Current Icing Potential For Alaska (cip-ak): Quality Assessment Report

Abstract

Current Icing Potential for Alaska (CIP-AK). CIP-AK was developed by the Inflight Icing Product Development Team of the Federal Aviation Administration Aviation Weather Research Program (FAA/AWRP), and is currently being considered for transition to an experimental product through the Aviation Weather Technology Transfer (AWTT) process. This report concentrates on verification results for the CIP-AK computed for the periods 23. This report summarizes the assessment of icing conditions produced by the July–11 September 2002 and 1 January–30 May 2003. The evaluation considers performance of the algorithm over three domains in Alaska where observation concentrations are highest. Voice pilot reports (PIREPs) were used to evaluate the CIP-AK performance and were augmented with additional PIREPs supplied during 23 July–11 September 2002 by PenAir Airlines. Because of the sparse nature of PIREP observations over Alaska, overall results are presented in the report, while further stratifications of the results are excluded from the analysis. The icing diagnoses were verified every 3-h using Yes and No icing observations from PIREPs indicating either “light or greater” icing severity or “no icing.” CIP-AK diagnoses were evaluated as Yes/No icing diagnoses by applying a threshold to convert the algorithm output to a Yes or No value. A variety of thresholds was applied to the algorithm output in order to examine the full range of CIP-AK performance characteristics. The verification analyses were primarily based on the algorithm’s ability to discriminate between Yes and No observations, as well as the extent of their coverage. To provide a standard of comparison, complimentary results for the Airmens’ Meteorological Advisories (AIRMETs), the operational forecasts issued by the National Weather Service Alaskan Aviation Weather Unit (AAWU), are presented in Appendix A. Several thousand individual CIP-AK diagnoses were considered in this evaluation. The number of icing observations considered (Yes and No) in the evaluation ranged from 3,070 for the largest of the three domains, CIP-AK Land domain, to 528 for the Anchorage-area subdomain. Results of the evaluation indicate that CIP-AK is skillful at discriminating between Yes and No icing conditions. CIP-AK provides relatively efficient diagnoses, covering comparatively small volumes for a given icing detection rate. Using a threshold of 0.10, CIP-AK correctly classifies 68% of the Yes PIREPs and 66% of the No PIREPs, while covering approximately 13% of the airspace volume over the CIP-AK Land domain. The forecast quality is maintained up to about 15,000 ft. The quality of CIP-AK diagnoses is relatively insensitive to variations in the PIREPs used for the analyses, and performance was best between 3,000 and 15,000 ft where the number of observations was greatest. Detection rates and volumes covered varied, with the most favorable numbers for the larger of the three domains (CIP-AK Land). Trends in CIP-AK performance over the two evaluation periods suggest that CIP-AK performs similarly in summer 2002 as in winter/spring 2003. However, the summer 2002 dataset had supplemental PIREPs from PenAir Airlines.