Louisville KY August 4 2009 Flash Flood Frank

  • Slides: 32
Download presentation
Louisville, KY August 4, 2009 Flash Flood Frank Pereira NOAA/NWS/NCEP/Hydrometeorological Prediction Center

Louisville, KY August 4, 2009 Flash Flood Frank Pereira NOAA/NWS/NCEP/Hydrometeorological Prediction Center

Motivation Recent high-profile flood events highlight the need for situational awareness of low probability,

Motivation Recent high-profile flood events highlight the need for situational awareness of low probability, yet high impact events. Louisville: Aug. 21, 2009 Nashville: May 1, 2010 Atlanta: Sept. 21, 2009

Outline • • • Event Summary Model and HPC Performance Any Indication of Impending

Outline • • • Event Summary Model and HPC Performance Any Indication of Impending Event? Spatial Density Plots Conclusions and Discussion

Louisville Impacts Rainfall amounts up to 6 inches fell between 1100 Z and 1400

Louisville Impacts Rainfall amounts up to 6 inches fell between 1100 Z and 1400 Z across central Louisville • Five inches fell in 90 minutes from 1145 Z to 1315 Z • In Louisville, nearly 200 people rescued from the tops of cars and houses. • No fatalities or injuries. • Courtesy of NWS WFO – Louisville, KY

SPC Analysis 04 Aug 12 Z 500 mb Height and Vorticity & 700 -400

SPC Analysis 04 Aug 12 Z 500 mb Height and Vorticity & 700 -400 mb Differential Vorticity Advection 850 mb Height, Temperature, Wind & Temperature Advection

SPC Analysis 04 Aug 00/12 Z Precipitable Water, Upwind Propagation Vectors & 1000 -500

SPC Analysis 04 Aug 00/12 Z Precipitable Water, Upwind Propagation Vectors & 1000 -500 mb Thickness

04 Aug 12 Z Nashville, TN Sounding PWAT = 1. 74 in.

04 Aug 12 Z Nashville, TN Sounding PWAT = 1. 74 in.

Radar and HPC Surface Analysis

Radar and HPC Surface Analysis

Verification 24 -hr Amounts Ending 04 Aug 12 Z observed Courtesy of NWS/OCWWS –

Verification 24 -hr Amounts Ending 04 Aug 12 Z observed Courtesy of NWS/OCWWS – National Precipitation Verification Unit HPC Ohio RFC NAM GFS

Verification 24 -hr Amounts Ending 05 Aug 12 Z observed HPC NAM Courtesy of

Verification 24 -hr Amounts Ending 05 Aug 12 Z observed HPC NAM Courtesy of NWS/OCWWS – National Precipitation Verification Unit Ohio RFC GFS

HPC Excessive Rainfall Graphics • Displays the probability that precipitation will exceed the flash

HPC Excessive Rainfall Graphics • Displays the probability that precipitation will exceed the flash flood guidance values issued by the River Forecast Centers (RFCs) Valid 04/12 Z – 05/12 Z Issued ~1000 Z

HPC Excessive Rainfall Graphics • Displays the probability that precipitation will exceed the flash

HPC Excessive Rainfall Graphics • Displays the probability that precipitation will exceed the flash flood guidance values issued by the River Forecast Centers (RFCs) Valid 04/15 Z – 05/12 Z Issued ~1400 Z

Moisture and Weak Warm Advection 12 -hour GFS Forecast Valid 04 Aug 12 Z

Moisture and Weak Warm Advection 12 -hour GFS Forecast Valid 04 Aug 12 Z Louisville 850 mb wind and PWATS Louisville 850 -700 mb Q-vector divergence & 850 mb warm air adv

SDF Forecast Sounding 09 -hour NAM Forecast Valid 04 Aug 09 Z Courtesy of

SDF Forecast Sounding 09 -hour NAM Forecast Valid 04 Aug 09 Z Courtesy of NWS WFO – Louisville, KY

High-Res Model Guidance • HPC investigating utility of high-res model guidance to anticipate heavy

High-Res Model Guidance • HPC investigating utility of high-res model guidance to anticipate heavy rainfall threats • Example from 4. 0 km WRF-NMM (run @ EMC) initialized with the 04/00 Z NAM

0700 UTC

0700 UTC

0800 UTC

0800 UTC

0900 UTC

0900 UTC

1000 UTC

1000 UTC

1100 UTC

1100 UTC

1200 UTC

1200 UTC

1300 UTC

1300 UTC

1400 UTC

1400 UTC

1500 UTC

1500 UTC

1600 UTC

1600 UTC

1700 UTC

1700 UTC

1800 UTC

1800 UTC

Integration of Hi-Resolution WRF guidance • High-resolution models are not accurate on the scale

Integration of Hi-Resolution WRF guidance • High-resolution models are not accurate on the scale of individual grid points • However, high-resolution models can capture realistic amplitude of events • Use neighborhood approach (e. g. , Schwartz et al. 2009) to give credit for the correct event/phenomenon, even if the placement is not perfect • Also known as “Spatial Density”

Neighborhood / Spatial Density Approach • Create binary field where threshold exceeded (Flash Flood

Neighborhood / Spatial Density Approach • Create binary field where threshold exceeded (Flash Flood Guidance) • Smooth the resulting binary (1 or 0) distribution (using a Gaussian Smoother) Model 1 h QPF 1” Binary Schwartz et al. (2009)

Creating the Exceeding FFG Density Plot Raw data (1 s & 0 s) are

Creating the Exceeding FFG Density Plot Raw data (1 s & 0 s) are run data through a Gaussian Weighted Filter to create an index of values from 0 -100 Model 1 h QPF > FFG Density Schwartz et al. (2009)

Density Plot • Used to raise forecaster’s situational awareness • Diagnostic Available from NCEP

Density Plot • Used to raise forecaster’s situational awareness • Diagnostic Available from NCEP High Res Window runs and experimental EMC run • Not a silver bullet – • Limitations of usingle models • Not calibrated (30% does not necessarily occur 30% of time) SPCWRF 4 (30 -HR QPF) QPF > 3 -hr FFG in 3 hrs

Conclusions Convection initiated ahead of warm front in a moist atmosphere along a weak

Conclusions Convection initiated ahead of warm front in a moist atmosphere along a weak low level jet • Event was poorly handled by lower resolution deterministic models and manual forecasts • High-resolution model input and spatial density plots may have been used to raise situational awareness •