linkage between wrf/nmm and cmaq daewon byun (pi) c.k. song & p. percell university of houston...

29
Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS) Coauthors: Jon Pleim, Tanya Otte, Jeff Young, Rohit Mathur ASMD, Air Resources Laboratory, NOAA In partnership with U.S. EPA and many others… Hsin-Mu Lin, David Wong, etc…

Upload: irene-singleton

Post on 31-Dec-2015

229 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Linkage between WRF/NMM and CMAQ

Daewon Byun (PI)C.K. Song & P. PercellUniversity of Houston

Institute for Multidimensional Air Quality Studies (IMAQS)

Coauthors:Jon Pleim, Tanya Otte, Jeff Young, Rohit Mathur

ASMD, Air Resources Laboratory, NOAAIn partnership with U.S. EPA

and many others…Hsin-Mu Lin, David Wong, etc…

Page 2: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Consistent governing set of equations & state variables

Consistent coordinates and grid structures

Consistent numerics & physics, and parameterizations

Flexible: able to help diverse stake holders (research – regulatory application – use of different emissions inputs)

Allow studying effects of using different basic input data (e.g., Land Use/Land Cover, topography, emissions, etc) separately

Same (*) numerics & physics, and parameterizations

Same (*) coordinates and grid structures

Same (*) governing set of equations & state variables

What are the main science issues of the NWP & AQM coupling?

Off-line

On-line* Need to check how closely the dynamics variables and trace species are matched

Page 3: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

WRF/nmm

WRF/nmm Postprocessors

(vertical/horizontal)

PREMAQ* (consistent vertical coordinate)

CMAQ

Loose coupling Tight coupling

WRF/nmm

WRF-CMAQ Interface Processor

CMAQ/E-grid

Components of Off-line Coupled system

Spatial interpolation

Lambert conformal projection C-grid

On rotated lat/long E-grid coordinateConsistent vertical coordinate

Page 4: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

zszs

zzs

szs Jf

s

vJ

m

Jm

t

JVτ

VV

VV

3

32 ˆ

ˆ)(

)(sss

sss J

s

p

z

s

g

mJp

mJ F̂

(Jsw)

tm2s

JswVz

m

(Jsw ˆ v 3 )

sJs

m

pss

sz Js F3

wQ

(Js )

tm2s

Jsˆ V s

m2

(Jsˆ v 3 )

sJsQ

QJs

vJ

m

Jm

t

Js

ssss

s

)ˆ(ˆ)( 3

22 V

(i Js )

tm2s

iJsˆ V s

m2

( i Jsˆ v 3 )

sJsQ i

)ln()ln(oo

doo

vd RT

TC

o

do p

Rpp

Fully Compressible Atmosphere (OOyama, 1990) used for CMAQ

•Follow coordinates/grid of met model

•Reproduce Jacobian

•Couple state variables consistently

Proper Coupling Requires

Page 5: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

WRF/NMMhttp://www.dtcenter.org/wrf-nmm/users/

Nonhydrostatic Mesoscale Model (NMM) core of the Weather Research andForecasting (WRF) system was developed by NOAA/NCEP

ARW (Advance Research WRF)

+ Terrain following hydrostatic P coord. or Terrain following sigma (ARW)+ Arakawa-C+ Conserves mass, momentum, dry entropy, and scalar

WRF/NMM

+ Hybrid sigma-pressure coord.+ Arakawa-E+ Conserves mass, momentum, enstrophy, TKE and scalar

WRF (ARW core) WRF (NMM core)

Page 6: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Hybrid Sigma-Pressure Coordinate

leveltoptheinpressureP

partsigmaofdifferencepressurethicknessPD

partpressureofdifferencepressurethicknessPD

PtyxPDetaPDetatyxP

T

TOP

TTOPhydro

)|,()()()|,,( 21

Page 7: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

kx 1ˆ3

kc

TOPx

kcx

PD

gx

p

p

z

x

zJPartUpper

PD

gx

p

p

z

x

zJPartLower

1ˆˆ

:

1

1ˆˆ

:

33ˆ

33ˆ

3

3

)(2

11ˆ:

))(

1(2

11ˆ:

2

)1()1(ˆ

0,:

)(,1:

3

3

213

21

21

21

TOP

T

TTOP

TOP

T

TTOP

TTOP

PD

PPxPartUpper

PD

PPDPxPartLower

etaetax

etaPD

PPetaPartUpper

PD

PPDPetaetaPartLower

PetaPDetaPDP

)2(1

:

)2(1

:

ˆ

3

3

3

ˆ

ˆ

topx

x

x

PDg

JPartUpper

PDg

JPartLower

x

zJ

tops

topkk pp

pp

Initial Terrain-Following Hydrostatic Sigma coordinate

Method 1

Method 2

kc : sigma interface of the lower and upper layersPD: pressure of top of lower layer

Define J for the Generalized Vertical Coordinate

Page 8: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Vertical Jacobian Discontinuity Problem & Solution

For example,

SIGMA LEVELS = 1.0000, .9976, .9948, .9920, .9890, .9858, .9825, .9790, .9754, .9718, .9679, .9637, .9590, .9538, .9480, .9415, .9340, .9251, .9144, .9020, .8883, .8736, .8582, .8420, .8253, .8079, .7900,.7714, .7523, .7326, .7124, .6915, .6699, .6477, .6248, .6015, .5779, .5540, .5300, .5057, .4812, .4566, .4319, .4070, .3822,.3576, .3333, .3100, .2881, .2679, .2494, .2316, .2135, .1936,.1707, .1445, .1159, .0863, .0569, .0282, .0000,

Case 1) Surface pressure = 101300 Pa & sigma(kc)=0.3822,

Pkc PD PDtop Js(lower) Js(upper)

JP method

41806 59494 36806

118988/ρg

73612/ρg

UH method

96300/ρg 96300/ρg

Case 2) Surface pressure = 70000 Pa & sigma(kc)=0.3822,

Pkc PD PDtop Js(lower) Js(upper)

JP method

41806 28194 36806

56388/ρg 73612/ρg

UH Method

45636/ρg 96300/ρg

g

zJ

PdRTk

hydrovk )(ln)()(11

One way to remove discontinuity

Page 9: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Horizontal E-Grid System of WRF/nmm:Rotated lat./long & Arakawa-E grid -> C-grid for CMAQ

If we use diamond gridC(C,R,L,S) -> C*(CR, L,S)

Page 10: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Dynamics with Semi-Staggered Arakawa E gridThe E grid is essentially a superposition of two C grids.

When only the adjustment terms in the equations of motion and continuity are considered, two large-scale solutions from each C grid may exist independently, and a noisy total solution results.

So, employ the forward-backward time differencing scheme to prevents gravity wave separation and thereby precludes the need for explicit filtering (Mesinger 1973: Mesingerand Arakawa 1976; Janji´c 1979).

(1,1) (2,1)

(1,2) (2,2)

(223,501)

(223,500)

(222,501)

(222,500)

dx

dy

dx dx

dx = 0.0534521 deg. (rotated Lon.)

dy = 0.0526316 deg. (rotated Lat.)

2dx

scalarvector

Advantages of usingE-grid with dynamicssolution

Page 11: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Dimension for Grid Point

Dot-Point Cross-Point Flux-Point

For MM5 (MCIP) (NCOL+1 , NROW+1) (NCOL , NROW)X-dir (NCOL+1 , NROW)Y-dir (NCOL , NROW+1)

For WRF/ARW (WCIP)

(NCOL+1 , NROW+1) (NCOL , NROW)X-dir (NCOL+1 , NROW)Y-dir (NCOL , NROW+1)

For WRF/NMM (NCOL , NROW) (NCOL , NROW) (NCOL , NROW)

Consistent coordinates and grid structuresWRF/EM & CMAQ utilize Arakawa-C Grid

Arakawa-B Grid (MM5) is linearly interpolated onto Arakawa-C Grid (CMAQ)

What to do with NMM E-grid data?

Page 12: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

How to Utilize Arakawa-E for CMAQ?

Develop a horizontal advection algorithm in CMAQ for Arakawa E-grids Split 2-D horizontal advection operator into 1-D operators and use

CMAQ-proven 1-D schemes, such as PPM, with alternation between appropriate X and Y directions

Work directly with meteorological variables on the E-grid - avoid spatial interpolation

Use rotated square cells (rotated B-grid then on C-grid)

Spatial distribution of dependent variables for a uniformly spaced Arakawa E-Grid

E-Grid with rotated square cells. Scalar variables are considered to be constant on each grid

Page 13: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Advantages Makes the E-Grid look like a B-grid whose “rows” and

“columns” are along diagonal SW→NE and SE→NW lines Can use 1-D algorithm, e.g. PPM, along these lines CMAQ (and preprocessors) are familiar with turning B-grid

data into C-grid flux point data

Disadvantages

Diagonal lines of cells have variable lengths, which requires non-trivial extra book-keeping (in EGRID_MODULE.F)

Requires interpolation of wind velocities to get flux point values Jagged boundary effect Parallelization could be more difficult

Page 14: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Grid geometry changes depending on whether thenumber of columns or rows is even or odd

Bookkeeping issues

Partitioning for parallelization

Page 15: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Jagged Boundary Effect

Boundary values propagate into the domain because boundaries are angled 45 degree

rotated B-grid then on C-grid

Page 16: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Option 1: rotated B-grid then on C-gridCMAQ C-grid

Comparison between regular CMAQ and Option 1

Page 17: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

START

get env./IOAPI variables

define grid/coord.- rotated Lat./Lon coord.- E-grid structure- calculate Dx & Dy- allocate memory xgrid and cgrid

get met. data

calculation for WRF/NMM- Eta1 & Eta2- geopotential height- hydrostatic pressure- hydrometeor

derive dynamic fld.

GRIDOUT

METCRO/DOTOUT

continue

END

Calculation Flow of WCIP/NMM Mapping Variables

Page 18: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

TEST Run

- Target Period : 00Z June 28 - 06Z June 30, 2006- Horizontal Resolution : ~ 12 km

Page 19: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Model ConfigurationC-Grid E-Grid

-----------------------------------------------------------------------------------------Met. MM5 v3.6.1 WRF/NMM v2.1

(w/ Eta forecast) (w/ Eta forecast)

MCIP MCIP v3.0 WCIP/NMM v1.0BCON BCON/Standard BCON/E-grid v1.0ICON ICON/ Standard ICON/ E-grid v1.0CMAQ CMAQ v4.4 CMAQ/ E-grid v1.0-----------------------------------------------------------------------------------------+ I.C. C-Grid UH-AQF/CMAQ 12km resolution output

00Z June 28, 2006 + B.C. C-Grid UH -AQF /CMAQ 36km resolution output

00Z June 28 – 06Z June 30, 2006 + Emisson None+ Chem. Mech. CB-IV

Page 20: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Domain ConfigurationC-Grid E-Grid

-----------------------------------------------------------------------------------------

Met. (MM5) (WRF/NMM)+ nx(dx) 100(12 km) 85(0.0780 deg.*)+ ny(dy) 100(12 km) 135(0.0724 deg.)+ nz 43 sigma 44 hybrid sigma-P

CMAQ+ nx(dx) 89** 57***+ ny(dx) 89 113+ nz 23 (see COORD_23L.EXT) 23 (JP & Dis.)

-----------------------------------------------------------------------------------------* ds=sqrt(dx**2+dy**2) ~ 12 km** As for DOT case of MCIP, nx and ny should be 90** As for CRO/DOT case of WCIP/NMM, nx(ny) should be 59(115)

Page 21: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Recommended Model Physics for WRF/NMM Microphysics : Ferrier Cumulus Convection : Betts-Miller-Janjic Shortwave Radiation : GFDL Longwave Radiation : GFDL Lateral diffusion : Smagorinsky PBL, free atmosphere : Mellor-Yamada-Janjic Surface Layer : Janjic Scheme Land-Surface : 4-layer soil model

Page 22: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)
Page 23: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

CMAQ ResultsNo emissions,

Transport & Chemistry Only

12Z (06 CST) June 28, 2006(12 hrs after initial time)

Page 24: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

C-Grid E-Grid Wind PBLHCO O3

Page 25: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

C-Grid E-Grid Wind PBLHCO O3

hr18

Page 26: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

C-Grid E-Grid ZH JabobianAir temp. U-wind

---- 13000 m---- 13000 m

discontinuity

Page 27: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

C-Grid E-Grid

CO CO

Page 28: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

C-Grid E-Grid

O3 O3

Page 29: Linkage between WRF/NMM and CMAQ Daewon Byun (PI) C.K. Song & P. Percell University of Houston Institute for Multidimensional Air Quality Studies (IMAQS)

Conclusion+ Presented a method to cast the WRF meteorological data on CMAQ grid & coordinate structures to represent transportation of pollutants.

+ Developed WCIP/NMM, BCON/E-grid, ICON/E-grid, and CMAQ/E-grid

+ Performed simulation (WRF/NMM -> CMAQ/E-grid) was successfully done

+ A simple evaluation with transport and chemistry was performedResults of CMAQ/E-grid simulation is generally consist with CMAQ/C-grid but reveal properly the discrepancy of meteorological fields

Future Work+ To solve some unsolved problems (WRF/NMM IOAPI, etc)+ More Evaluations & Documentation+ Deliver the developed codes to NOAA/EPA for National AQF