Title: Status and Plans for the RSA LAPSMM5 Implementation July 2005
1Status and Plans for the RSA LAPS/MM5
Implementation July 2005
- John McGinley, Steve Albers, Ed Szoke, Dan
Birkenheuer - NOAA Forecast Systems Laboratory
In collaboration with the Cooperative Institute
for Research in the Atmosphere (CIRA), Colorado
State University, Fort Collins, Colorado
2Overview
- Review of System Design
- Case Studies - Verification
- Work Accomplished Since 2004 TIM
- Recommended Schedule
- Issues
- Discussion
3System Design Review
- Basic requirements
- New 24-h forecast every 6 hours
- High spatial resolution (1 km objective)
- Support launch weather operations and range
safety requirements for dispersion modeling - Integrate with display system (AWIPS)
- Trades considered
- Size of domain and nesting options
- Forecast length vs. scale predictability
- Computational resources vs. cost-benefit
- Future upgrade path
4Basic Solution
- LAPS coupled with MM5 NWP model
- Use diabatic initialization (hot start)
- Utilize parallel code on Linux cluster
- Integrate with AWIPS in a modular fashion
5AWIPS Integration
NOAAPORT
LDAD
Obs/Radar/Sat/NCEP
LAPS Anal/Fcst Grids
AWIPS Data Server
AWIPS Workstation
Modeling Server
6Hardware Configuration
- Linux Cluster Modeling Server
- 1 master/8 compute nodes (18 processors)
- Dual P-III 1GHz, 1GB RAM on each node
- 60GB RAID array on front-end
- Myrinet inter-connect
- Interacts with AWIPS DS via NFS
- LAPS analyses and 2-h MM5 update forecast run on
master node - 6-hourly MM5 and post-processing use compute
nodes
7LAPS Domain Configuration
- Triple Nest Domain
- 97x97
- 10.0/3.3/1.1 km ?x
- 10km for Model Init
- 3.3/1.1 for NowCasting
- LAPS Runs
- Hourly analyses
- 41 Pressure levels
- Runs at H20 min
- Available at H30 min
WR
ER
8RSA LAPS Data Sources
- National Data (NOAAPORT SBN Feed)
- Eta (Grids 211, 212, and 215)
- Regional narrowband WSR-88D reflectivity
- GOES imagery (Vis, SWIR, 2 LWIR)
- MDCARS
- RAOBs
- METARs/Ship Reports/Buoys
- National Profiler Network
9RSA LAPS Data Sources
- Sea Surface Temperature Data
- NCEP internet FTP feed
- GFS model
10RSA LAPS Data Sources
- Local Data (via LDAD)
- Local wideband WSR-88D (Z and Vr)
- Local ASOS Observations
- 50/915 MHz Wind Profilers
- MiniSODAR Wind Profiles
- RASS
- Tower observations
- AMPS Soundings
- Local MM5 forecast grids
11MM5 Forecast Model
- MM5 v3 used for RSA application
- Upgraded to release 3.5 in June 2002
- FSL modifications for diabatic initialization
- FSL-developed runs scripts suitable for
operations - Concurrent post-processing
- Supports multiple output formats
- Allows viewing on workstation as model runs
- Hourly temporal output
12MM5 Forecast Model
- Two configurations for RSA
- Update run to provide first guess for LAPS
- Used for downscaling national model
- Runs every hour out to 2 h on master node
(serial) - Initialized with national first guess (Eta
forecast) and NCEP SST - Domain 1 (10km) only
- Not displayed on AWIPS
- Forecast run used for operational forecasts
- Diabatically intialized with LAPS
- Runs every 6 hours
- Triple nest (10/3/1.1) with forecasts out to
24/12/9
13MM5 Run Configuration(Development System Config)
- 30-second timestep
- MRF PBL Scheme
- Explicit Schultz II microphysics on all domains
- 2-way feedback between nests
- Domain 1 initialized with LAPS (diabatic)
- Domains 2 and 3 interpolated from parent
- Lateral boundaries provided by NCEP Eta
140-3 h QPF Verification
150-6 h QPF Verification
160-12 h QPF Verification
17JAX WRF QPF Verification
0600 UTC 3, 6, and 9-h QPF Verification, 1 Oct
03 thru 4 May 04
18(No Transcript)
19Vertical Level Configuration
41 s Levels
20LAPS-MM5 Production Cycle
MM5 Forecast (Every 6 h) - 10km to 24h - 3.3 to
12h - 1.1 to 9h
National Model Forecast (Eta via SBN)
Lateral Boundary Conditions
2-9 h Forecast for IC/LBC
Hourly Cycle
MM5 Update Cycle (10km)
1-h Forecast for First Guess
Diabatic Init. Cond.
Observations via SBN and LDAD
21MM5 Forecast Run Timeline
Actual Hour
Valid Hour
All 3 Domains (2.31 Ratio)
Run time based on 12 processors for model 1
processor for each post job
2 Domains (7.81 Ratio)
1 Domain (361 Ratio)
22Cape Canaveral 6-hour QPF on 1-km Grid and Radar
Verification9 Feb 04
23RSA Examples
- Eastern Range
- June 8, 2005
24(No Transcript)
25(No Transcript)
26(No Transcript)
27(No Transcript)
28(No Transcript)
29(No Transcript)
30(No Transcript)
31(No Transcript)
32(No Transcript)
33Series of plots for 15z 10 km run, model
reflectivity is heavy cyan contours, observed is
image
34(No Transcript)
35(No Transcript)
36(No Transcript)
37(No Transcript)
38(No Transcript)
39(No Transcript)
40Analysis and beginning of the 21z/10 km run.
Doesnt get echoes to start.
41(No Transcript)
4215z 3 km run
43(No Transcript)
44(No Transcript)
45(No Transcript)
46(No Transcript)
4715z 1 km run at 1 km cells do develop faster
48(No Transcript)
49(No Transcript)
50(No Transcript)
51(No Transcript)
52(No Transcript)
53Next 2 are from 09z run. Seeing if it got the
overnight land breeze line off the east coast.
54 55Build History
- RSA LAPS/MM5 V1 Dec 2001
- Initial installation at WR
- RSA LAPS/MM5 V2 Aug 2002
- Initial installation at ER
- Included easy install CD, provided to LMMS
- RSA LAPS/MM5 V2.1 July 2003
- Delivered to LMMS on CD
- RSA LAPS/MM5 V2.2 July 2004
- Delivered to LMMS on CD
56Current Status
- Current RSA status
- Fully adapted to AS 2.1 OS
- Testing criteria delivered to LM
- Model status
- Improved Schultz microphysics
- Shallow Cu scheme
57Current Activities
- Current activities
- Data failure test simulated at FSL
- /data/fxa disk mount disabled for 40 minutes
- Model forecast continued to run with degraded
analysis - Analyses were interrupted, changes since made to
help prevent this
58Next few month deliverables
- Deliverables (next few months)
- New LAPS/MM5 build in July 2005
- Accommodates AS 2.1/2.3
- Deliver installation CD
- Test CD installation needed? (what procedure)?
59Future Deliverables
- Future Deliverables
- MM5 to WRF transition
- New modeler on staff July 18, 2005
60Deficiencies in Build V2.1
- Deficiencies
- TKE PBL schemes in MM5 under-predict diurnal
temperature range - Microphysics require saturation in cloudy grid
boxes
61Accomplishments Since June 2004 TIM
- LAPS Analysis Improvements
- QC flag codes 13-20 now interpreted
- Edit flag now checked for AMPS data
- Dense SODAR levels processing improved
- Met TOWER QC now applied
62Accomplishments Since June 2004 TIM
- LAPS Analysis Improvements
- Surface analysis parameters adjusted based on
study of Hurricane Charlie - 3-D reflectivity mapping improved to minimize
artifacts in high-resolution domains - Moisture analysis component changes to use GOES
gradient structure
63LAPS Moisture Variational Step now Relies on
Satellite Gradients
- Other aspects of the moisture analysis have not
changed - The International H2O Project identified problems
with satellite derived product moisture (too
moist, or more specifically moist biased) - Two approaches are now underway to address this
problem - Correct the bias problem in the GOES moisture
data - Modify the analysis system to ignore this
moisture bias by using moisture gradients from
this source
64The new approach is to utilize only satellite
gradients in the analysis
65Accomplishments Since June 2004 TIM
- MM5 Forecast Improvements
- Implementing daily GFS SST analyses
- Provides better forecasts than pseudo-SST
- Provides a first guess for LAPS SST analysis
- Added TKE diagnosis to post-processor
- Allows use of better MRF PBL while still meeting
need for TKE field for HYPACT integration - Updated scripts for new OS (Nov 2004)
- Implemented new Schultz II microphysics
- Deactivated K-F cumulus parameterization
66Accomplishments Since June 2004 TIM
- Verification Package
- Surface observations vs. MM5 forecast
- Requires MySQL database
- Interactive query capability
- Command line or web based
- User customization options
- Being set up on development system for 10-km MM5
grids - Still testing interface scripts
67Recommended Schedule
- Deliver LAPS/RSA v3.0 July 2005
- CD delivery to LMMS with installation scripts
- Includes all changes discussed previously
- Deliver LAPS/RSA v3.1 Jan 2006?
- Incorporate WRF forecast model
- 30-minute analysis cycle
- Adds AMPS ingest
- improved verification package
68Outstanding Issues
- FSL access to RTAMPS data
- Complete real-time data preferable to partial
archived - Helps evaluate spatial distribution and QC
performance - Analysis bulls-eyes and QC
- Land/sea weighting function
- NOAAPORT buoy data timing
69Outstanding Issues (cont)
- Additional SST testing
- Ingest /Analysis of Soil sensor data
- Interface to GOES Sounder on NOAAPORT?
- 30-min LAPS cycle
70Narrowband / Soil
- Narrowband (Level-III) radar
- AWIPS to LAPS NetCDF converter needs repair
- Soil moisture
- Overall Soil Moisture currently in Big File
- Add other soil related fields?
- What soil observations are at Ranges?
- Data formats?
71Verification
- Verification package
- Can run on more powerful headnode
- Requires another software package (MySQL)
- Web server running on headnode for interactive
displays - Forecasters use web browser to access
verification info - Product reliability monitor will be included
72Model output to AMU
- Model output to AMU
- Exportation of bigfile via ftp or LDAD
- Multiple options possible
- Should model grids be enlarged?
- More computer power may be needed for this
73SBN data limitations
- SBN data limitations
- Partially mitigated by use of NCEP SST
- LMMS EDS project should fully mitigate
74Summary
- July build on-track
- Improved LAPS analysis with additional local data
and improved algorithms - Better forecasts from new microphysics, better
SST - Apache server installed for verification display
at ranges - Continue planning for future
75Discussion?