Ken Cook – SOO NWS Wichita, KS (ICT)
An Assessment of Using the Mean Field Bias Correction to Improve
Precipitation EstimatesKen Cook and Maggie Schoonover
NOAA/National Weather Service Office2142 South Tyler Road
Wichita, KS 67209Phone: (316) 942-8483 Fax: (316) 945-9553
Email: [email protected]
Ken Cook – SOO NWS Wichita, KS (ICT)
Outline• Introduction• Prompted Use• Methodology of Analysis• Assessment
– Examination Results – Case Studies– Challenges– Advantages
Ken Cook – SOO NWS Wichita, KS (ICT)
Ken Cook – SOO NWS Wichita, KS (ICT)
What is the Mean Field Bias?• A statistical analysis between the gauge observations
and the radar bin that matches that gauge (NPair)• Performed hourly• Uses a minimum of 10 NPairs
– If 10 cannot be ascertained during the current hour, then looks back in time until 10 is reached
– User adaptable parameter• Applies this statistical analysis (one number) to the entire
coverage area• Software part of Multi-sensor Precipitation Estimator
(MPE)
Ken Cook – SOO NWS Wichita, KS (ICT)
Why Use It?• June 8th, 2005 case• 2-3X observed
precipitation• Latest in a number of
cases where precipitation estimates were less than desirable
• Saw media partners using this bad data
• Needed to improve forecasters confidence in radar estimated precipitation
• Improve service/warning meteorology
Ken Cook – SOO NWS Wichita, KS (ICT)
Once Implemented• Results were instantly improved• Noticed some underestimation during
various events• How much have we improved?• How can we make the system better?
– Local training/learning– National science sharing/improve
development
Ken Cook – SOO NWS Wichita, KS (ICT)
Methodology• Radar Data (ICT)
– 12Z STP level III data from NCDC– Nexrad Exporter (create shapefiles)– ArcGIS 9, create rasters
• Observation Data– Gathered 12Z Rain Gauge (Tipping Buckets/COOP)
reports– Imported into Microsoft Access– Loaded then added XY Data in ArcGIS 9
• Compared Datasets for the period March through June 2006
• Resulted in ~ 400 G-R Observation Cases• Gauge data assigned a “bin” value
– .45 gauge value assigned to the .3 to .6 radar “bin”
Ken Cook – SOO NWS Wichita, KS (ICT)
Results
• Slight underestimation evident
Ken Cook – SOO NWS Wichita, KS (ICT)
Results
Ken Cook – SOO NWS Wichita, KS (ICT)
Results
Ken Cook – SOO NWS Wichita, KS (ICT)
Results• There seems to be a clear signal of slight
underestimation• More noticeable as
– Event grows in size– Amount of precipitation increases
Ken Cook – SOO NWS Wichita, KS (ICT)
Case Studies• 32 Gauge
Observations• Light Rain
Event• Average Gauge
Observation .14 inches
• Highest Gauge Observation .22 inches
• Estimates were outstanding
• Hourly MFB Calculations Very Consistent
Ken Cook – SOO NWS Wichita, KS (ICT)
Case Studies• 43 Gauge
Observations• Moderate Rain Event• Average Gauge
Observation 1.10 inches
• Highest Gauge Observation 2.85 inches
• Poorest estimation of the cases
• Largest number of observations for one case (largest coverage)
• Highest average precipitation
• Hourly MFB Calculations somewhat less stable
Ken Cook – SOO NWS Wichita, KS (ICT)
Challenges - How to Improve the MFB
• Use proper overlays in MPE to inspect suspect gauges
• Inspect gauge table in MPE
• Take out bad gauges from MPE ingest filter
• Inspect MPE Local Bias for areas that may be over/underestimating as compared to Mean Field Bias
Ken Cook – SOO NWS Wichita, KS (ICT)
Advantages - Using the MFB• Better Precipitation
Estimations• No change to Z/R
relationship necessary
• Updated hourly• Reacts to a warm
rain process with no interaction
• Improved credibility & customer service
Ken Cook – SOO NWS Wichita, KS (ICT)
Future• Continued assessment
– Alter adaptable parameters?– Alter npairs?
• Science sharing with developers other users– Incorporate differences from second/third runs of
bias?– Assume 1:1 bias after no precipitation?
• Goal: Improve performance
• Thank you – Questions??
Ken Cook – SOO NWS Wichita, KS (ICT)
Resources• Cook, Kenneth (SOO – ICT), 2006: WFO Wichita Science and Training Intranet Page (
http://204.194.227.45/soo/soopage.htm)– Training Materials Also Available
• Hunter, S. M., 1996: WSR-88D rainfall estimation: Capabilities, limitations, and potential improvements. National Weather Digest 20 (4), 26-38
• WHFS Field Support Group Web Site (http://www.nws.noaa.gov/om/whfs)