92nd American Meteorological Society Annual Meeting (January 22-26, 2012)

Wednesday, 25 January 2012
Verification of Oklahoma Weather Lab Student Forecasts From 2009 to 2011
Hall E (New Orleans Convention Center )
David John Gagne II, Univ. of Oklahoma, Norman, OK; and T. A. Supinie, A. J. MacKenzie, A. D. Losey, and B. R. Smith

The Oklahoma Weather Lab (OWL) is a student-run weather forecasting organization at the University of Oklahoma. As OWL has grown, many potential clients and other meteorologists have inquired about the quality of OWL's forecasts. Until recently, OWL could provide only limited information about its forecast quality. In fall 2009, OWL began archiving all of its forecasts for 12 cities in Oklahoma and Key West, Florida. A Python framework was developed to extract forecasts from the archive and match them with corresponding ASOS observations. Temperature, precipitation, wind speed, wind direction, and sky cover forecasts were evaluated. Initial and modified forecasts for all lead times were evaluated by semester and across forecasting shifts. An interactive, automated web interface has been developed to allow the general public and forecasters to view OWL's verification scores. Preliminary results show: (1) a 3 to 6 degree Fahrenheit temperature error on average, (2) positive skill compared to climatology for probability of precipitation forecasts for all lead times, and (3) an approximate 5 mph error in wind speed forecasts. Future plans include comparing model output statistics (MOS) and the National Digital Forecast Database (NDFD) to OWL forecasts. OWL forecasts will be verified with Oklahoma Mesonet observations in addition to ASOS.

Supplementary URL: