A research aircraft verification of the Integrated Icing Diagnostic Algorithm (IIDA)

The Integrated Icing Diagnostic Algorithm (IIDA) was designed to determine the potential for icing and supercooled large droplets (SLD) to exist at locations across the contiguous United States (CONUS) and southern Canada. It is currently employed by several regional airlines for dispatch information regarding icing. Past verification exercises using icing pilot reports (PIREPs) have shown that IIDA is a relatively efficient algorithm, when compared to other automated icing algorithms. While PIREPs allow for evaluation of an algorithm's ability to identify general icing conditions, they are difficult to use for verification of no-icing, SLD, and no-SLD forecasts/diagnoses. Negative icing reports are relatively uncommon, and the assumption of "no-icing" based on a lack of positive icing PIREPs is not reliable. Verification of SLD/no-SLD aloft is extremely difficult since PIREPs do not contain information regarding droplet size, except for the rare mention of FZDZ or FZRA in the weather or comments fields.

During the winter of 1997-98, the NASA-Glenn Twin Otter research aircraft completed more than 50 hours of flight in a variety of conditions over the Great Lakes region. Numerous encounters with FZDZ and FZRA are complemented by a large amount of data for conventional icing and no-icing situations. The aircraft was well equipped for sampling and identifying these conditions, and data collected by it provides reliable information for accurately determining the 3-D locations of icing, no-icing, SLD, and no-SLD. In this paper, Twin Otter data are directly compared with IIDA output to examine the algorithm's ability to identify SLD, no-SLD, and no-icing.

This paper provides the first absolute verification of diagnoses or forecasts SLD conditions aloft. The technique applied is fairly strict, as only IIDA output for the adjacent horizontal grid points that were within 1000 feet of the aircraft altitude. For each 3-D grid point, IIDA provides estimates of the potential for icing and SLD. This scaled, rather than binary, icing field allows for an assessment of the ability of the algorithm to discriminate between higher and lower likelihood of icing and SLD. Results are very encouraging. IIDA indicated high or moderate potentials for icing and SLD for most of the Twin Otter icing and SLD encounters, respectively. IIDA also demonstrated capability for differentiating between "yes" and "no" icing and SLD situations.