An Extended CAD Spectrum

1
An Extended CAD Spectrum

• CSTAR co-lab presentation,
• 20 July 2001

2
Objectives of Todays Session

• Review CAD detection algorithm and climatology
• Review workings of algorithm
• briefly summarize some results

• Outline proposed extension of CAD classification
• Explain how CAD classification algorithm works
• Provide evidence that CAD sub-types matter

• Show some preliminary composite results

• Solicit YOUR feedback

3
CSTAR CAD Objectives

• Develop an objective CAD detection algorithm

• Identify a large number of historical CAD events
  with which to build a flexible database

• Test original subtype hypothesis develop an
  objective CAD climatology

• Stratify sample and generate composites
• elucidate subtype differences
• e.g., differentiate slowly and quickly eroding
  cases
• solidify conceptual picture of CAD subtype
  evolutions
• build flexible database for future research

4
The Original CAD Spectrum
Classical
CAD Intensity
Hybrid
In-Situ
ltlt1
?1
gtgt1
Diabatic Processes
Dry Synoptic Processes
5
ClassicalDry synoptic forcingMay have
diabatic contributionStrong, northerly parent
high
In-Situ Diabatic processes essential High
location unfavorable Little/no CAA

• Hybrid
• Equal contributions from
• dry synoptic diabatic Parent high
• Right place, too weak
• Progressive

6
CAD Detection Algorithm

• Based solely on surface observations (hourly)

Pressure, ? Laplacians computed for lines 1-3
Check for threshold value (1 standard deviation
above average /- values)
Line 4 checked for along-barrier pressure drop
(surrogate for NE ageostrophic flow)
Criteria must be met for 6h
7
CAD Detection Algorithm Findings

• 374 events identified (1984-1995)

8
CAD Detection Algorithm Findings

• Original damming subtypes emerged, but...

• Many cases that did not fit the original subtypes
  were also identified
• Parent high too far south or too weak for
  classical label, with no precipitation, but
  damming still evident

• Weak, short-lived warm-season cases not
  accompanied by precipitation at onset, but
  maintained in part by solar sheltering

• Events in which a cyclone to the south provided
  the along-barrier pressure drop

• Extended CAD spectrum (perhaps broadened
  definition of CAD to include cases that were
  previously dubbed lookalikes)

9
The Extended CAD Spectrum
Classical (Diabatically Enhanced)
Classical (Dry Onset)
CAD Intensity

• Weak/Southern Onset
• dry onset
• high lt 1030 mb or S of 40N
• or progressive

Hybrid
In-Situ
Unclassified
ltlt1
?1
gtgt1
Diabatic Processes
Dry Synoptic Processes
10
Classification based on characteristics of
parent high

• Classical (dry onset and diabatically. Enhanced)
• High gt 1030 mb
• gt24 h duration

In-Situ

• Weak Onset
• High lt 1030
• or lt 24h duration
• dry onset

• Hybrid
• High lt 1030
• or lt 24h duration
• precip. W/in 6h of onset

• Southern Onset
• high S of 40
• dry onset

In-Situ
11
Automated CAD Classification
12
Do the Differences Between the Subtypes Matter??
20
70
23
43
17
15
7
Cold Season Events
13
Cold Season Events
14
Average Event-Total Precipitation by Sub Type
15
Average CAD duration by Sub Type
16
Preliminary Composite Results Cold Season

• Classical (Diabatically Enhanced)
• E-W elongation of high
• confluence, little ridging _at_ 500
• cold front nearby

• Classical (Dry Onset)
• N-S elongation of high
• confluence AND ridging _at_ 500
• cold front far offshore

20 cases
70 cases
Onset time
Onset time
Sea level pressure
Legend
500 mb height
17
Is the difference in composite mean SLP
statistically significant?
Classical (dry onset) minus Classical (diab.
enhanced)
95 significance
99 significance
Onset time
SLP higher in dry
SLP lower in dry
Legend
18
Conclusions, the Next Step

• CAD detection, classification algorithms
  functional
• Results
• algorithm identified more warm-season cases than
  anticipated
• additional subtypes required to complete spectrum
• sensible weather signatures are statistically
  distinctive
• composites indicate fundamental differences in
  synoptic patterns differences statistically
  significant

• Check out http//www4.ncsu.edu/cmbailey
• How can we make these results useful?

19
Thanks to

• All of you for your comments, attention, and
  support!
• Gail Hartfield for the sensible weather subtype
  data and graphics, and guidance throughout
• Kermit Keeter for continuous support and guidance
• Scott Sharp for help testing the detection
  algorithm (archiving, analyzing and classifying
  real time cases)
• Reid Hawkins, Larry Lee, Neil Stuart, Rick
  Neuherz, Mike Rusnack, Ed Delgado, Dan Riley, Pat
  Moore and others who have provided useful
  comments
• Jonathan Blaes for frequent support with web
  presentations, case studies, comments and more
• Ron Humble for help setting up the CAD database
  on the Linux PC