Impacts of high thin cloud on the Current Icing Product (CIP)

Supercooled cloud liquid poses a great risk to general, commercial and military aircraft through its effects on aircraft icing. In order to facilitate prediction of regions with high aircraft icing probability, NCAR has developed the Current Icing Product (CIP). This algorithm, by utilizing a variety of input datasets, predicts where regions of supercooled liquid water may occur and at what level of probability.

Preliminary studies indicate that cloud top heights in CIP, which are based on passive infrared measurements from radiometers such as those onboard the Geostationary Operational Environmental Satellites (GOES), are biased toward higher altitudes with respect to measurements from CloudSat's Cloud Profiling Radar (CRP). Two possible causes for this bias are: 1) unaccounted-for temperature inversions in the vertical profile, and 2) the presence of unaccounted-for semi-transparent clouds cirrus in the profile which impart a cold bias to satellite longwave infrared measurements of the lower cloud. Both factors can result in the assignment of an infrared cloud top temperature for the detected cloud which is too low, and therefore a cloud top which is too high. The vertical misplacement of a liquid water cloud into an environment where liquid can only exist in a super-cooled state gives rise to false alarms in the CIP product.

This study investigates the second issue mentioned above—the impact that overlying optically thin cold clouds may have on the determination of supercooled cloud liquid water regions as specified within CIP. Here, we utilize a recently available multi-satellite cloud layer product, the 2B-GEOPROF-Lidar combined cloud geometric profile product derived from CloudSat and the Cloud Aerosol Lidar and Infrared Pathfinder Satellite (CALIPSO). This combination of active satellite observing systems allows us to identify multi-layer cloud profiles readily and assess their possible impacts to CIP. The 2B-GEOPROF-Lidar data are collocated to the CIP 20 km grid. Using only the cloudy pixels, the resulting product is then broken into two groups, one with high thin cloud overlying a cloudy scene, and a second with a cloudy scene and no high thin cloud. These two sets of data are then compared for differences in their cloud top height biases. The results will provide better insight on how to develop an improved infrared cloud height determination and cloud filter quality-control scheme within CIP.