Verification of In-flight
Icing and Turbulence Algorithms:
A Gridded Approach
Research Applications
Program
National Center for
Atmospheric Research
Boulder, Colorado
Author Contact Information: Michael Chapman, NCAR, PO Box 3000, Boulder, CO.
80307. Phone: (303)497-8371; Fax: (303)497-8401; e-mail: mchapman@ucar.edu
In years past, many products, both computer-based (e.g. Integrated Turbulence Forecasting Algorithm (ITFA) and Integrated Icing Diagnosis Algorithm (IIDA)) and human-based (e.g. AIRMETs), have been developed in order to better diagnose and forecast in-flight icing and turbulence. Concurrently, the need for improved verification methods for these products has also become apparent. The most common and widely used technique for verification is a point-to-point method incorporating the use of pilot reports (PIREPs) as verifying observations. This approach has numerous limitations due to non-systematic reporting by PIREPs, PIREP biases in both location and time, and the strong biases toward reporting a non-representative number of positive events.
It has been suggested that the utilization of PIREPs with a “gridded” technique might be a useful way to avoid some of the problems that exist within the point-to-point method. The use of a uniform grid over a specific amount of time ideally would correct for the temporal and spatial problems that exist with current methods. Although the idea of having a gridded data set for verification initially appears to be a quick solution, a deeper examination of this method has exposed several problems.
The main problem, which became apparent through this study, is the scale on which the verification analysis must occur. Many of the computer-based algorithms use the 40 km grid from the Rapid Update Cycle, and forecast over three hour time periods. The AIRMETs are human-based and cover a six-hour time period on a synoptic-scale similar to ITFA and IIDA. A quick calculation of the amount of points needed in order to cover an area the size of the United States over a 40 km grid in the horizontal and 1 kft levels in the vertical (up to approx. 40 kft.) over a six hour period shows that the number of PIREPs available is considerably less than is needed to sufficiently cover a grid of this magnitude. Also, calculations of both over flights and departures from Air Route Traffic Control Centers (ARTCC) show that even if reports were received from all available flights over the CONUS the data still would be insufficient to cover a grid of this size. Through this study, an investigation is taking place over the possibility of using this technique on a smaller scale over large cities or in ARTCC regions that receive a high volume of PIREPs.