220 likes | 335 Views
Polygon Warning Statistics For 2004 & 2005 Polygon Warning Team Meeting Fort Worth TX. Ken Waters NWS Regional Scientist National Weather Service Pacific Region Headquarters Honolulu, Hawaii October 25, 2005. Outline. The Process Quality Control Issues
E N D
Polygon Warning Statistics For 2004 & 2005Polygon Warning Team MeetingFort Worth TX Ken Waters NWS Regional Scientist National Weather Service Pacific Region Headquarters Honolulu, Hawaii October 25, 2005
Outline • The Process • Quality Control Issues • Overview of warnings for 2004 & 2005 • Results for WFOs participating in the test vs. nonparticipating WFOs
The Process • Collect the warnings from OCWWS’ FTP server • Scan all warnings, parsing relevant information (VTEC, polygon, etc.) and write to a single warning file • Quality control the data • Create shapefiles for TOR, SVR, FFW, and SMW from the data • Quality control the data more based on visual inspection of the GIS graphics • Repair “ring order” geometry (Warngen is not consistent in the ordering of the vertices; GIS software requires clockwise ordering)
The Process • Set the datum to North American NAD 83 and project into the Albers Equal Area projection for North America • Calculate the area (PWA) for each polygon warning and repopulate it back into the data attributes for the shapefile • Write out the shapefile as a “CSV” (comma separated variables) file • Intersect each shapefile with the counties shapefile to determine the CEA (county equivalent area) values • Import results into Excel to organize and compute the CAR (county area ratios)
Quality Control Issues • Data quality issues • Warnings with no polygons • Much less common in 2005 than 2004 • Have to completely remove the record since there is no value without a polygon
Quality Control Issues • Warnings with “one-point” polygons • Presumed bug in Warngen • Identified to OCWWS and FSL to correct • Again, have to remove the record as there’s no useful polygon
Quality Control Issues • Last digit of longitudes missing • Have to add a zero to longitude to prevent it from going back to the Prime Meridian • Comms error? • Happens several times each year
The Numbers(after the unusable warnings are removed) *SMWs were archived but not used for CAR statistics due to their marine extent
Methodology • Compute area of each polygon warning (PWA=Polygon Warning Area) • Sum up area of all counties identified by their UGCs in the warning (CAE=County Area Equivalent) • Compute CAR (County Area Ratio):
Data Issues Affecting CARs • Polygons stretching from one CWA to adjacent CWA • Reduces CAE due to not counting counties in adjacent CWA; increases CAR • Overextending pathcast • Increases PWA, thus increases CAR
The Final Results TEST NON-TEST
Conclusions • Average size of warnings increased slightly (why?) • Areal percentage of counties decreased overall • Areal percentage of counties decreased most notably for Polygon Warning test sites • Several significant quality control issues dealing with Warngen and policy that should be addressed