1. Northern Hemispheric Flood Fingerprints
Katie Walls
The Pennsylvania State University
December 16, 2008

2. Motivation
Flooding causes billions of dollars in damage every year across the Northern Hemisphere
The synoptic patterns involved share common patterns, or fingerprints
The configuration of the anomalies in relation to the flooding event
Root et al. (2007) found repeatable U.S. mid-Atlantic flood fingerprints

3. Hypothesis
There are common fingerprints for the mid-latitudes (20° to 60°) of the Northern Hemisphere
Four different atmospheric variables involved in identification
Sea Level Pressure SLP
Precipitable Water PWAT
850 hPa U-wind
700 hPa V-wind

4. Hypothesis Cont.
Six fingerprints identified using rotated Empirical Orthogonal Functions (EOFs)
Each rotated EOF represents one synoptic pattern and consists of four atmospheric variables

5. Case Selection Over 300 flooding events from three sources
National Climate Data Center's event database
CMORPH
Encyclopedia Britannica disaster articles from 1979 to 1999
Surface rainfall totals unavailable outside of the U.S. – need an objective method to assess rainfall
Japanese 25-year Reanalysis (JRA25) data
Validated via comparison of JRA25 data with North American COOP reports

6. Flood Event Locations
200 Events taken from North America, Europe, and Asia
where at least 2 in. (~50mm) of rain fell in an hr. period

7. Standardized Anomalies
Four atm. variables extracted from NCEP-NCAR Global Reanalysis dataset (17 pressure levels on 2.5º by 2.5º grids) for each event
Standardized anomalies SA calculated for each field on a 9 by 9 grid (22.5º by 22.5º) surrounding flood event L
Sea Level Pressure

8. Mitigate effects of seasonal variation
At each grid point:
A is the atmospheric variable value
µ is the mean of the 17-day average centered on the flood date
σ is the value of one standard deviation based on a 30-year climatology.
Next, calculate the eigenvectors of the SA fields

9. Rotated EOFs
Basically reduce a data set to a smaller data set that contains linear combinations of the original data
Reveal synoptic patterns by displaying the most variation amongst all cases - fingerprints
Six-factor oblique rotation reveals most meaningful meteorological synoptic patterns
Six distinct fingerprints represent 61% of the variance of floods across the Northern Hemisphere
Flooding events are described by a combination of multiple EOF rotations

14. Summary
Flood events are categorized using obliquely rotated EOF analysis, based on standardized anomalies
Oblique rotation reveals 6 fingerprints that explain 61% of the variance amongst 200 mid-latitude Northern Hemispheric floods
Identification of such patterns can advance flood forecasting and prevent life and property loss in the future

15. Future Work Add more atmospheric fields to methodology
Use standardized anomalies from ERA and JRA datasets
Search for flood fingerprints in predicted model anomaly fields
Apply methodology to other weather hazards

16. Acknowledgements Paul Knight Nels Shirer George Young Sue Ellen Haupt
Rich Grumm
Sonya Miller
The Pennsylvania State Climatology Office
COOP
University of Reading-World Weather News
Those who have supported me throughout my graduate experience