Status of the NOAA WRF-CMAQ Regional Air Quality Forecast System

Status of the NOAA WRF-CMAQ Regional Air Quality Forecast System

Jeff McQueen, Pius Lee, Marina Tsildulko, Geoff DiMegoSarah Lu and Bert Katz

NOAA/NCEPEnvironmental Modeling Center

Rohit Mathur, Hsin-Mu Lin, Daiwen Kang, George PouliotU.S. EPA/ORD/ASMD and NOAA/OAR/ARL

2006 Developments• WRF Transition

– Improved vertical coupling– Continued convective mixing & vertical layering tests– Continued lateral boundary condition tests

• CMAQ Improvements– Optimized Advection scheme– Emissions upgraded for 2006

• AQF system retrospective & Real-time testing– July 2005 with experimental CONUS configuration

• Real-time Verification– Spatial map comparisons to observations (03 & PBL hgt)– Inclusion of AIRNOW PM 2.5 observations– Use of ACARS, ESRL Profilers for PBL hgt verification– Inclusion of NESDIS GASP AOD products

• Improved Analyses & Visualizations– Focus group, TEXAQS06 & SHENAIR projects– Vertical gas-phase and aerosol chemistry profiles

NAM-WRF Implementation

• Major Components of Upgrade Package for NAM– Replace Eta Model with WRF version of NMM

(Nonhydrostatic Mesoscale Model)• WRF Common Modeling Infrastructure• Non-hydrostatic dynamics• Use of hybrid sigma-pressure vertical coordinate with top at 2 mb• Refined advection, diffusion, numerics and physics

– Replace Eta 3D-Var analysis with Gridpoint Statistical Interpolation (GSI) analysis

• Unified (regional + global) 3D-Variational analysis adapted to WRF• Begin use of background errors based on WRF-NMM to 2 mb• Use of new variable for moisture analysis• Use of tendency in constraint terms• Use of dynamically retuned observational errors


• Model initialization– Begin use of:

• high resolution (1/12th degree) SST• high resolution (4 km) snow

– Common specification of terrain, land-sea mask

• Data assimilation changes– Use of bias-corrected observed precipitation analysis

values in land-surface physics – Start assimilating

• WSR-88D Level II radial wind data• GPS-Integrated Precipitable Water (IPW)• NOAA-18 radiances

– Drop use of • GOES Precipitable Water retrievals• SSM/I Total Precipitable Water retrieval

Hybrid vs Step (Eta) Vertical Coordinate Systems

ground MSL

ground

Pressure domain

Sigma domain

δ = 0

δ= 1 η = 1

Ptop Ptopη= 0

HYBRID Step (Eta)

268 grid cells

259gridcells

East “3x” Domain

Forecast Domains (2006)

CONUS “5x” Domain1. WRF-CMAQ2. WRF-CMAQ-PM

2006 SystemsWRF-CMAQ 48 h forecasts

System Domain Vertical coupling

Convective mixing

LBCs

Operational

Prod: 06/12Z

32p, 45 min

Eastern U.S. (3X)

22L Loose: interp from WRF

Walcek (1980) RADM

GFS ozone at model top: 100mb

Experimental Para: 06/12Z

65p, 70 min

CONUS (5X) 22L Tight: Common hybrid

Asymetric Convective Model (ACM-Pleim)

Clean, static profiles

Developmental

Blue: 12Z

127p, 150 min

CONUS w/ PM (5X)

22L Tight:

Common hybrid

ACM-Pleim Clean, static profiles

Operational Physics Coupling

Current

Capability

Met Model (WRF/NMM)

AQ Model (CMAQ)

Core/Dynamics

Rotated Lat-Lon E Hybrid sigma-P

Arakawa C Grid Sigma-P

Cloud micro-physics

Full Ferrier Cloud Microphysics

NAM cloud water for aqueous chemistry

Convective mixing

Betts-Miller Janjic RADM: Entrainment from top turned off

Radiation Lacis-Hansen SW & LW

CMAQ J Tables

PBL Mellor-Yamada TKE NAM PBL hgt for Pleim-Xiu 1st order closure PBL

Land Surface NOAH common NAM canopy conductance terms for Pleim-Xiu LSM

Experimental CONUS Coupling

Run Summer 2006 PlannedPlannedDomain Interp to CMAQ Secant

Lambert-Conf. C gridCommon Rotated Lat-Lon E grid

Vertical Coordinate Common NMM Hybrid coord Common NMM Hybrid coord.

Improved vertical resolution

Photolysis NAM Sfc Radiation for Photolosis Scaling

3-D Radiative fluxes

PBL NAM PBL height into P-X NAM TKE/Kh to drive mixing

Clouds Aqueous

Mixing

NAM cloud water

Asymetric Convective Model (ACM) mixing

NAM cloud water,graupel & ice

GFS-RAS mixing into CMAQ

LBCs Clean, Static profiles GFS in strat, static below

PM from GFS-GOCART

0

100

200

300

400

500

600

700

800

900

1000

1100

Improved Vertical Coupling(Hsin-Mu Lin, EPA)

http://virtualskies.arc.nasa.gov/weather/tutorial/tutorial2d.html NAM 60L CMAQ 22L

Summer 2006 Evaluations(Errors not uncovered w/retro tests)

• Operation Runs: Ozone over-predictions esp. under cool, cloudy conditions

– WRF-Post temperature interpolation to CMAQ sigma errors

• Sigma level 1 and 2 temps were too warm for thermally stable conditions (near lakes, water bodies)

• STATUS: Crisis Fix implemented on July 28. 2006– Small impact on forecast Slightly less ozone

produced

– PREMAQ Emission Processing• Some arrays for mobile, point, area, biogenic sources were

undefined• Some point and mobile source emission files not upgraded

for 2006

• STATUS: Implemented 18Z August 22, 2006– Preliminary tests show minor impact

Impact of WRF temperature interpolation error on Oper CMAQ

Summer 2006 Evaluations(Errors not uncovered w/retro tests)

• Experimental/Developmental Runs: Significant under-prediction esp. near coasts, lakes

– PREMAQ Biogenic Isoprene emissions • Turned off for tight vertical coupled codes

• STATUS: Implemented in exper/dev runs on Aug. 5– Major impact on forecast Increased photochemistry esp. in

California

– PREMAQ Emission Files and Processing• Some arrays for mobile, point, area, biogenic sources were undefined• Some point and mobile source emission files not upgraded for 2006

• STATUS: Corrected in exper/dev run on August 5– Preliminary tests show minor impact

– PREMAQ Jacobian Calculations for vertical diffusivity• Status: Should be implemented at 12Z on Sept. 5, 2006

July 17,06 Ozone event (06Z forecast)

1h daily max valid 06z 7/18/06

3X loose coupled5X tight coupled

5X forecasts 1 category less than 3X in general

California AQ Forecasts

Impact of Isoprene Emissions Error on Exper. CMAQ

R. Mathur, EPA

MAX Hrly O3 Differences (Corr Tight – Old Tight)

Additional NAM ChangesImplemented August 15, 2006

• WRF-NAM PBL often too cool and moist esp. in complex terrain and over/near water bodies

• Over intense storms off West coast

• Corrections Made:– Increased vertical mixing for stable PBLs

– Turned on horizontal mixing at and near all coastal points and in complex terrain

– Fixed SST initializations at Western Lakes

• Double horizontal diffusion everywhere: to be implemented on Sept. 5, 2006

NAM vs NAM-Y evaluations2m Temperature forecast error (12 Z Cycle)

WESTERN U.S.OBS NAMNAM-Y

EASTERN U.S.

NAM vs NAM-Y evaluations2m Dewpoint Temperature forecast error

(12 Z Cycle)

WESTERN U.S.OBS NAMNAM-Y

EASTERN U.S.

NAM vs NAM-Y evaluations 10m WIND forecast error (12 Z Cycle)

WESTERN U.S.OBSNAM NAM-Y

EASTERN U.S.

NAM-CMAQ Aug 3 case

The importance of met driver to CMAQ forecasts

Cloud cover2 m Temperature

NAM-CMAQ Aug 3 case

The importance of met driver to CMAQ forecasts

Cloud cover2 m Temperature

O3 NAMY-NAM

CMAQ under-predictions in California

NAM Temperature NAM Vertical Motions

coast coast

California Ozone ForecastInland Along Coast Across Coast

N S coast

Impact of Lateral Boundary ConditionsCMAQ 3X (w/ GFS Top LBC) vs CMAQ 5x (w/ static, clean LBC)

Figure 2. Ozonesonde measurements (black line) and CMAQ operational 3X (red dashed) and developmenat CONUS (green dashed) ozone predictions for Wallops Island, VA at 18 UTC Feb. 10, 2006. Ozone reported in ppb. Model sigma pressure used as vertical axis.

Wallops Island Chicago

Summary

• Major errors in WRF-Post, biogenic emissions, PREMAQ code and emission files– Not found with retrospective testing– Delayed by final WRF retro configurations and

Summer 05 retro WRF NDAS runs

• True NAM & NAM-Y and vertical structure impact can be assessed on/after August 5 when major isoprene error was corrected

Proposed Plans for FY07• Current CONUS gas-phase run operational

• CONUS gas + aerosol run– Examine impacts of vertical resolution– “ “ Tight horizontal coupling– “ “ Convective mixing scheme in CMAQ– “ “ Direct WRF turbulence predictions for CMAQ PBL mixing

(e.g: TKE, exchange coefficients, Zi calculations…)

• Developmental run to cover Alaska, Hawaii ??– Include wild fire smoke emissions– GFS-GOCART aerosols for LBCs

• ShenAir evaluation

BACKUPS

• BACKUPS

Loose Old Tight Isoprene Fix Tight

Max.-8hr O3: 7/19/06

Max. 8 O3 7/19/06

Loose

Old Tight

Isoprene Fix Tight

2005 Retrospective Analyses

• Time Periods– 12Z July 9, 2005 – 12Z July 20, 2005 - Hurricane Dennis, East coast near exceedences– 12Z August 23, 2005 – 12Z August 30, 2005 - West Coast event

• Configuration:– WRF V2.1 initialized with fully cycled IC (restart file)– Use WRF EGRD3D post file to collapse to CMAQ vertical coordinate (22-40 layers, LCC C grid)– All runs with CMAQ gas-phase chemistry on CONUS domain

RETROSPECTIVE CONUS TESTINGRuns: P. Lee, M. Tsidulko

Analysis: R. Mathur, D. Kang, J. Pleim,…

• 2005 Base: 2005 Operational run• S0: Reflects changes due to NAM-X (WRF)• S1: S0 + Increased vertical layering• S2: S1/S0 + Convective mixing based on NAM-X

conv. Precip rates • S3: S1/S0 + PV ozone BCs• S4: S1/S0 + new mass continuity constraints• S5: S2+S3 (Conv + LBCs)• S6: S2+S4 (Conv + mass cont)• S7: S2+S3+S4 (Conv + LBCs + mass cont)

2005 Retrospective NAM-Eta and WRF(NAM-X)

Verification

• From 12 Z run, 24-48 hr forecasts • UPA : 00 Z valid time

• Low level temperatures: Sfc-700 mb• Low level Winds• Low level RH• 24 h Precip valid at 48 hrs

Precipitation Verification

EAST WEST

Retros Verification 2 m Temperature

EAST U.S. WEST U.S.

2005 Retros Verification 10 m RH

EAST WEST

2005 Retros Verification 10 m Winds

EAST WEST

2005 Retros Verification UPA CONUS Temperatures

RMSE BIAS

2005 Retros Verification UPA Winds RMSE /RH BIAS

Real-time Ozone Verification

• From 2005 3X and 5X 48 hr forecasts• 12 UTC cycle only• AIRNOW hrly ozone monitors• RMSE, BIAS, threat scores, hit rates

FVS O3 1h Aver. Aug 20051x - 3x - 5x over 3x domain

RMSE by Forecast Hour BIAS

Summer 2005 PerformanceSub-Region Bias

NE, SE Regions1X vs 3X 3X All Regions

1X

3X SE NE

LMV

Objective Verification July 26, 2005 Case

•Operational Run : NE US •8 hr daily max

• Experimental Run: Eastern 2/3 US

• 8 hr daily max

CONUS Domain Performance July 12, 2005 Case

• 8h daily max obs vs predicted

CONUS Domain Performance July 12, 2005 Case

Hurricane Dennis

NCEP Graphical Products

Predicted Sfc Ozone(1, 8h & daily max)

NAM Ventilation IndexNAM PBL hgt

Air Quality ForecastingUser Access

✔ Operational Eastern U.S. Domain :

✔ Public: NDGD and TOC ftp server

✔ Surface ozone predictions

✔ State Forecasters: HPC web site

✔ Sfc O3 & met plots

✔ Daily (2pm) conference calls

✔ HPC forecasters trained

✔ Experimental Domain (CONUS):

✔ Focus group: EMC web site

✔ Expanded met plots (pbl hgt, sw rad, ventilation index….)

✔ Sfc & upper level O3 and precurser plots (NOx, NOy,CO,SO2)

✔ Developmental Run (CONUS Aerosols)

✔ Sfc PM

Data Assimilation &Global System Tasks

• CMAQ data assimilation– Plan for surface ozone assimilation– Correlate sfc ozone w/ precursers (Nox VOCs)

• GFS-Chem: Provide Regional boundary conditions– Ozone:

• Include tropospheric product/loss rate terms• Test reduced ozone chemistry (U.Wisc-RAQMS)• Begin testing assimilation of AURA/OMI• CMAQ LBC impact studies

– Aerosols:• Include NASA-GOCART reduced biomass burning/dust and emission processes• Begin testing assimilation of MODIS & AURA/TESS• CMAQ LBC impact studies

Potential Collaborations

• Evaluation of NCEP WRF-CMAQ ozone & aerosol simulations– Experimental & rural obs networks (eg: ETOS,

AERONET, REALM lidar network)

– GOES/MODIS AOD satellite evaluation

• Assimilation of AIRNOW data • Improved cloud mixing, aqueos chemistry

PBL coupling with WRF-CMAQ• Testing of WRF-Chem on-line system

Operational Requirements• Driven by NCEP Operational Meteorological Model (NAM-WRF)• I/O Formats:

– Only machine binary, GRIB and BUFR, disk space limitations

• Time Requirement:– 12 Z 48 hour forecast available by 17:25 Z (1:25 pm EDT)– 06 Z 48 hour forecast available by 13:00 Z ( 9 am EDT)– 65 IBM Power 4 procs available– 12 Z start after Eta is complete (14:30 Z)

• Robustness:– Thoroughly tested & evaluated with retrospective and real-time

experimental runs– Available to NWS Gateway, NDGD: 99% reliability, 24x7 NCEP

support– Accuracy: 90% exceedence hit rate

Implementation Tasks

• Transfer parallel experimental system to Operations:– Complete agreed upon Charter w/ NCO– Provide additional NAM/GFS fields from Postprocessors– * Transfer upgraded CMAQ to EMC – Add internal documentation, refine scripts, adjust IO & dataset names– Support GRIB2 hrly gridded outputs– Perform retrospective tests w/ upgraded NAM– Perform real-time parallels w/ updated emissions files– System evaluation against AIRNOW w NCEP FVS– Prepare estimates of cpu/disk resources for NCO– Prepare Job Implementation Form (JIFs) requests to NCO:– Send out Change Notices, update web page change logs

• Maintain/improve operational graphics, verification plot web pages– May require additional output to GRIB files

Basic Feature Comparison (for NAM): WRF-NMM vs Meso Eta Model

Feature Meso Eta Model WRF-NMM

Vertical Advection schema

T, u, v, TKE: quadratic conservative, 2nd order

Water vapor and cloud water: piecewise linear (Mesinger and Jovic, 2002)

T, u, v: quadratic conservative, 2nd order

TKE, water vapor and total condensate: upstream, flux-corrected, positive definite, conservative (Janjic, 1997)


Feature Meso Eta Model WRF-NMMHorizontalDiffusion

2nd order “Smagorinsky-type” COAC= .27 w/ lower limit on deformation

forward, nonlinear 2nd order “Smagorinsky type” (Janjic, 1990) COAC= .075 with no limit deformation

VerticalDiffusion

Sfc layer: viscous sublayer, similarity theoryAbove sfc lyr: mod Mellor-Yamada Level 2.5 (Janjic,1994,1996, 2001)

Sfc layer: based on similarity theory, viscous sublayer over land/water (Janjic 1996; Chen et al. 1997) Above sfc lyr: 1-D prognostic TKE w/ local vertical mixing (Janjic,1990,1996,2001,2002)

Divergence damping

Coeff = 6.0 Grid-coupling=1.

Coeff = 5.7 Grid-coupling < 0.5

Physics Feature Comparison (for NAM): WRF-NMM vs Meso Eta Model

Physics Meso Eta Model WRF-NMM

PBL & Turbulent mixing

Mellor-Yamada Level 2.5 dry

Mellor-Yamada Level 2.5 w/ moist processes Density variations included (elim. Boussinesq approx.)

Surface exchange

…+ Paulson functions

…+ Holtslag and de Bruin functions

Sfc heat exchange mod

Land-Surface

NOAH LSM with 4 soil layers

May 2005 “glitch”

NOAH LSM with 4 soil layers

Fixed glitch

http://www.emc.ncep.noaa.gov/mmb/mmbpll/Spring2005.NAMUpgrade_newweb/Spring2005.NAMUpgrade.html

http://www.emc.ncep.noaa.gov/mmb/research/FAQ-eta.html

Basic Feature Comparison: Eta 3D-Var vs GSI

Feature Eta 3D-Var GSIGFS & WRF Connectable

No Yes

Unified use of satellite radiances (JCSDA)

No - OPTRANS

Yes – new CRTM

Background error GFS basedNMC method

NMM basedMonte-Carlo

Normalized RH as moisture analysis variable

No – uses specific humidity

Yes

Dynamic constraint using tendencies

No - simple thermal wind

Yes

Adaptively tuned ob errors

No Yes

Use of reported heights No YesBuilt-in cross-validation

No Yes

NCEP Operational NAM and Nest Grid Domains

• NAM: 12 km

4x/day to 84 hours

• One ARW at 6.1 km

• One NMM at 5.5 km

• 48 hr forecasts

• 00Z: AK, HI

• 06Z: West

• 12Z: Central

• 18Z: East, PR

http://www.emc.ncep.noaa.gov/mmb/mmbpll/nestpage/


Feature Meso Eta Model

WRF-NMM

WRF Common Modeling Infrastructure

No Yes – allows faster tech transfer and partnering with community

Dynamics Hydrostatic Hydrostatic + efficient treatment of complete nonhydrostatic corrections

Horizontal grid spacing

12 km semi-staggered E-grid rotated lat/long projection

12 km semi-staggered E-grid rotated lat/long projection – identical

Domain North America

North America – identical

NMM Vertical Domain Compared to Eta

Model top moved to 2 hPa

18 pressure layers above

~420hPa

42 sigma layersbetween

surface and ~420 hPa

1st layer interface above 420 hPa is bottom of first fixed pressure layer

420 hPa

NMM 60 –Layer Distribution

18 layers

42 layers

2 hPa

1st layer 40 m at sea-level then thinner 1st layer 20 m at sea-level then thicker

Eta 60 –Layer Distribution


• Product changes - added various output parameters required by NWS regions, NCEP service centers, Air Quality Forecast system

• Simulated radar reflectivity• Height of the top of the planetary boundary layer• Vertical velocity dz/dt• Ceiling height• Instantaneous clear sky incoming SW flux at the surface• Instantaneous outgoing LW flux at the top of the atmosphere• Dominant precipitation type (replaces Baldwin)• Pressure of LCL• Total column integrated supercooled liquid water and

melting ice• Base and top heights of supercooled liquid water layer • GOES IR look-alike


Feature Meso Eta Model WRF-NMMVertical coordinate

60 step-mountain eta layers

60 hybrid sigma-pressure layers

Top pressure

25 hPa 2 hPa

Gridscale Terrain

Unsmoothed with peak (silhouette) and valley enhancement, lateral boundaries set to sea-level

Lightly smoothed, grid-cell mean everywhere

Sub-gridscale terrain effects

Form drag – Zo effective from terrain variability

Zo depends on elev, subgrid variability & vegetation type

Gravity wave drag

No No



Timesteping schema: fast / inertial gravity waves

forward-backward (Mesinger, 1974) for inertial gravity wave adjustment 30 sec

forward-backward (Janjic, 1979) for all fast waves 26.7 sec Implicit for vertically prop. sound waves

Timesteping schema: advection

first-forward then off-centered 60 sec

Turbulence and moist processes 300 sec

T,u,v: horizontal Adams-Bashforth,vertical Crank-Nicholson off-cntr 26.7sec TKE, water species: Explicit, iterative, flux-corrected 53.3 sec Turbulence and moist processes 160 sec



Horizontal

Advection schema

T, u, v, TKE: energy and enstrophy conserving, quadratic conservative, 2nd order (Janjic, 1984)

Water vapor and cloud water: upstream, flux-corrected, positive definite, conservative (Janjic, 1997)

T, u, v: energy and enstrophy conserving, quadratic conservative, 2nd order (Janjic, 1984)

TKE, water vapor and total condensate: upstream, flux-corrected, positive definite, conservative (Janjic, 1997)

Eta versus NMM Terrain

ETA NMM

• Based on global 30-arc second USGS dataset.• Model terrain height is the average of the USGS values

located within the grid box.• Isolated mountain peaks smoothed with 5-point filter.• Differences with Eta silhouette terrain are small.

NAM: WRF-NMM Implemented on June 20, 2006

Eulerian Mass-Core: NCAR physics suite (MM5 & Eta conversions) (w/options)

NOAH unified 5-layer land-surface model

Ferrier gridscale cloud and microphysics

Kain-Fritsch convection Yong-Sei University PBL Dudhia shortwave RRTM longwave [Also adapted to use NCEP physics]

NMM Core: NCEP physics suite (NMM = modified Eta)

NOAH unified 5-layer land-surface model

Ferrier gridscale cloud and microphysics

Betts-Miller-Janjic convection Mellor-Yamada-Janjic 2.5 PBL Lacis-Hansen shortwave Fels-Schwartzkopf longwave [Also adapted to use NCAR physics]

Status of the NOAA WRF-CMAQ Regional Air Quality Forecast System

Documents

Transcript of Status of the NOAA WRF-CMAQ Regional Air Quality Forecast System