Use of Aqua Satellite Data to Improve Shortterm Weather Forecasts

1
Use of Aqua Satellite Datato Improve Short-term
Weather Forecasts

• Keith Brewster Center for Analysis and
  Prediction of Storms
  
• University of Oklahoma

Presented at ASPRS Central Fall Technical
Session, September 28, 2006Corresponding author
kbrewster_at_ou.edu
2
Collaborators

• Shanna SampsonOU School of Meteorology MS
  Student
  
• Dr. Fred Carr, OU School of Meteorology
• Gary Jedlovec and Brad ZavodskyNASA SPoRT

3
Aqua Satellite

• Launched May 2002
• Provides data to improve the specification of the
  earths hydrological cycle
  
• Polar orbiting satelliteSun-synchronous orbit,
  revisits twice per day, 100am LST and 100pm
  LST
  
• Low-Earth Orbit, 730 km

4
Aqua Instruments - AIRS

• Atmospheric Infrared Sounder
• Measures upwelling infrared energy from the earth
  and atmosphere at high spectral resolution
  
• Grating Spectrometer
• 2378 Channels
• 3.7-15.4 mm
• 13.5 km footprintat nadir

5
Data Sampling by AIRS
-NASA Animation
6
Aqua Instruments AMSU/A

• Advanced Microwave Sounding Unit
• 15 Microwave Channels 15-90 GHz
• 40 km footprint at nadir
• All-Weather

7
(No Transcript)
8
Radiation Measurements to Vertical Soundings

• AIRS and AMSU data combined to create vertical
  soundings of temperature and humidity
  
• Air and/or water vapor at various heights
  (pressures) contribute to the total radiation
  measurement viewed from space.
  
• The contribution peaks at different pressures for
  different wavelengths

9
Sample Weight Functions For a Few IR Channels
10
Radiation Measurements to Vertical Soundings

• AIRS and AMSU data combined to create vertical
  soundings of temperature and humidity
  
• Air and/or water vapor at various heights
  (pressures) contribute to the total radiation
  measurement viewed from space.
  
• The contribution peaks at different pressures for
  different wavelengths
  
• Inversion processed used to solve for air
  temperatures and water vapor from the radiances
  
• Quality control of cloud-contamination is critical

11
Project Motivation Goal

• The ability of the satellite to obtain data over
  the Gulf of Mexico promises to improve forecasts
  of air mass modification and return flow from the
  Gulf
• Aim to improve short-term numerical prediction of
  high impact weather events such as severe
  thunderstorms and flash floods
  

12
Two Parts of AIRS Study

• Examine soundings to quantify the inherent
  smoothing, any biases and standard error
  
• Test impact of AIRS on the analysis and forecast

13
Examination of the Data

• Use retrieved soundings
• Require some knowledge of the error associated
  with the data in order to use it properly in data
  assimilation
  
• The statistics computed deal with 10 of the
  levels provided between 925-150 mb

14
Comparison Soundings

• AIRS retrievals reported as point observations
• Due to nature of radiation measurements on which
  they are based, the values are representative of
  temperatures over layers
  
• Smooth validation sounding data before comparing
  with AIRS in order to find the filter parameter
  that best matches the AIRS data

15
Exponential Weighting

• Use exponential function (as in Barnes analysis)
  as the filter of choice to smooth sounding in the
  vertical

• rm pressure difference between mth
• observation and AIRS point (units mb)
• k is the filter shape parameter (units mb)
• Comparison soundings first interpolated to
• 1 mb increments to be evenly spaced

16
Two Sources of Comparison Data

• Atmospheric Radiation Measuring Program (ARM)
  Southern Great Plains (SGP) site at Lamont, OK
  land
  
• August 20, 2005 April 19, 2006
• ARM Tropical Western Pacific (TWP) site on Nauru
  Island ocean
  
• September 16, 2005 April 17, 2006

17

• Use soundings launched 5 min before Aqua
  overpass
  
• Not a rigorous validation, but a sample dataset
  chosen to get an estimate of statistics to be
  used in data assimilation
  
• 70 km limit for collocation

18
Temperature Filter Match
Ocean (TWP)
Land (SGP)
19
Relative Humidity Profiles
Land (SGP)
Ocean (TWP)
20
RMS RH
SGP
TWP
QC Flag Top, Mid, Bot, Sfc
21
RMS T
SGP
TWP
QC Flag Top, Mid, Bot, Sfc
22
Summary of Results for Validation

• Found a value of filter parameter, k, that
  minimized error for T and RH.
  
• Satellite data fit smoothed profiles better
• With proper filter matching most levels meet
  instrument goals of 1K RMS and 20 relative
  humidity RMS
  
• TWP-AIRS Ocean soundings agree better than
  SGP-AIRS Land

23
Impact of AIRS on Analysis and Forecast

• First look at impact on initial analyses
• Want to know if addition of AIRS temperature and
  moisture profiles over the ocean improve a high
  resolution forecast
  
• Impact on humidity analysis
• Impact on thunderstorm forecast
• Use ARPS model

24
ADAS

• Use ARPS Data Analysis System (ADAS) to
  assimilate the soundings into ARPS
  
• ADAS is a Bratseth successive correction
  statistical analysis that converges to optimal
  interpolation.
  
• Flexible system of ingesting data having varying
  sources and observation densities.
  
• Error characteristics of the data can be
  specified by each source and by height above
  ground level.
  
• Includes complex cloud analysis procedure that
  integrates cloud information from surface
  stations, visible and IR satellite data, and
  radar reflectivity.

25
Case April 9, 2005

• The NAM (Eta) model underpredicted the moisture
  return along the Gulf coast of Texas on the day
  preceding an outbreak of severe weather in
  northeast Texas and eastern Oklahoma
• The Aqua satellite passed over the region around
  19 UTC on April 9 and therefore 19 UTC is used as
  the initialization time of the ARPS model
  
• Archived NAM forecasts with 40-km resolution used
  as background field

26
Corpus Christi, TX Soundings
12z to 00z Change in Observation
NAM vs Observation 00z
27
Aqua MODIS Composite Image19Z 09 April 2005

• Used AIRS soundings over the ocean
• Clear overpass, very little cloud cover

28

• Other Sources
• Surface Aviation Observations (METAR)
• Buoy
• Model Resolution
• 12 km horizontal resolution for ADAS analyses and
  boundary conditions, 3 km resolution for
  forecasts
  
• 350 m average vertical resolution

29
ADAS Analyses
Name Key N No Modification B Bias Correction
E Updated Error Information
S Smoothing of Background O Ocean
30
850 mb Specific Humidity 19Z
28V4.0 BEBias Removed
28V4.0 BESOBias RemovedBkgd Smoothed
28V4.0 N No Modification
31
850 mb Specific Humidity Difference Fields
(28V4.0 N-CTRL)
(28V4.0 BE-CTRL)
(28V4.0 BESO-CTRL)
32
Surface Specific Humidity Difference Fields
(28V4.0 N CTRL)
(28V4.0 BE CTRL)
(28V4.0 BESO CTRL)
33
Summary of ADAS Analyses

• Increase in moisture at 850 mb
• Decrease in moisture at the surface
• Greatest increase at 850 mb when bias correction
  is applied

34
ARPS Forecasts

• Use 28V4.0 BESO and CTRL to produce two separate
  forecasts
  
• Model initially run at 12 km
• Use 12 km run as background and boundary
  conditions for storm-resolving3 km grid
  forecast
  

35
12-hour Forecast Specific Humidity850 mb
Forecast
Difference Field 28V4.0 BESO - CTRL
36
24-hour Forecast Specific Humidity850 mb
Forecast
Difference Field28V4.0 BESO - CTRL
37
Forecast DifferencesSurface Specific Humidity
12-hour
24-hour
38
Reflectivity Plots at 2200 UTC on 10 April 2005
CTRL
28V4.0 BESO
Actual
39
Reflectivity 00 UTC 11 April 2005
Actual
CTRL
28V4.0 BESO
BESO w/o 925
40
Reflectivity Plots No AIRS at sfc and 925 mb
2200 UTC 10 April 2005
0000 UTC on 11 April 2005
41
Summary and Conclusions

• Possible reasons impact not stronger
• Significant increase in moisture at 850 mb, but
  some decrease in moisture at surface
  
• Small improvement if 925 and sfc levels omitted
• Aqua overpass may have missed deepest of
  modifying air mass as it did not cover extreme
  western portion of Gulf on this pass
  

42
Future AIRS Work

• Verify against surface dataExclude surface and
  buoy data from analysis
  
• Compare Filter Response to Individual Band Weight
  Functions
  
• Explore means to identify when 925-Sfc data may
  be validImproved moisture and QC flags in next
  version of retrievals
  
• Use SST data in combinationEarlier overpass to
  allow for BL mixing?
  
• Study additional cases

43
Aqua Instruments AMSR-E

• Advanced Microwave Scanning Radiometer for EOS
• 12-Channels, 6 frequencies 6.9-89.0 GHz
• dual-polarization
• 5.4-56 km footprint at nadir
• All weather

44
AMSR-E Products

• Precipitation Rate
• Cloud Water
• Surface wind speed over oceans
• Sea Surface Temperature
• Ice, Snow and Soil Moisture

45
Concept Extending Radar Reach
46
Fusion of Different Data Sources

• NEXRAD Network Radars3-D profiles of
  precipitation hydrometeorsLimited to 230 km
  radius from radar
  
• Surface ObservationsCloud base heights over land
  sites
  
• GOES Satellite Visible2-D Cloud-vs-No Cloud in
  daylight
  
• GOES Satellite 10mm IR2-D Cloud-top Temperature
• AMSR-E Satellite2-D surface rainfall estimates

47
Method Schematic
48
Hurricane Ivan, 2004
49
Initial Condition
Without AMSR-E
With AMSR-E
50
Cloud Cross-Sections
West-to-East Cross Section
South-to-North Cross Section
51
30-min Forecast
Without AMSR-E
With AMSR-E
52
1-Hour Forecast
Without AMSR-E
With AMSR-E
53
Compared to Radar Composite
Coastal Radar at 20Z
With AMSR-E at 20Z
54
AMSR-E Rainfall Conclusions

• Successfully integrated AMSR-E rainfall
• Improved spin-up of model to only a few minutes
• Model with cloud analysis without AMSR-E data
  able to spin-up hurricane rainfall on its own in
  about one hour

55
AMSR-E Future Work

• Use additional AMSR-E data products
• Combine with AIRS data products
• Verify longer model forecasts
• Try with early stages of Hurricane Katrina just
  east of Miami