An Examination of the Climatology and Environmental Characteristics of Flash Flooding in the Binghamton, New York County Warning Area

1
An Examination of the Climatology and
Environmental Characteristics of Flash Flooding
in the Binghamton, New York County Warning Area

• Stephen Jessup
• M.S. Student
• Dept. of Atmospheric Science
• Cornell Univ.

2
Project Objectives

• Develop a long-term climatology of flash flood
  events for the BGM CWA.
• Identify any spatial differences in flash flood
  frequency and flood producing meteorological
  conditions across the CWA.
• Analyze a set of meteorological variables to
  quantitatively identify combinations of variables
  that are associated with flash flooding.
• Compare the conditions associated with flash
  floods to the conditions associated with
  non-events

3
Flash Flood Climatology

• Spatially FF's most common in NY/PA border
  counties and in eastern NY counties
• Diurnally
• Peak in late afternoon/early evening
• Secondary max. in morning
• Seasonally
• Peak in summer (June max.)
• Min. in late fall/winter

4
16
7
5
8
6
6
16
8
7
9
18
6
16
13
25
24
13
7
5
28
8
4
2
11
5
Flash floods per county area
13.2
9.0
18.5
7.6
17.8
11.5
16.0
16.0
21.3
10.1
12.6
35.4
12.9
25.0
58.8
11.1
6.1
24.3
13.4
9.6
5.0
17.4
7.3
12.3
6
Mostly Spring/Summer
Mostly Fall/Winter/Spring

7
(No Transcript)
8
Antecedent Precipitation

• Determined for one week (7 days) and one month
  (30 days) prior to flash floods
• Climatology for comparison
• Consists of all non-flood years (from 1986-2003)
  for each flash flood date
• To test hypothesis that floods tend to occur
  during periods of above-normal precipitation
• Flash floods tend to occur in anomalously wet
  periods

9
(No Transcript)
10
Climatology vs. Flash Flood Antecedent Precip.
(7-day)
11
Climatology vs. Flash Flood Antecedent Precip.
(30-day)
12
(No Transcript)
13
(No Transcript)
14
Improving FF Forecasting Procedures for NWS BGM
Methodology

• Construct independent databases of flash flood
  and null event cases
• Determine meteorologically significant parameters
  and their values during events
• Find combinations of variables that improve
  predictability
• Plot composites to determine whether the synoptic
  situations of FF's and null events differ
• Merge these into a forecasting methodology

15
Datasets

• Warm-season flash flood cases (May-Oct), n51
• Separated by at least one week (7 days)
• Drawn from 1986-2003
• Warm-season heavy precipitation events, n36
• At least 1 in one hour, at least 1.5 in six
  hours
• Separated from each other FF's by at least one
  week
• Drawn from 1986-2003
• Random days representing same seasonality as
  FF's, n51
• Random year (1986-2003) assigned to the date of
  each FF case
• Separated from each other FF's by at least one
  week
• Watches/warnings that did not verify, n17
• Separated from each other by at least one week
• Drawn from 1995-2003

16
Dataset Methodology

• NCEP Regional Reanalysis used as primary data
  source 32 km, 3-hour resolution
• Three time periods used time closest to the
  flood, two time steps of three hours prior
• Most variables averaged over quadrilateral area
  containing FF counties
• Area for prior time periods determined by
  backtracking four corners of this area using 850
  mb wind
• 850 wind, storm motion vectors backtracked an
  extra timestep to reflect inflow
• Backtracking not used for several variables
  classified as synoptic parameter representing
  large-scale field used instead

17
Highlights Current BGM FF Checklist

• Winds/storm motion
• Slow storm movement (MBE/Corfidi vector)
• Low level jet gt 20 kts
• 700 500 mb winds lt 30 kts
• Weak mid-level (700-500 mb) shear
• Upper level divergence
• Atmospheric Moisture
• Mean 1000-500 relative humidity gt 70
• Precipitable water gt150 normal or gt 1.4 inches

18
BGM FF Checklist, continued

• Synoptic-scale features
• Nearby surface boundary
• Low-level theta-e axis
• Upper level ridge axis
• 1000-500 mb thickness diffluence
• Other parameters
• Tall and skinny CAPE
• Warm cloud depth exceeding 3-4 km

19
Summary Results

• Thresholds in checklist generally agree with FF
  results, but are often exceeded in non-events
• Exception Low-level jet apparently not as
  important for flash flooding, but more common for
  heavy rainfall non-events
• Measures of antecedent soil moisture a good
  first-guess criterion between both flood/heavy
  and flood/watch
• Properties of the 850-mb theta-e field differ in
  both flood/heavy and flood/watch comparisons
• Measures of 850-mb and 700-mb moisture (dewpoints
  and RH) differ for flood/heavy and flood/watch
• Notable differences in large-scale mid and low
  level wind patterns
• Notable differences in 850-500 relative humidity
  patterns

20
Heavy
Flood
Watch
Sea Level Pressure
21
850-mb wind speed
Expect strong LLJ
22
850-mb wind direction
Mostly SW to W
Some SE
23
850-mb wind speed vs. 850-mb wind direction
F flood H heavy W watch R random
NW often non-events
Fast winds either
SE mostly floods
24
Weak LLJ, convergence
Stronger LLJ
Strong LLJ, convergence
850 mb wind vector
25
Storm Motion Speed
Not necessarily slow storm motion
26
Storm motion direction
Primarily SSW to W
27
Mid-level (500 mb 700 mb) Shear Speed
Weak shear
28
Mid-level (500mb-700mb) shear direction
Weak directional shear
29
Mid level shear direction vs. speed
Larger shear can flood!
30
Weak, convergence
Strong
Strong, convergence
700 mb wind vector
31
Weaker
Stronger
Stronger
500 mb wind vector
32
250 mb wind vector
33
Precipitable water ( normal)
34
Weekly Ant. ppt. vs. Precipitable Water (
normal)
35
Localized moisture
Frontal signature?
Perhaps some of both?
Precipitable Water (anomaly)
36
850 mb Theta-e vs. weekly antecedent precipitation
Floods lower theta-e and wetter antecedent
37
CAPE
Long, skinny CAPE?
38
850 Wind Direction vs. 850 Dewpoint
Floods lower 850 Td
39
Floods have higher RH than heavy
850 mb RH
40
700 mb RH
41
500 mb RH
42
Summary

• LLJ not as significant in FF's as in null events
• Composites suggest theta-e ridging is less
  significant in FF's than in null events
• Atmospheric moisture greater and more localized
  in FF's than in null events
• Possible connection between antecedent soil
  moisture and local maximum in moisture content
  during FF's?
• Caveat small sample size, small spatial domain!

43
Acknowledgements

• COMET Outreach Project S05-52254
• Art DeGaetano, Cornell University
• Mike Evans, NWS Binghamton