Combining multiple satellite datasets for the detection and verification of aircraft icing

In the past, the primary observation set used for the verification of aviation icing products has been PIREPs. The drawback of PIREPs is that they are subjective-- they are reported by pilots at their discretion based on experience, and are consequently subject to inconsistencies in location and intensity. In addition, PIREPs are an unrepresentative sample of the atmosphere given that pilots attempt to avoid icing. Therefore, it is important to identify alternative datasets that can provide more objective information with which to verify icing products. One such data source is satellite data. The use of geostationary and polar orbiting satellite data for detection of cloudy regions, as well as icing conditions, is not new. Mecikalski, et al. (2007) provides a summary of many approaches used to derive icing information from GOES and MODIS satellite imagery. These products provide a cloud base/top and ice base/top from which cloud and icing layers are derived. Additionally, CloudSat and CALIPSO provide a very detailed view of clouds from space (Stephens et al. 2002), which, when using the cloud classification products together can provide upper and lower bounds for regions of icing.

This study uses GOES, MODIS, and CloudSat/CALIPSO, as well as RAP temperature data, in collocated regions to assess gridded icing forecast and analysis products. We identify areas of icing in each satellite product using 'relaxed' and 'strict' requirements, as well as areas where icing should not exist (clear air and outside an appropriate icing temperature range). Collocated cross sections of each satellite product are examined. The areal extent of the icing and clear air fields are compared to explore the differences in satellite products and the effects on verification. Preliminary results show discrepancies in icing layer depth and icing base altitude that lead to inconsistencies in verification. An additional assessment of cloud phase derived from satellite will be compared with identified icing regions.