Objective method to determine the Typhoon intensity using IR1 images

1
Objective method to determine the Typhoon
intensityusing IR1 images

• 20/03/2007 NESDIS
• Japan Meteorological Agency
• Koji Kato

2
Todays Topic

• Preliminary objective approach to determine the
  typhoon intensity.
• Input the position of typhoons and IR1 image of
  typhoon then get the typhoon intensity
  automatically!

960hPa MAX103kt
Analysis Program
3
Example
Comparison between and manual result 2004
TC0024(Japan)
4
Procedure

• Brief introduction of conventional method
• Introduction of objective hurricane analysis
  method.

5
What is Dvorak Technique?

• Tropical storm analysis method using IR (11µm)
  image of meteorological satellites.
• (Especially minimum pressure and maximum wind
  velocity)
• At the JMA, we have no operational objective
  method for Typhoon analysis.

6
?????(BD???)
?? CSC???
??????????????????? CSC???????????(TI)??????Step
1A(??)???
???????????? 1)????????????????????????????????
2)E??4.5??????????????????????????0.5???
1
?????????? (?????Step3??????)
??????2A2E????????? Step36???????????????Step2??
?
2
?????W?????DT??0.5????????????1.0???????2C(?????)?
?????
??? ? ?
?????????????????
BAND???? (10log?????????? ???????)
2A
SHEAR???? (CSC?????????? ??????)
2B
DT1.50.5 DT2.5
DT3.0
DT3.5
Step4?
???(Eadj)
YES
EYE???? (????????????? ?????????)
24????T? ?2.0
2C
CF??? CFEEadj
E6.5 E6.0 E5.5 E5.0
E4.5 E4.5 E4.0
NO
Step2A or 4?
BF??(BF)
12????T? ?3.5
EMBED???? (CSC??????0.4??)
YES
2D
DT??? DTCFBF
CF5.0 CF5.0 CF4.5 CF4.0 CF4.0
CF3.5
NO
Step4?
Step2A or 4?
7
How to do? Part1
IR
Special Coloring
Determine the pattern
TY 05 2005/07/15?12UTC
TY 05 2005/07/15 12 UTC
8
How to do? Part2
Tropical Number 3.5
9
How to do? Part3
10
Weak point of Dvorak Technique

• 1. Depend on forecasters skill.
• 2. Improvement of satellite sensor doesnt
• contribute the accuracy.
• 3. Get intensities in 6 hours interval.

11
Pattern determination
EYE Patter or Band Pattern ?
12
Weak point of Dvorak Technique

• 1. Depend on forecasters skill.
• 2. Improvement of satellite sensor doesnt
• contribute the accuracy.
• 3. Get intensities in 6 hours interval.

Why?
13
Can we see the difference?
14
How to overcome the problems?
15
Objective Hurricane Analysis method

• ODT (Objective Dvorak Technique)
• (Velden, Zehr 1998)
• AODT (Advanced ODT)
• (Olander, Velden 2002-present)

16
What is AODT?

• National Hurricane Center and Wisconsin
  university group developed AODT. Only one
  objective method to determine the hurricane
  intensities.
• Input the center position of the hurricane and
  the IR(10 µm) image, then output the hurricane
  intensity.
• Algorithm is not open!
• Output the intensity every hour!

17
(No Transcript)
18
I want an objective method!I am envious of
them.But I cant get the AODT...!!Make by
myself!
19
My method
1.Use IR picture and center position of typhoon.
2. Divide typhoon area into concentric circles.
3. Classify the typhoon using maximum
temperature, minimum ring temperature and
variance.
Tropical Number
20
Procudure1
Divide into rings (0.05width) to 1.5radius. And
find the minimum temperature ring and calculate
the variance of temperatures.
TCenter
TRing
21
Procedure2

• From the temperature distribution classify the
  typhoon automatically

22
Procedure3
23
Procedure4
Ultra EYE
Active Cb
Low Cloud
Weak EYE
Tropical Number
Flat EYE
Determine the Tropical number using tendency and
previous Tropical number.
unknown
24
Example
Ultra EYE
Low Cloud
Active Cb
25
Ultra EYE
Warm area in the center area
Smooth ring (low variance)
Very low temperature
Ultra EYE
26
Active Cb
Very low temperature around the center
Temperature distribution is rough
spatially-temporally
Totally low temperature
Active Cb
27
Low cloud
High temperature around the center
Asymmetric shape
High temperature area
Low Cloud
28
Condition 1

• Ultra EYE
• TBBMAX-TBBMINgt30K
• sTBBMINlt8K
• TBBMAXgt240K
• TBBMINlt220K

29
Condition 2

• Active Cb -a
• TBBMINlt230K
• TBBMAXlt8K
• sTBBMINgt20K
• Or Average6(TBBMIN)gt16K

30
Condition 2.1

• Active Cb -ß
• TBBMAX-TBBMINlt 7K
• sTBB1.5Ring?20K

31
Condition 2.3

• Active Cb -?
• TBBMINlt230K
• Ave3(sTBB0.25)-AvePre3(sTBB0.25)

32
Condition 3.1

• Low Cloud - a
• TBBMAX?260K
• TBBMIN?240K

33
Condition 3.2

• Low Cloud ß
• TBB0.25?230K
• TBBMAX?260K
• sTBB0.25?15K

34
Condition 4

• Weak EYE
• TBBMAX-TBBMINgt15K
• sTBBMINlt10K
• TBBMINlt230K

35
Condition 5

• Flat EYE
• TBBMAX-TBBMIN?5K
• Ave3(sTBBMIN)lt8K
• TBBMINlt210K

36
Data

• Period 2004 TC0004-TC0040
• Satellite IR1(GOES-9)
• Using regression analysis to determine the
  T-number

Data Resolution 0.05 10bit
37
Is this method is valid?

• Introduce 5 examples.
• TC0007 TC0009 TC0013 TC0021 TC0024

38
TC0007
39
TC0009
40
TC0013
Expect the typhoon to become more stronger!
41
TC0021
Human error
42
TC0024
43
Features

• In T?4.5 (under 979hPa),
• comparable to human
• In Tlt4.0 (over 987hPa),
• Auto classification doesnt work well.

44
Further Study

• Improve the recognition ability especially in the
  developing and decaying stage
• Reduce biases between satellites

45
TC0517
Different Satellite! (MTSAT-1R) Bad Result!
46
Thank you for your attention!
Sorry, I stopped developing these tools. But Id
like to try again someday!
Thank you for your attention!
47
Fin