Operational Scatterometry 1

1
NOPP Operational Scatterometry Activitiesat
COAPS
Mark Bourassa Center for Ocean-Atmospheric
Prediction Studies The Florida State
University With contributions from Kyle
Hilburn, Jim OBrien, and Ryan Sharp
2
Partners

• James J. OBrien COAPS
• Mark A. Bourassa COAPS
• Robert L. Bernstein Sea Space Corporation
• Paul Chang NOAA/NESDIS
• Mike Clancy Fleet Numerical Met. and Oceanog.
  Center
• John Kindle NRL-Stennis
• James P. Rigney Naval Oceanographic Office
• David E. Weissman Hofstra University, Dept. of
  Engineering

3
SeaWinds on QuikSCAT

• Satellite
• The QuikSCAT satellite
• Polar orbiting satellite
• Average orbital height of approximately 800 km
• One orbit in approximately 100 minutes
• Sensor
• SeaWinds scatterometer
• Active microwave sensor
• Responds to short water waves
• Which respond very rapidly to changes in vector
  wind
• Measures wind speed and direction

4
SeaWinds Daily (22 hour) Coverage
Ascending Node
Descending Node
From Paul Chang (NOAA/NESDIS) http//manati.wwb.n
oaa.gov/quikscat/
5
Latency of Near Real-Time QSCAT Winds
Estimated from an 8 day average (Feb. 21 March
1, 2002). Delays greater than 6 hours result in
missing data, and are not considered here.
6
Missing Data (1 day Feb. 19, 2002)Relative to
RSSs Science Quality Data
7
Missing Data (8 days Feb. 25 March 3)Relative
to RSSs Science Quality Data
8
Missing Data (8 days Feb. 15 22)Relative to
RSSs Science Quality Data
9
Research Quality Global Gridded Wind Fields

• Objectively derived, 1x1?, 4x daily, research
  quality
• based solely on SeaWinds observations (Pegion et
  al. 2000 MWR).
• Available through COAPS web site in data files
  and animations.

• Animations for 37 regions
• Various spatial scales
• One week per animation
• Available through our web site
• Example
• Hurricane Lenny (1999)
• Formed late in the season
• Moved opposite the usual direction

10
The Calm Mediterranean Sea
11
Main Objectives (1)

• Improved scatterometer winds (near real-time)
• Better rain flag
• Currently, winds are grossly over-flagged
• Determine when rain leads to substantial errors
  (Weissman et al., JAOT, 2002) and flag
  accordingly
• Methodology related to
• SeaWinds-derived surface pressures (Zierden et
  al., JGR, 2000)
• SeaWinds estimates of rain rate
• Development of scatterometer-based pressure maps
  (in and near swath)
• Forecasters find surface pressure maps very easy
  to work with
• Comparison to model pressures helps evaluate
  model product
• Based on extension of current pilot study

12
Main Objectives (2)

• Development of slightly delayed gridded wind and
  stress products
• Method adapted from Pegion et al. (MWR, 2000)
• Global product
• delayed 28 hours
• Local products
• delayed lt4 hours

• Test the utility of these fields
• Ocean model of the Gulf of Mexico (in house)
• Regional model off California (collaborators)

13
Additional Activities

• Early detection and location of tropical cyclones
• Apply a vorticity-based test to locate potential
  tropical cyclones (Sharp et al. 2002, BAMS)
• Use scatterometer-derived pressures to better
  locate the system
• Working in collaboration with NHC and AOMLs HRD

14
Example Rain Flag Problems

• Over-flagging
• Areas of rain, where the rain impact is small
• Missed serious rain impacts

15
Radar Cross-Section Wind and Rain
Wind 3 m/s
In collaboration with David Weissman and Jeff
Tongue (Weissman et al. 2002, JAOT).
16
Computing Surface Pressure

• ETA model mean sea-level pressure is used as the
  background field, interpolated in time to
  coincide with the latest swath (usually falls
  between the model initialization and the 24 hour
  forecast).
• Winds are rotated and adjusted to free atmosphere
  values.
• Converted from gradient winds to geostrophic
  winds.
• Relative vorticity is computed from the adjusted
  satellite winds.
• Variational method assimilates the QuikSCAT
  relative vorticity in the swath, treating it as
  geostrophic vorticity.
• Smoothing and kinematic constraints adjust
  correct model pressures near the swath.
• Boundary Conditions
• ETA pressures are assumed correct on boundaries
  over land.
• ETA pressure gradients are assumed correct on
  boundaries over water.

17
Near Real-Time Winds and Pressures
18
Near Real-Time Winds and Pressures (24 hours
later Snow forecast bombs)
19
NorEaster (Dec. 20, 2001)
20
NorEaster (Dec. 20, 2001)
21
Scatterometer-Derived PressuresTS Keith
Statistics Best Track20.8N 94.9W988 mb70
mph QSCAT20.75N 94.75W989.9 mb Development of
this technique was inspired by Patoux and Brown
(2001)
22
Differences in Position Relative to Best Track
Locations
23
Hurricane Isaac 21Z, Sept. 26, 2000
kts
24
Vorticity as a Detector
Ascending swaths of 19 September 1999
25
Methodology

• Calculate the Average Vorticity
• Determine local vorticity based on swath wind
  vectors
• Average these vorticities within a 7 by 7 (175
  km) in-swath box

Pre-Emily
15 m s-1
15 m s-1
Local Vorticity
Averaged Vorticity
14N
14N
12N
12N
10N
10N
50W
48W
46W
44W
50W
48W
46W
44W
5
10
15
20
25
X10-5 s-1
Vorticity
26
Threshold Test

• Criteria for detection must be determined
• Subjectively Derived from the 1999 Atlantic
  Hurricane Season (storms had to be directly hit
  by the swath and free from any landmasses)
• Three Criteria for identifying a potential TC
• Average vorticity must exceed 10 X 10-5 s-1
• The maximum rain-free wind speed in the averaging
  box must exceed a certain minimum wind speed (we
  selected 10 m s-1)
• The above two criteria must be met at least 25
  times within a system (area of 15000 km2)
• Domain includes the Gulf of Mexico, the Caribbean
  Sea, the tropical Atlantic from 10oN to 25oN, and
  the Eastern tropical Pacific.

27
Detection Times Relative to NHC Classification

2001 Atlantic Season
Moved out of study domain
2001 Eastern Pacific Season
28
Statistics for the Detection Method

• Atlantic TCs
• TC in Swath?
• Yes No
• Method Yes 56 32
• Detects a
• TC in No 10 1230
• Swath?
•  
• Probability of Detection 0.85
• False Alarm Rate 0.36
• Critical Success Index 0.57

• Eastern Pacific TCs
• TC in Swath?
• Yes No
• Method Yes 121 74
• Detects a
• TC in No 7 797
• Swath?
•  
• Probability of Detection 0.95
• False Alarm Rate 0.38
• Critical Success Index 0.60

29
Why Did We Fail to Detect Them Early?

• Too close to land during development
• In a QuikSCAT coverage gap during development
• The storm developed rapidly
• Outside of our domain (north of 25oN)
• Multiple errors in ambiguity selection because of
  heavy rains

30
Summary

• SeaWinds algorithms are not currently designed to
  consider rain.
• Improved rain flags should indicate magnitude of
  rain-related problems.
• Will improve ambiguity selection, and the
  quantity of data for severe weather
• SeaWinds-based surface pressures can help
• Alert forecasters when NWP forecasts that are
  seriously in error
• Estimate central pressure for TS and Tropical
  Depressions
• Locate the centers of circulation, when aircraft
  reconnaissance is not available
• SeaWinds on QuikSCAT can be useful for early
  identification of potential tropical cyclones
• Particularly effective in tropical Western
  Pacific Ocean

31
(No Transcript)
32
Related Web Sites
Jim OBrien and Mark Bourassa Center for
Ocean-Atmospheric Prediction Studies http//airse
a-www.jpl.nasa.gov/quikscat/ http//coaps.fsu.edu/
scatterometry/ http//manati.wwb.noaa.gov/quikscat
/ http//www.ssmi.com/qscatinfo.html/ http//www.i
fremer.fr/cersat/ACTIVITE/ERS/MISSION/E_ERS.HTMsa
t/ http//www.ee.byu.edu/ee/mers/Seawinds-1.html/
http//airsea-www.jpl.nasa.gov/seaflux/
33
Example Winds Within a Swath (Hurricane Cindy
cat. 1)
34
Example Winds Within a SwathEarly Detection of
Tropical Circulations
35
Results for the 1999 Atlantic Hurricane Season
36
Early Detection Times for 1999
37
Results for the 2000 Atlantic Hurricane Season