Date post: | 30-Dec-2015 |
Category: |
Documents |
Upload: | myra-reeves |
View: | 214 times |
Download: | 0 times |
Quantitative Design: The Right Way to Develop the Composite Observing System
A presentation to the GOES R Conference
Alexander E. MacDonald
NOAA Forecast Systems Lab – Boulder
May 11, 2004
Quantitative Design: The Right Way to Develop the Composite Observing System
Talk Summary
1. Observing subsystems MUST BE TREATED AS PART OF A COMPOSITE SYSTEM – not as stand alone systems.
2. We are now capable of credible simulations:
* Continental Scale
* Global Scale
3. Quantitative design is the right way to develop the composite observing system.
Quantitative Design: The Right Way to Develop the Composite Observing System
Talk Summary
1. Observing subsystems MUST BE TREATED AS PART OF A COMPOSITE SYSTEM – not as stand alone systems.
2. We are now capable of credible simulations:
* Continental Scale
* Global Scale
3. Quantitative design is the right way to develop the composite observing system.
Global Hawk could be the Unmanned Aerial Vehicle platform:
* Range: 14,000 miles
* Speed: 350 knots
* Altitude: 60,000 feet
* Payload: 1960 lbs
* Lease Cost: $ 4 M /plane
* Year ops: $ 3 M per plane
* ConOP: 2 aircraft, 25% duty cycle
* Prime: Northrop
AEM in situ
System Description:
The main idea of the Global Unified Profiling System is to take the most accurate possible profiles from the stratosphere to deep in the ocean over as much of the earth as possible. (Land too!)
The profiles should include state (T,p,u,v,q in atmosphere, temperature, current and salinity in the ocean), forcing, and chemistry.
Quantitative Design: The Right Way to Develop the Composite Observing System
Talk Summary
1. Observing subsystems MUST BE TREATED AS PART OF A COMPOSITE SYSTEM – not as stand alone systems.
2. We are now capable of credible simulations:
* Continental Scale
* Global Scale
3. Quantitative design is the right way to develop the composite observing system.
Results From the FSL Regional Lidar
OSSE•NOAA/FSL
- Steve Weygandt- Stan Benjamin- Steve Koch- Tom Schlatter- Adrian Marroquin- John Smart- Dezso Devenyi
• NOAA/NWS/NCEP - Michiko Masutani
• NOAA/ETL - Mike Hardesty- Barry Rye- Aniceto Belmonte- Graham Feingold
• NCAR- Dale Barker- Qinghong Zhang
Relationship between Global and Regional OSSEs
Global Nature Run
(ECMWF)
Global AssimilationRun (GFS)
Regional Nature Run
(MM5)
Regional Assimilation Run (RUC)
Global
Regional
Nature Run
AssimilationRun
Simulated Observations
Boundary Conditions
Boundary Conditions
Simulated Observations
Lidar Data Coverage
• Three satellite swaths per 12 h
• Profiles of two VLOS components at each point
0300UTC
0130UTC
0000UTC
0430UTC
OBSERVATION DATA COUNTS
Ob type Variables 12z 15zRaob (Z,T,Q,U,V) 3700 0Prof/VAD (U,V) 2600 2600ACARS (T,U,V) 1200 1300METAR/Buoy (T,Q,U,V) 1600 1600
Lidar (Vr) 1500 1500
Approximate no. of obs data points
• Lidar adds ~8% more wind obs at raob init times (00z, 12z)
• Lidar adds ~14% more wind obs at non-raob init times (06z, 18z)
Regional OSSE CalibrationDoes simulated-data impact (OSSE)
match real-data impact (OSE) for an existing observation type?
Real DataVerify against raobs
4-16 Feb 2001
15-20 Feb 1993
Simulated Data
Verify against nature run
Compare real-data and simulated-data ACARS denial
ACARS denial yields similar % degradation for real-data and OSSE simulated-data
Normalize Errors
NEGATIVE VALUE % degradation
POSITIVE VALUE % improvement
CNTLerror – EXPerror
CNTLerror
Impact of denying ACARS obson 6-h fcst vector wind RMSE
% degradation
• Lidar obs improve fcst more at non-raob init times• Lidar obs improvement greatest aloft
6- hourforecast
Non-raobinit time(06z,18z)
Raobinit time(00z,12z)
Assimilation of lidar observations
(but no lidar obs in boundary conditions)
Impact of adding lidar obson 6-h fcst vector wind RMSE
% improvement% degradation
The relative impact of the profiler data
0 5 10 15 20 25 30 35
100
200
300
400
500
600
700
800
900
"Relative" impact of ACARS and profiler datacompared to nodata exp on 3h RUC forecasts
profilerACARS
( denial - cntl ) / (nodata - cntl) X 100%
pre
ss
ure
(m
b)
profiler domain01035-01047 average
3-h Model forecast improvement
% improvement due to profiler and ACARS data
Profiler/ACARS impact calibrated by difference between 13-day experiments with all data and no observations (lateral boundary conditions only)
Fact: Profilers are the best data source for the lower part of the atmosphere within the network.
Satellite image taken at 0045 UT, during tornado outbreak.
When the profiler data is included, it doubles the “storm energy” that was predicted for the May 3, 1999 Oklahoma tornadoes.
Observing System Simulation must be an important part of our efforts to add new observing capabilities on the geostationary satellites.
The Potential Impact of Space-based Lidar Winds on Weather Prediction: Update on recent experiments at
the NASA DAO
Robert Atlas
Data Assimilation OfficeNASA Goddard Space Flight Center
Quantitative Design: The Right Way to Develop the Composite Observing System
Talk Summary
1. Observing subsystems MUST BE TREATED AS PART OF A COMPOSITE SYSTEM – not as stand alone systems.
2. We are now capable of credible simulations:
* Continental Scale
* Global Scale
3. Quantitative design is the right way to develop the composite observing system.
Important Community Efforts Should Embrace Quantitative Design of Observing Systems:
• Joint Center for Satellite Data Assimilation
• NCEP
• OAR Boulder Labs
• SSEC
• NAVY
• University community
• International community
• etc