Transcript No Slide Title

Incorporation and use of the NOAH LSM in
the Coupled/Ocean Atmosphere Mesoscale
Prediction System (COAMPS)®
Teddy Holt
Naval Research Laboratory
Marine Meteorology Division, Code 7533
Monterey, CA
[email protected]
5th WRF Land Surface Modeling Workshop
NCAR, Boulder CO
13-15 September 2005
COAMPS® is a registered trademark of the Naval Research Laboratory
Outline
• NRL WRF integration strategy (including LSM)
• Long-term (8-month) simulations
• Ongoing and future plans
• Issues
NRL WRF integration strategy
Strategy
• Build WRF standard I/O and physics interfaces into COAMPS® to allow for:
• Initial and lateral boundary condition input from any model; standard output
• Interchangeable physical parameterizations
• Investigate/modify WRF infrastructure to merge COAMPS® dynamic core into WRF
• NCAR (ARWRF)
• NCEP (NMM)
I/O
COAMPS® Framework,
Core
• NRL (COAMPS)
Plug-and-Play
Physics
• Cumulus
• Radiation
• LSM
• Boundary Layer, . .
I/O
PostProcessing
Pre-Processing,
Input
Parameters
Model Framework,
Dynamic Cores
NRL WRF integration strategy
Accomplishments
• NetCDF, GRIB, and binary output options for COAMPS®
• FNMOC, NRL discussing specific details with AFWA for
operational implementation
• Working with NCAR to develop improved WRF Standard
Initialization (SI)
• candidate to replace existing COAMPS® pre-processing
program
• Prototype physics interface developed and tested
• Physics conversion status:
•
•
•
•
•
•
Cumulus Parameterization [Adapted from WRF]
Radiation [On-going]
Land-surface [Complete]
Surface/Boundary Layer [On-going]
Explicit Moist Physics [On-going]
Aerosols [On-going]
Long-term data assimilation simulations
COAMPS® Operational Europe domain (81 km)
Terrain
3 km
1.5
Nest 1: 111 x 80 x 30 levels
0
48-h forecasts from 25 January to 16 September 2004
Three experiments:
1. cont
-- operational COAMPS®: Slab surface model
2. lsm
-- WRF LSM (“unified” v1.3; Dec 2003) with no soil initialization (climatology)
3. agr
-- WRF LSM (“unified” v1.3; Dec 2003) with AGRMET (AFWA) initialization
LSM data assimilation experiments
• cont: CONTROL uses COAMPS® existing soil slab model with force-restore surface energy budget
• lsm: uses NOAH LSM with climatological soil and vegetation initialization
• cold start initialization: Soil temperatures = ground temperatures (constant in depth)
Soil moisture = Soil porosity * climatological ground wetness
Unfrozen soil moisture = soil moisture
Canopy water content = snow water equivalent = 0.0
Vegetation greenness = climatology
• warm start: previous forecast soil/vegetation fields used as analysis fields for next forecast
• agr: uses NOAH LSM with AGRMET soil and vegetation initialization
• Air Force Weather Agency (AFWA) Agricultural Meteorology (AGRMET) modeling system
• near real-time, off-line, global 47-km resolution agricultural meteorology analysis model
• input data:
• first guess fields from NOGAPS analyses (surface winds, isobaric
temperatures, geopotential heights, and relative humidity)
• surface observations
• three-hourly SSM/I rain rate analysis
• precipitation analysis based on rain gauge observations
• AFWA SNODEP snow depth analysis based on surface and satellite observations
• AFWA CDFSII global cloud analysis for up to four levels
• NOAH LSM, outputting soil temperatures, soil moistures, and unfrozen soil moistures
at 4 levels, and canopy moisture content, snow water equivalent, and greenness
• runs four cycles per day, each at approximately +0430 hours
• warm start: reinitialize soil/vegetation fields using AGRMET analyses
LSM data assimilation experiments
Operational COAMPS® Europe domain (81 km): 25 January to 16 September 2004
48-h forecasts valid 12 UTC (daytime)
2-m air temperature
2-m dew point temperature
Average
number of
observations
per forecast
used for
verification =
~1200
Climo soil initialization (lsm) is too dry for the entire period; yet daytime air temperature is warmer (better) only from
Apr onward (not in winter). Why? Snow-related? Spin-up?
AGRMET soil initialization (agr) differs little from slab (cont); largest differences: agr drier from Aug-Sep
LSM data assimilation experiments
Operational COAMPS® Europe domain (81 km): 25 January to 16 September 2004
48-h forecasts valid 00 UTC (nighttime)
2-m air temperature
2-m dew point temperature
Average
number of
observations
per forecast
used for
verification =
~1200
Climo soil initialization (lsm) is too dry for the entire period and daytime air temperature is colder for the entire period
AGRMET soil initialization (agr) similar to slab model (cont) until mid Apr, then agr too dry and cold
LSM data assimilation experiments
Operational COAMPS® Europe domain (81 km): 25 January to 16 September 2004
All forecasts from 00 UTC
2-m air temperature rmse
day
night
day
2-m dew point temperature rmse
day
night
cont
agr
cont
Average
number of
observations
per forecast
used for
verification =
~1200
agr
lsm
2-m air temperature bias
cont
day
night
agr
lsm
day
lsm
2-m dew point temperature bias
day
night
day
cont
agr
lsm
AGRMET soil initialization (agr) and slab (cont) similar during daytime
Climo soil initialization (lsm) too cold at night and warmer (best of three) during the day
Climo soil initialization too dry both day and night
Ongoing and future plans
WRF-COAMPS® Soil Analysis System
• Off-line WRF-COAMPS® LSM system that can run on
single or multiple processors (MPI)
• Forcing fields:
• COAMPS® near surface wind, temperature, and dew point
temperature analyses (MVOI or 3DVAR)
• AGRMET radiation, precipitation, and snow analysis
• NRL SSMI/TRMM precipitation analysis
• Data assimilation
• Assimilation of high-resolution skin temperature retrieved
from IR radiance (MODIS and AVHRR)
Issues
• COAMPS® LSM System
• will be transitioned to operations 30 Sept 2005
• will not be “turned on” until viable soil initialization is available
• funding expires 30 Sept 2005
• WRF-COAMPS® Soil Analysis System
• testing slowed by MSRC downtime due to Katrina
• merger with HRLDAS/ new WRF system ?
• viability/use for non-CONUS areas
• AGRMET global resolution (15 km?)
• spin-up time/data availability
• What version of LSM code do we have/use?
• configuration management
• How does WRF-COAMPS® SI integrate into LSM?
• databases (global) applicable for high-resolutions (~1km)
• NASA LIS (?)
• Validation/Evaluation
• Integration of LSM with urban parameterization
LSM data assimilation experiments
Europe Nest 1 (81 km): 00 UTC Jan 1 to 00 UTC 15 Jan 2003
Tobolsk, Russia
10-m air temp (C)
10-m mixing ratio (*0.1 g/kg)
cont
lsm
agr
Obs
cont
lsm
agr
36-h fcst valid 12 UTC 10 Jan 2003
10-m air temp (C)
10-m mixing ratio (*0.1 g/kg)
Obs
cont
lsm
agr
36-h fcst valid 00 UTC 11 Jan 2003
Control
LSM
AGR
cont
lsm
agr
LSM data assimilation experiments
COAMPS® 1-15 January 2003 CONUS nest 2 (27 km)
Dew point
temperature
(deg C)
pressure (hPa)
pressure (hPa)
Air
temperature
(deg C)
All
36-h
forecasts
COAMPS® 1-15 January 2003 CONUS nest 2 (27 km)
agr 0- to 36-h fcst from 00 UTC 01 Jan 2003
0
10-cm soil temperature Near Billings, MT
10-cm soil temperature (C)
2.8
2.1
9.3
-0.9 -1.9
-1.6
0.8
-0.6
2.8
1.7
0.6
3.8
1.8
1.4
1.9
4.5
2.0
6.8
10.0
6.5
6.9 9.4 16.5
13.1
11.2
13.1
SCAN
12.6
Observations
15.6
0.08
0.11
5
AGR
-5
0.4
0
0.3
AGR
0.2
-5
0.1
-10
Near Atlanta, GA
15
SCAN obs
0.03
0.07
0.45
• COAMPS AGR
soil moisture
too moist for the
entire forecast
SCAN obs
0.0
0.6
• COAMPS AGR
soil temperature
too cold over last
12-h of forecast
10-cm soil moisture
21.0
0.37 0.27
0.37
0.04
0.31
0.39
0.18
0.37
0.09
0.39
0.37
0.36
0.38
0.37 0.38 0.37
0.25
0.28
0.40
0.23
0.54
0.05
0.13
0.34
0.06
0.25
SCAN obs
-4
5.2
10-cm soil moisture (vol. fraction)
0.19
0.33
-2
4.3 4.3
agr 12-h fcst valid 12
UTC 01 Jan 2003
0.23
10
-3
-1.5
5.5
-1
AGR
10
• COAMPS AGR
soil temperature
too cold over the
entire forecast
10-cm soil temperature
0.3
0.15
0.4
AGR
0.2
0.12
0.0
0.0
SCAN obs
10-cm soil moisture
forecast hour
• COAMPS AGR
soil moisture
slightly moister
than observations
for the entire
forecast