Andrew Burton

1
Use of Scatterometer Winds in TC
Forecasting Tropical Cyclone Warning Centre
Perth
Andrew Burton Bureau of Meteorology, Perth,
Australia
2
Application of Scatterometer to Tropical Cyclone
Forecasting

• Formation
• Size (radius of gales)
• Wind distribution
• Not for absolute intensity
• (winds saturate at gt60 90? knots)

3
Where to Get Scatterometer Data

• NRL Monterey
• http//www.nrlmry.navy.mil/tc_pages/tc_home.html
  
• NOAA/NESDIS QuikSCAT
• http//manati.wwb.noaa.gov/quikscat
• Storms page includes ambiguities
• http//manati.wwb.noaa.gov/cgi-bin/qscat_storm.p
  l
• Alternative NOAA site, with SSMI wind
  speeds
• http//polar.wwb.noaa.gov/winds/globdata.html
• FNMOC
• http//152.80.49.210/PUBLIC/SCAT
• or http//www.fnmoc.navy.mil/PUBLIC/SCAT
• Remote Sensing Systems
• http//www.ssmi.com

4
NRL Monterey http//www.nrlmry.navy.mil/tc_page
s/tc_home.html
5
NOAA/NESDIS QuikSCAT http//manati.wwb.noaa.gov/q
uikscat
6
FNMOChttp//152.80.49.210/PUBLIC/SCAT/
FNMOC Ambiguity Removal over SSMI
Near Real-Time Ambiguity Removal
7
Remote Sensing Systemshttp//www.ssmi.com
8
Differences

• Wind retrieval
• RSS uses KU-2000 wind retrieval method
• Others use QuikSCAT1 wind retrieval method
• Rain Flags
• Generally Multidimensional Histogram (MUDH)
  procedure a statistical method based on
  noisiness of data
• RSS has similar approach though it is less
  conservative and hence rain affected areas are
  often smaller

9
Scatterometer Coverage
10
QuikSCAT SeaWinds Measurements
V-pol H-pol
From Dr. M. Freilich, Oregon State University
11
SeaWinds Swath Geometry
Subtrack View (4 solns,but small angle
var)
Forward Look
Red V-pol Blue H-pol
Edge View (2 solns) V-pol only
Ideal View (4 solns, 90 deg var)
Backward Look
From Dr. M. Freilich, Oregon State University
12
Scatterometry 2-Look Solutions
1 2 3
4
Solution wind 10m/s at ?? deg
From Dr. M. Freilich, Oregon State University
From Dr. M. Freilich, Oregon State University
13
Scatterometry 4-Look Solution(s)
1 2
3 4
Most-likely solution 10m/s at 40deg
From Dr. M. Freilich, Oregon State University
From Dr. M. Freilich, Oregon State University
14
Ambiguities The Chicken-Scratch Diagram
15
Location, Intensity, Wind Distribution
Source Tropical storms discussion group
16
(No Transcript)
17
Small system (X) analysed for 3 days --no help
from NWP model
20S
170W
170W
O
O
X
X
X
O
25 Jan 1800Z 26 Jan 0517Z
26 Jan 1800Z
170W
160W
170W
160W
O
X
O
27 Jan 1712Z
28 Jan 0427Z
18
Interpretation Challenges

• Edge of swath ( 7 wind vector cells)
• Rain effects
• Sensitivity to errors in NWP
• Practical wind regime 3-45? m/s (problems with
  both very light and very strong winds)
• Resolution (25km) impact in tight gradients
• Ambiguity Removal Process and rain flag process
  can affect final solution

19
Edge Problems
Along the whole edge or small portion
FNMOC DISPLAY
20
Rain Flags MUDH vs RSS
TC Ando
RSS
Less Coverage
FNMOC-NOGAPS
FNMOC-NRT
Same Coverage
21
Rain Effects tear drop

• Position using the curvature outside rain block
  region.
• Look for good north-south winds.

22
Streamlines
Beware of winds perpendicular to the swath, even
when they are not flagged
X
Look for non-rain flagged winds
TC Chris 03/02/02 0914Z
23
Isotachs
X
Look for min speed near centre
TC Chris 04/02/02 1002Z
24
Errors in NWP
TC Guillaume 19/02/02 1341Z
Wrong Model Position?
25
Where is TC HUDAH?No circulation!
Try to fix in trough equator-ward of the
strongest winds
?
Max Winds 95 knots
26
Model initialization errors
In this case, poor model initialization combined
with a lower skill nadir position, picks proper
wind speed, but NO circulation center
20/2356Z
AVN 19/12Z tau 24
(Light winds?) -----low skill
c
10S
c
10S
?
Max Wind 55 KTS
?
20S
TC Paul
20S
27
Comparing Different Solutions
FNMOC-NOGAPS
FNMOC-NRT
28
NRCS imageryNormalised Radar Cross Section
29
Microwave Imagery vs Quikscat
Scatterometer winds give wrong estimate for
centre
10 S
14 S
18 S
92 E
88 E
84 E
92 E
88 E
84 E
Comparison between Quikscat solution from NESDIS
30/11/2001 at 0023Z and fair LLCC seen in SSMI
near 14.4S 89.1E 30/11/2001 at 0218Z
30
Analysis Methods - Summary

• Ignore the bad - streamline the good
• Tear-drop curved end
• TCs equatorward side of max wind
• Compare different solutions
• Isotach method ignore direction

31
Conclusions

• Provides coverage over data sparse areas
• Wind speeds generally good useful for areas of
  gales etc
• Use the data if it makes sense
• Be aware of low skill areas and different
  ambiguity removal processes (compare!)
• Do not use in isolation