Tuesday, 14 January 2020: 2:15 PM
251 (Boston Convention and Exhibition Center)
The Weather Prediction Center’s (WPC) Excessive Rainfall Outlook (ERO) is a
probabilistic forecast of rainfall exceeding one, three, and six hour flash flood guidance
(FFG). Recently verification on the ERO has been developed and extended, but
forecasters lacked tools to serve as first-guess fields or to evaluate single-event
performance. Practically Perfect (PP) is commonly used at the Storm Prediction Center
for severe weather and uses a smoothed field of binomial observations to create a sort of
“Probabilistic Observation.” This talk will detail how PP can be applied to the plethora of
flooding observations and proxies to serve both as a verification and a first-guess field for
WPC’s ERO.
Due to the complicated nature of flash flooding, PP utilizes both flooding proxies
(e.g. Stage IV exceeding FFG and Stage IV exceeding the 5-year Average Recurrence
Intervals) and flooding observations (e.g. local storm reports and United States
Geological Survey river gauge data). Multiple sources are used because flash flooding is
often difficult to identify and single source observations exhibit biases in coverage and
classification. The PP values are created by applying a Radius of Influence (ROI) around
the observation/proxy and then smoothing the binomial field (100%=yes, 0%=no) with a
Gaussian filter. Since PP values are inherently ad-hoc, sensitivity studies are conducted
comparing the 2017 ERO to varying values of the ROI (between 5 km and 40 km) and
Gaussian filter (sigma values between 90 km and 120 km). The PP sensitivity run with
the smallest error and bias when compared to the 2017 ERO is selected as WPC’s
optimal configuration.
This talk will detail the selection of the optimal PP run and how PP can be used to
verify specific events at WPC. In addition, caveats of the PP method will be discussed
with some examples presented. Finally, the development of an ERO first-guess field by
applying PP to higher percentiles of WPC’s Probabilistic Quantitative Precipitation
Forecasts will be presented.
probabilistic forecast of rainfall exceeding one, three, and six hour flash flood guidance
(FFG). Recently verification on the ERO has been developed and extended, but
forecasters lacked tools to serve as first-guess fields or to evaluate single-event
performance. Practically Perfect (PP) is commonly used at the Storm Prediction Center
for severe weather and uses a smoothed field of binomial observations to create a sort of
“Probabilistic Observation.” This talk will detail how PP can be applied to the plethora of
flooding observations and proxies to serve both as a verification and a first-guess field for
WPC’s ERO.
Due to the complicated nature of flash flooding, PP utilizes both flooding proxies
(e.g. Stage IV exceeding FFG and Stage IV exceeding the 5-year Average Recurrence
Intervals) and flooding observations (e.g. local storm reports and United States
Geological Survey river gauge data). Multiple sources are used because flash flooding is
often difficult to identify and single source observations exhibit biases in coverage and
classification. The PP values are created by applying a Radius of Influence (ROI) around
the observation/proxy and then smoothing the binomial field (100%=yes, 0%=no) with a
Gaussian filter. Since PP values are inherently ad-hoc, sensitivity studies are conducted
comparing the 2017 ERO to varying values of the ROI (between 5 km and 40 km) and
Gaussian filter (sigma values between 90 km and 120 km). The PP sensitivity run with
the smallest error and bias when compared to the 2017 ERO is selected as WPC’s
optimal configuration.
This talk will detail the selection of the optimal PP run and how PP can be used to
verify specific events at WPC. In addition, caveats of the PP method will be discussed
with some examples presented. Finally, the development of an ERO first-guess field by
applying PP to higher percentiles of WPC’s Probabilistic Quantitative Precipitation
Forecasts will be presented.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner