Derived Motion Fields
from the GOES Satellites

Jaime Daniels
(NOAA/NESDIS Office of Research and Applications Forecast Products Development team)

Donald G. Gray
(NOAA/NESDIS Goes Product Manager Office of Systems Development)

Topics

Philosophy
Review of GOES visible, IR, WV channels
Basic methodology
GOES - Next optimized data processing strategies
GOES wind products - What’s new?
Verification
Current and new/planned applications
Summary
Product availability and recommended reading
Discussion/questions

 

Philosophy

 
Assumptions that make the calculation of cloud drift winds possible:

• Clouds are “passive” tracers of winds at a single level
  • use infrared and visible radiances
• Water vapor features (i.e., moisture gradients) are “passive” tracers of winds
  • both in clear air and cloudy conditions
  • use water vapor infrared radiances
• We can properly assign height of tracer

 

 

GOES Visible, IR, WV Channels

Imager

• Water vapor channel 3 (6.7 µm)
• Longwave IR window channel 4 (10.7 µm)
• Visible Channel 1 (0.65 µm)

Sounder

• Water vapor channel 10 (7.3 µm)
• Water vapor channel 11 (7.0 µm)
  

Basic Methodology

• Image Acquisition
• Automated Registration of Imagery
• Target Selection Process
• Height Assignment of Targets
• Target Tracking
• Quality Control (Autoeditor)

 

 

Image Acquisition

• Select 3 consecutive images in time
• Which channels are selected is a function of which wind product (cloud-drift, water vapor, visible) is to be generated
• Extended Northern Hemisphere
• Southern Hemisphere

Auto-registration of Imagery

• Registration is a measure of consistency of navigation between successive images
• Landmark features (i.e., coastlines) must remain stationary from image to image
• Satellite-derived winds are much more sensitive to changes in registration than to errors in navigation
• Navigation of the 3-axis stabilized GOES satellites much more difficult
• Manual registration corrections applied operationally to imagery 5% of the time
• New automated registration QC:
  • hundreds of landmarks used
  • each landmark is sought in all images
  • middle image in loop is assumed to have “perfect” navigation
  • mean line and element correction is computed and possibly applied for the 1st and 3rd image

Target Selection Process

• Consider small sub-areas (target area) of an image in succession
• Perform a spatial coherence analysis of all targets. Filter out targets where:
  • scene is too “coherent”
  • multi-deck cloud signatures are evident
• Locate maxima in brightness
  • Select target/feature associated with strongest gradient
  • Target density is controlled by size of target selector area

Height Assignment of Targets

• Infrared window technique
  • oldest method of assigning heights to cloud-motion winds
  • not suitable for assigning heights of semi-transparent cloud (i.e., thin cirrus)
  • still provides a suitable fallback to other methods
• CO₂ Slicing Technique
  • most accurate means of assigning heights to semi-transparent tracers
  • utilizes IR window and CO₂ (13 µm) absorption channels viewing the same FOV
  • however, CO₂ absorption channel absent on current GOES imagers
• H₂O Intercept Method
  • utilizes WV channel 3 (6.7 µm) and longwave IR window channel 4 (10.7 µm)
  • algorithm: these two sets of radiances from a single-level cloud deck vary linearly with cloud amount
  • adequate replacement of CO₂ slicing method

Target Tracking Algorithm

• Define tracking area centered over each target
• Search area in second image that best matches radiances in tracking area
• Confine search to “search” area centered around guess (AVN Forecast) displacement of target
• Two vectors per target: 1 for images 1 and 2; 1 for images 2 and 3

Quality Control (Autoeditor)

• Functions
  • Target height reassignment
  • Wind quality estimation flag
• Method (4 Steps)
1. 3-dimensional objective analysis of model forecast wind field on 1st pass
2. Incorporate satwinds into analysis on 2nd pass. Remove those differing significantly from analysis
3. Target heights readjusted by minimizing a penalty function that seeks the optimum “fit” of the vector to the analysis
4. Perform another 3-dimensional objective analysis (at reassigned pressure) and assign quality flag

Height Assignment Related to Satellite Wind Type
(Approximations)

High Density Water Vapor GOES Winds

GOES High Density Cloud Drift Winds

GOES High Density Winds
(Cloud Drift, Imager H₂O, Sounder H₂O)

GOES High Density Visible Winds

GOES High Density Visible Winds:
Tropical System Circulations

Optimal Data Processing Strategies

• Take advantage of new sensor technology
  • silicon photodiode detectors (improved signal-to-noise)
  • higher spatial resolution and bit depth
  • improved spectral sampling and sampling rates
• Take advantage of automation techniques and processing power
  • eliminate manual labor-intensive tasks
  • increase data volume
    

Optimal Data Processing Strategies

• Take advantage of improved viewing capability
  • temporal sampling (including rapid scans)
  • independent imager and sounder
• Optimize processing strategy
  • high data volume/density (x, y, z, t) coverage
  • multi-spectral data integration (H₂O winds)
  • multi-satellite (data fusion)

Optimal Data Processing Strategies

• Focus processing strategy towards the meteorology
  • circulations and environmental features
• Adapt the data quality control
• Take advantage of improved communications
  • timely data dissemination

GOES-10 Visible Winds Impact
of Higher Sampling Rates

GOES Wind Products: What’s New?

Current and New/Planned Applications

• Mid-latitude Oceanic Analyses
  • NWS offices have access to high density wind products via internet; AWIPS access to follow
• Numerical Weather Prediction (NWP) and Data Assimilation
  • What’s happening at NCEP/EMC?
  • ECMWF is utilizing GOES high density wind products

Current and New/Planned Applications

• Tropical cyclone analysis and forecasting
  • Tropical Prediction Center (TPC) has access to the GOES multi-spectral wind data sets
  • GFDL & NRL are performing model impact studies using the GOES multi-spectral winds
    to improve tropical storm track forecasts
  • CIMSS routinely generating water vapor and visible winds from GMS-5

NWP and Data Assimilation

EMC Status/Plans:

• Operational use of high density cloud drift winds in global and
  regional forecast models began in December 1997.
• Evaluation of high density water vapor (imager and sounder) and
  visible winds planned for 1999 -- focus on assimilation of layer wind estimates

NWP and Data Assimilation

NESDIS Status/Plans:

• Routine production of GOES sounder WV and VIS winds began in late 1997
• Work with EMC to support evaluation in EMC operational database in 1999
• NESDIS/CIMSS and FSL will coordinate on model impact study involving
  the generation of multi-spectral (vis, ir, wv) winds and their
  assimilation into the MAPS/RUC models

Verification

• Sources of errors in satellite-derived winds
• Satellite winds vs. rawinsondes vs. model collocation statistics
• Model impact studies
• Satellite minus forecast wind field
• Mean tropical storm track forecast errors

Comparison of Model Forecast
and Satellite Derived Wind Fields

yellow=AVN forecastred=AVN forecast + sat winds

Impact of GOES Winds - Hurricane Edouard 1996

Impact of GOES Winds - Hurricane Fran 1996

Sources of Errors

• Assumption that clouds and water vapor features are passive tracers
  of the wind field
• Image registration errors
• Target identification and tracking errors
• Inaccurate height assignment of target

Summary

• Higher resolution data, improved science, and full automation--resulted in satwinds
  that are superior in both quality and quantity to any done
  previously at NOAA/NESDIS
• Improved automated QC is the most significant change in the winds
  processing systems over the past 5 years
• Improved target selection avoids mix-level scenes and concentrates on providing greater targeting density for features of interest.
• Water vapor intercept method. Numerous applications

Product Availability

E-mail:
jdaniels@nesdis.noaa.gov
Donald.G.Gray@noaa.gov

Websites:
http://cimss.ssec.wisc.edu
http://orbit35i.nesdis.noaa.gov/goes/winds/

Reference Material

• Nieman et al., 1997: Fully automated cloud-drift winds in NESDIS operations.
  Bull. Amer. Meteor. Soc., 78, 1121-1133.

• Velden et al., 1997: Upper-tropospheric winds derived from geostationary satellite water vapor observations. Bull. Amer. Meteor. Soc., 78, 173-195.