Improving WRF/CAMx Model Performance using Satellite Data ...
Transcript of Improving WRF/CAMx Model Performance using Satellite Data ...
Improving WRF/CAMx Model Performance using Satellite Data Assimilation Technique for the
Uintah Basin
19th AMS Conference on Mountain MeteorologyJuly 15, 2020
Huy TranTrang Tran
AcknowledgementsUtah Division of Air Quality
Uintah Impact Mitigation Special Service DistrictUtah Legislature SB118
Manual and forcible corrections to WRF had to be made to capture surface albedo)
Corrected WRF Albedo in winter 2013 ozone episode
2
(Ref.: Crosman et al., 2015; Neemann et al., 2015)
Neemann, E. M., Crosman, E. T., Horel, J. D., and Avey, L.: Simulations of a cold-air pool associated with elevated wintertime ozone in the Uintah Basin, Utah, Atmos. Chem. Phys., 15, 135–151, https://doi.org/10.5194/acp-15-135-2015, 2015.
Albedo
Surface albedo corrections are:• Performed on case by case basic, and there is no standard method
to perform the correction.• Heavily relies on knowledge and observations of modelers on the
actual surface conditions during simulated episode
• May results in unrealistic albedo estimations over model domain
Utilizing satellite observations (MODIS) to correct surface characteristics
Surface Shortwave Albedo (ALBEDO)• MCD43A1: MODIS/Terra+Aqua BRDF/Albedo Model Parameters Daily L3 Global - 500m V006
• MCD19A1: MODIS/Terra+Aqua Land Surface BRF Daily L2G Global 500m, 1km and 10km SIN Grid V006
• MCD19A2: MODIS/Terra+Aqua Land Aerosol Optical Thickness Daily L2G Global 1km SIN Grid V006
Surface Snow Cover Fraction (SNOWC)• MOD10A1: MODIS (Terra) Snow Cover Daily L3 Global 500m Grid
Leaf Area Index (LAI)• MCD15A3H: MODIS/Terra+Aqua Leaf Area Index/FPAR 4-Day L4 Global 500 m (4 days composite)
3
MODIS Data Assimilation Processes6SV RoutineMCD43A1
fiso(𝜆𝜆), fvol(𝜆𝜆), fgeo(𝜆𝜆)
BRDF Albedo Determination:
Black − sky Albedo 𝛼𝛼𝐵𝐵𝐵𝐵𝐵𝐵 𝜃𝜃𝐵𝐵𝑆𝑆𝑆𝑆, 𝜆𝜆 = 𝑓𝑓𝑖𝑖𝑖𝑖𝑖𝑖 𝜆𝜆 𝑔𝑔𝑖𝑖𝑖𝑖𝑖𝑖 𝜃𝜃𝐵𝐵𝑆𝑆𝑆𝑆 + 𝑓𝑓𝑣𝑣𝑖𝑖𝑣𝑣 𝜆𝜆 𝑔𝑔𝑣𝑣𝑖𝑖𝑣𝑣 𝜃𝜃𝐵𝐵𝑆𝑆𝑆𝑆 + 𝑓𝑓𝑔𝑔𝑔𝑔𝑖𝑖 𝜆𝜆 𝑔𝑔𝑔𝑔𝑔𝑔𝑖𝑖 𝜃𝜃𝐵𝐵𝑆𝑆𝑆𝑆
White-sky Albedo 𝛼𝛼𝑊𝑊𝐵𝐵𝐵𝐵 𝜃𝜃𝐵𝐵𝑆𝑆𝑆𝑆, 𝜆𝜆 = 𝑓𝑓𝑖𝑖𝑖𝑖𝑖𝑖 𝜆𝜆 𝑔𝑔3𝑖𝑖𝑖𝑖𝑖𝑖 + 𝑓𝑓𝑣𝑣𝑖𝑖𝑣𝑣 𝜆𝜆 𝑔𝑔3𝑣𝑣𝑖𝑖𝑣𝑣 + 𝑓𝑓𝑔𝑔𝑔𝑔𝑖𝑖 𝜆𝜆 𝑔𝑔3𝑔𝑔𝑔𝑔𝑖𝑖
Blue-sky Albedo 𝛼𝛼𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵= 𝛼𝛼𝐵𝐵𝐵𝐵𝐵𝐵 1 − 𝐷𝐷 + 𝛼𝛼𝑊𝑊𝐵𝐵𝐵𝐵𝐷𝐷
ESMF Re-gridding
MCD19A1𝜃𝜃𝐵𝐵𝑆𝑆𝑆𝑆 , 𝜗𝜗𝐵𝐵𝑆𝑆𝑆𝑆, 𝜃𝜃𝑉𝑉𝑖𝑖𝑔𝑔𝑉𝑉 ,
𝜗𝜗𝑉𝑉𝑖𝑖𝑔𝑔𝑉𝑉
MCD19A2AOD
Lookup Table for Diffuse Radiation
Fraction𝐷𝐷 (𝜆𝜆, 𝜃𝜃𝐵𝐵𝑆𝑆𝑆𝑆 , 𝜗𝜗𝐵𝐵𝑆𝑆𝑆𝑆, 𝜃𝜃𝑉𝑉𝑖𝑖𝑔𝑔𝑉𝑉, 𝜗𝜗𝑉𝑉𝑖𝑖𝑔𝑔𝑉𝑉,𝐴𝐴𝐴𝐴𝐷𝐷)
MCD15A3HLeaf Area Index
WRF Surface Nudgingwrfsfdda CAMx
MOD10A1Snow Cover
WRF
4
wrfcamx
wrfinput
Modifications to WRF source codes are required
/RegistryRegistry.EM_COMMON
/physmodule_physics_init.Fmodule_fddagd_driver.Fmodule_fdda_psufddagd.F
/shareoutput_wrf.F
/dyn_em.module_first_rk_step_part1.F
/run/namelist.input…&fddaif_ramping = 1,dtramp_min = 60,grid_fdda = 1,grid_sfdda = 1,sgfdda_interval_m = 360, galb_sfc = 0.950,gsnc_sfc = 1.000,/
WRF
Input files:wrfinput (LAI)wrfsfdda (ALBEDO, SNOWC)
5
WRF Model Configurations
6
Parameters Values
Grid size (x,y) 298 x 322 in 1.3km horizontal resolution
Vertical levels 37
Vertical coordinates Terrain-following Eta (non-hybrid)
Vertical grid spacing 12-16 m in the boundary-layer
Topographic dataset USGS GTOPO30Land use data set modified NLCD2011Veg parm table variables modified for winter simulations
SNUP, MAXALB
Snow cover initialization SNODAS
Re-initialize Every 5 days
WRF Treatment Option SelectedMicrophysics ThompsonLongwave radiation RRTMGShortwave radiation RRTMGLand surface model (LSM) NOAHPlanetary boundary layer (PBL) scheme
MYJ
MODIS data assimilation results in better estimations of ALBEDO
Default WRF (REF)
7
MODIS assimilation
Modeling episode Feb 01- 28 2011Albedo
MODIS data assimilation results in “better” estimations of Snow Cover (SNOWC)
Default WRF (REF)
8
MODIS assimilation
Modeling episode Feb 01- 28 2011
SNOWC
Effect of MODIS data assimilation
9
5
10
15
20
25
30
2/1
2/2
2/3
2/4
2/5
2/6
2/7
2/8
2/9
2/10
2/11
2/12
2/13
2/14
2/15
2/16
2/17
2/18
2/19
2/20
2/21
2/22
2/23
2/24
2/25
2/26
2/27
2/28
Snow
Wat
er
Equi
vale
nt(k
g/m
2)
20
70
120
170
2/1
2/2
2/3
2/4
2/5
2/6
2/7
2/8
2/9
2/10
2/11
2/12
2/13
2/14
2/15
2/16
2/17
2/18
2/19
2/20
2/21
2/22
2/23
2/24
2/25
2/26
2/27
2/28Sn
ow D
epth
(cm
) REF MODIS
0
0.2
0.4
0.6
0.8
1
2/1
2/2
2/3
2/4
2/5
2/6
2/7
2/8
2/9
2/10
2/11
2/12
2/13
2/14
2/15
2/16
2/17
2/18
2/19
2/20
2/21
2/22
2/23
2/24
2/25
2/26
2/27
2/28
Snow
Cov
er
Frac
tion
Effect of MODIS data assimilation (Cont.)
10
0
500
1000
1500
2000
2/1 2/3 2/5 2/7 2/9 2/11 2/13 2/15 2/17 2/19 2/21 2/23 2/25 2/27
PBLH
(met
er A
GL) PBLH at Ouray (OU) REF MODIS
-50
-40
-30
-20
-10
0
10
20
30
2/1 2/3 2/5 2/7 2/9 2/11 2/13 2/15 2/17 2/19 2/21 2/23 2/25 2/27
Laps
e Ra
te a
t Our
ay (C
/km
)
REF MODIS
extremely stable
moderate stable
slightly stable
neutral
unstable
MODIS data assimilation does NOT necessarily result in better overall WRF performance
11
MODIS data assimilation does NOT necessarily result in better overall WRF performance
12
Modifications to CAMx source codes are also required
CAMx /Includescamxfld.inc
/Mod_srccamxfld.f
/IO_binsreadinp.fmetinit.f
/CAMxgetalbedo.fstartup.ftstep_init.f
wrfcamx(ALBEDO, SNOWC)
13
• CAMx calculates its own ALBEDO and SNOWC as the function of snow water equivalent, snow age, and landuse types
• At the initialization of CAMx simulation, CAMx-ALBEDO maybe higher than WRF-ALBEDO
Higher ALBEDO obtained in WRF-MODIS did not translate to higher photolysis rate in CAMx
14
0102030405060708090
100110120130140150
01/3
1-17
02/0
1-17
02/0
2-17
02/0
3-17
02/0
4-17
02/0
5-17
02/0
6-17
02/0
7-17
02/0
8-17
02/0
9-17
02/1
0-17
02/1
1-17
02/1
2-17
02/1
3-17
02/1
4-17
02/1
5-17
02/1
6-17
Ozo
ne P
hoto
lysi
s Rat
e (h
r-1) OURAY (OU)
REF MODIS
MODIS data assimilation is not the magic wand to solve ozone underperformance issue
15
0
20
40
60
80
100
120
140
160
2/1 2/3 2/5 2/7 2/9 2/11 2/13 2/15 2/17
Ozo
ne (p
pb)
Ouray Observation NAAQS REF MODIS
Summaries
16
• We have developed a methodological approach in improving WRF/CAMxperformance using satellite data.
• MODIS data assimilation “improves” WRF and CAMx model performance.
• Positive effect of MODIS assimilation varies with different simulation episode and length of the episode.
• Effect of the technique could be more substantial in direct coupling meteorology-chemistry model.
• Better resolution and more frequent satellite dataset in future will enhance the benefit of satellite- base data assimilation technique.
• The same approach could be applied for assimilating other satellite data product to improve performances of both meteorology and chemistry model.