2011 1500 (Max O3 =121 ppb)
Mar 2, 2011 1500 (Max O3 =123 ppb)
Mar 10, 2011 1500 (Max O3 = 84 ppb)
Mar 12, 2011 1500 (Max O3 = 121ppb)
Winter Ozone WRF Conclusions
• Need more focused evaluation for high ozone
days
• Some promise with low bias for winds and
temperature
– Large error not unexpected given slow wind speed
• Vertical temperature matched reasonable well
– Worse near the surface
• 3SAQS 2011 WRF 36/12/4 km can provide
starting point for focused higher resolution WRF
model of winter ozone episodes using winter
ozone WRF configuration in 3SAQS 2015 SOW
24
Precipitation Evaluation Approach
• Monthly total precipitation from PRISM analysis fields
– Parameter-elevation Regressions on Independent Slopes Model
– Interpolates precipitation observations using regression
equations to account for elevation and other factors
– Regrid ~1 km PRISM data to WRF domains
– Note: PRISM analyses fields only for lower 48 USA
• Canada, Mexico and Oceans not included
• Qualitative comparison of monthly WRF and PRISM
precipitation
– PRISM also has daily products
– Plans to develop quantitative PRISM evaluation tool in future
25
January Monthly Precipitation (inches)
PRISM
WRF
26
March Monthly Precipitation (inches)
PRISM
WRF
27
May Monthly Precipitation (inches)
PRISM
WRF
28
July Monthly Precipitation (inches)
PRISM
WRF
29
August Monthly Precipitation (inches)
PRISM
WRF
30
September Monthly Precipitation (in)
PRISM
WRF
31
October Monthly Precipitation (in)
PRISM
WRF
32
December Monthly Precipitation (in)
PRISM
WRF
33
3SAQS 2011 WRF Precipitation MPE
• Very good agreement between PRISM and WRF
monthly precipitation spatial fields and
magnitudes during winter and adjacent months
– WRF able to simulate synoptic storm systems well
• PRISM/WRF agreement during monthly total
precipitation in summer months not as good
– Convective precipitation has been historically difficult
for WRF/MM5 to simulate
– Questions whether PRISM interpolation scheme
appropriate for summer thunderstorms
34
3SAQS 2011 WRF Conclusions
• WRF surface meteorological performance consistent achieves
model performance benchmark across CO, UT and WUY and 4
km domain
– Some indication model has difficulty simulating nocturnal inversions in
complex terrain
– Appears to be a slow wind speed bias of ~-0.5 m/s year-round
– Mixing ratio overestimated in cooler and underestimated in warmer
months
• Monthly precipitation performance in winter and adjacent
months quite good
– Maybe some issues with summer convective precipitation
• Overall, 3SAQS 2011 WRF performance was reasonably good
and as good or better than the performance of other
WRF/MM5 applications in the Rocky Mountain region
35
3SAQS 2011 WRF Next Steps
• Update draft 3SAQS 2011 WRF MPE report
– Over next week or two
• Focused high-resolution WRF sensitivity
modeling for winter ozone events (2015 SOW)
– Leverage work of University of Utah and others
– Snow cover and cloud evaluation
• In addition to surface, upper-air and precipitation
• Stratospheric/Tropospheric Exchange (STE)
– Need PGM modeling results to evaluate WRF
36
Download
Related flashcards
Create Flashcards