Thursday, 24 January 2008: 3:30 PM
An evaluation of the performance of the Forecast Icing Product (FIP)
226-227 (Ernest N. Morial Convention Center)
Michael B. Chapman, NOAA/ESRL/GSD/CIRES/Univ. of Colorado, Boulder, CO; and J. Mahoney, S. Madine, and J. K. Henderson
The Forecast Icing Product (FIP) is a physically-based situational fuzzy-logic algorithm that generates three icing fields: severity, probability, and super-cooled large drops (SLD). This algorithm integrates various model variables (e.g. relative humidity, temperature, precipitation, etc.) using 20-km output from the 13-km Rapid Update Cycle (RUC) forecast model. In 2006 and 2007, FIP was evaluated and approved as an experimental product by the Federal Aviation Administration's Aviation Weather Technology Transfer (FAA / AWTT) board. The next step in the AWTT board process is for FIP to be considered for unrestricted supplemental access to various aviation decision makers (e.g. pilots, dispatchers, controllers, etc.).
In this study, the three FIP fields are evaluated by using both standard and modified user-based verification techniques. The severity field is evaluated on its ability to discriminate between different observed icing intensities. The probability field is evaluated by the accuracy of the calibration as well as by the skill of the algorithm at discriminating between YES and NO icing observations. The performance of the SLD field is evaluated by using: (1) icing pilot reports (PIREPs) that indicate both “severe” icing intensity and “clear” icing type; (2) PIREPs with any indication of freezing precipitation, and (3) surface stations (METARs) that report freezing precipitation. Also evaluated in this study is the severity field at different probability thresholds and the severity field constrained to the AIRMET boundaries. The time periods chosen during 2006 are Jan-Mar, May, July-August, and October, which are representative of all four seasons of the year.
Supplementary URL: