observing system simulation experiments at cimss
DESCRIPTION
Observing System Simulation Experiments at CIMSS. By CIMSS/OSSE Team : Bob Aune ; Paul Menzel ; Jonathan Thom Gail Bayler ; Chris Velden ; Tim Olander and Allen Huang Cooperative Institute for Meteorological Satellite Studies University of Wisconsin 7 June, 1999. Current Research Plan. - PowerPoint PPT PresentationTRANSCRIPT
Observing System Simulation Experiments at CIMSS
By
CIMSS/OSSE Team :
Bob Aune ; Paul Menzel ; Jonathan Thom
Gail Bayler ; Chris Velden ; Tim Olander
and Allen Huang
Cooperative Institute for Meteorological Satellite Studies
University of Wisconsin
7 June, 1999
MESOSCALE OBSERVING SYSTEMSIMULATION EXPERIMENTS
(OSSE)GOAL
To assess the contribution of environmentalobserving systems to operational mesoscalenumerical weather forecasts in a controlledsoftware environment.
Future observing systems can be tested usingprojected instrument characteristics.
Current Research Plan
Initial Impact Study: Geostationary Interferometer
Simulated Products from “Nature” Atmosphere:
Soundings (T, Td)
Winds (Cloud drift / Water Vapor)
Soundings plus Winds
Soundings, Winds, Conventional Data
Radiances
Derived Products from Simulated Radiances
Retrievals (T, Td, derived products)
Winds (automated wind algorithm)
Direct radiance assimilation
PROCEDURES
Observations are synthesized from forecastsgenerated by a numerical prediction model thathas a known history calibrated against reality.
These forecasts represent truth and are referred toas the "nature" atmosphere.
Synthesized observations must mimic, as close aspossible, observations from the real observingsystem that is being evaluated.
Synthesized observations are assimilated into anassimilation system that is independent of the"nature" model.
Pilot Experiment
HYPOTHESIS:Information from a geostationary-basedinterferometer will significantly improve theaccuracy of numerical weather forecasts over thecurrent geostationary radiometer.
Temperature and moisture retrievals aresimulated by superimposing estimatedobservation errors on the "true" profilesgenerated by the "nature" run.
OSSE Design
An OSSE can be subdivided into four basic steps:
1) Generate a "nature" atmosphere
2) Compute synthetic observations
3) Assimilate the synthetic observations
4) Assess the impact on the resulting forecast
Each step is performed with the goal of minimizing anyexternal influences, which may compromise the value of thesynthesized observations, the assimilation process, or theresults of the numerical forecasts.
This OSSE is being conducted over a limited areadomain. The influence of pre-specified lateral boundariesmust be minimized.
1. Generate a "nature" atmosphere
University of Wisconsin, Nonhydrostatic Modeling System The horizontal domain is chosen to be as large as practical
to isolate the influence of the pre-defined lateral boundaryconditions. Horizontal resolution = 60 km.
Boundary conditions: NCEP Eta forecast model, NCEP 104grid.
Ideally, the "nature" atmosphere should be two to four timesthe resolution of the simulated observing system.
The model vertical resolution is chosen to be a minimum oftwo-times the resolution of the observing system to besimulated. Vertical levels = 38.
A 12hr forecast was generated to allow the model to "spinup".
UW-NMS Domain in the Eta 104 grid
“Nature” Calibration
2. Simulate observations
Temperature and moisture profiles from the "true"atmosphere are modified using realistic observationerrors.
Profiles of temperature and moisture are generated athourly intervals over the 12-hour analysis period.
A cloud mask is used to simulate gaps in the coverage.
NMS Grid Locations with Cloud Mask
RAOBS Surface
ACARS Profilers
Observation Errors
Ob type Count RMS Error BIASRAOB Temperature 98* 0.3 CRAOB Height 98* 8-32 mRAOB Dewpoint 98* 0.5 CRAOB Wind 98* .8 - 1.3 m/sSFC Temperature ~600 0.3 CSFC Dewpoint ~600 0.5 CSCF Wind ~600 0.4 m/sACARS Temperature ~3000 1.0 CACARS Wind ~3000 1.0m/sProfiler Wind 31* 1.0 m/sGEO-R Temperature ~3500* 1.9 - 2.1 C .27 CGEO-I Temperature ~4000* ~1.0 C 0.1 CGEO-R Mixing ratio ~3500* ~1.0 g/Kg .053 g/KgGEO-I Mixing ratio ~4000* ~0.5 g/Kg .02 g/Kg
(* indicates a profile)
Satellite Wind Errors
Ob type Level GEO-R GEO-IWinds, clear Count ~7000 ~10000
200mb na 3.5 m/s300mb 5.0 m/s 3.2 m/s400mb 4.5 m/s 3.0 m/s500mb 4.0 m/s 2.6 m/s700mb na 2.0 m/s
Winds, cloudy Count ~2000 ~4000200mb 4.5 m/s 3.0 m/s300mb 4.0 m/s 2.6 m/s400mb 3.5 m/s 2.3 m/s500mb 3.5 m/s 2.3 m/s700mb 3.0 m/s 2.0 m/s850mb 2.5 m/s 2.0 m/s
Simulated Error for TemperatureGEO-I GEO-R
0
100
200
300
400
500
600
700
800
900
1000
1100
0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5
Degrees C
Pre
ssu
re h
Pa
3. Assimilate the synthesized observations
The operational 40km Rapid Update Cycle (RUC) wasused to assimilate the observations at hourly intervals.
Boundary conditions: NCEP Eta model, projected ontothe AWIPS 211 grid (80km resolution).
Four assimilation experiments were performed:
1) Conventional observations only (CONV)
2) Geostationary radiometer (GEO-R); assimilate profilesadjusted to emulate a GOES-type system
3) Geostationary interferometer (GEO-I): assimilate profilesfrom a proposed geostationary interferometer
4) Perfect observation (BEST); assimilate "true" profilesextracted directly from the "nature" run
Note: The CONV and BEST experiments represent the range ofperformance that can be expected from the RUC system.
4. Assess the impact of the observations on theresulting forecasts
The impact of the observations will be assessed byobjectively measuring the ability of each observing system tosteer the resulting 12-hour forecasts toward the “true”atmosphere
Sensitivity of RUC analysis to retrieval density
GEO-R versus GEO-I
Retrieved T and Td
GEO-R versus GEO-I
Retrieved T and Td
GEO-R versus GEO-I
Retrieved T and Td
GEO-I results are significantly improvedover those from the GEO-R.
500 hPa temperature errors are reducedby 0.2 C root mean square (rms) over theextended CONUS (contiguous UnitedStates) and 700 hPa relative humidityerrors are reduced by 2%.
To compare the impact of the geostationaryinterferometer against the geostationary radiometer arelative score was computed using the No Observationrun (NO) and the Perfect Observation run (PO) tonormalize the verification statistics.
The RMS errors for temperature and relative humiditywere summed over four layers (700hPa, 500hPa, 400hPa,300hPa) and normalized between the RMS error sumsfrom the NO run and the PO run. A score of 10 matchesthe PO run.
Soundings + Winds 700hPa RH Validation
2
4
6
8
10
12
14
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
RMSE
(%)
CONV
GEO-R
GEO-I
BEST
Soundings + Winds 700hPa RH Validation
-6-5-4-3-2-1012
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
Bia
s (%
)
CONV
GEO-R
GEO-I
BEST
Soundings + Winds 700hPa RH Validation
30
35
40
45
50
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
S1 S
core
CONV
GEO-R
GEO-I
GEO-R versus GEO-I
Retrieved T and Td
Sat Winds
Conventional
Soundings + Winds 850hPa RH Validation
25
30
35
40
45
50
55
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
S1 S
core
CONV
GEO-R
GEO-I
Soundings + Winds 850hPa RH Validation
0
5
10
15
20
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
RM
SE
(%)
CONV
GEO-R
GEO-I
BEST
Soundings + Winds 850hPa RH Validation
-8
-6
-4
-2
0
2
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
Bia
s (%
)
CONV
GEO-R
GEO-I
BEST
GEO-R versus GEO-I
Retrieved T and Td
Sat Winds
Conventional
Soundings + Winds 500hPa T Validation
-0.4
-0.2
0
0.2
0.4
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
Bias
(deg
C0
CONV
GEO-R
GEO-I
BEST
Soundings + Winds 500hPa T Validation
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
RMSE
(deg
C) CONV
GEO-R
GEO-I
BEST
Soundings + Winds 500hPa T Validation
35
40
45
50
55
60
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
S1 S
core
CONV
GEO-R
GEO-I
GEO-R versus GEO-I
Retrieved T and Td
Sat Winds
Conventional
Soundings + Winds 300hPa T Validation
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
Bias
(deg
C)
CONV
GEO-R
GEO-I
BEST
Soundings + Winds 300hPa T Validation
0
0.2
0.4
0.6
0.8
1
1.2
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
RMSE
(deg
C) CONV
GEO-R
GEO-I
BEST
Soundings + Winds 300hPa T Validation
35
40
45
50
55
60
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
S1
Sco
reCONV
GEO-R
GEO-I
GEO-R versus GEO-I
Retrieved T and Td
Sat Winds
Conventional
Soundings + Winds 300hPa U
1
1.5
2
2.5
3
3.5
4
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
RMSE
(m/s
)
CONV
GEO-R
GEO-I
BEST
Soundings + Winds 300hPa U
30
35
40
45
50
55
60
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
S1 S
core
CONV
GEO-R
GEO-I
Soundings + Winds 300hPa U
-0.20
0.20.40.60.8
11.21.4
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
Bia
s (m
/s)
CONV
GEO-R
GEO-I
BEST
GEO-R versus GEO-I
Retrieved T and Td
Sat Winds
Conventional
Soundings + Winds 300hPa V
1
1.5
2
2.5
3
3.5
4
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
RM
SE
(m/s
) CONV
GEO-R
GEO-I
BEST
Soundings + Winds 300hPa V
-0.6-0.4-0.2
00.20.40.60.8
1
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
Bias
(m/s
)
CONV
GEO-R
GEO-I
BEST
Soundings + Winds 300hPa V
30
35
40
45
50
55
60
0 2 4 6 8 10 12 14 16 18 20 22 24
Hour
S1
Sco
reCONV
GEO-R
GEO-I
GEO-R versus GEO-I
Retrieved T and Td
Sat Winds
Conventional
Plans for the Future
* Simulate winds from radiances
* Assimilate retrievals from radiances
* Assimilate radiances with 3DVar
* 14 day test periods (winter and spring)
* Resolve boundary condition issues
* Low Earth Orbit (LEO) OSSE
* Test other observing systems
Wind Experiment Using Simulated Radiances
1) Simulate radiances from GOES and from a geostationary interferometer using forward radiative transfer.
2) Put simulated radiances into the automated wind algorithm and generate cloud drift and water vapor winds
Hurricane Bonnie Wind and Cloud Fields
Wind Vectors :
Red - 1 km level
Green - 14 km
level
Clouds :
Light gray -
Ice Cloud
Dark Gray -
Water Cloud
GOES Radiances Simulation Verification
Wind Tracking Verification
Wind Tracking Verification - Continued
Tracking Interferometer Radiances
Tracking Mixing Ratio from Model