IHL plot

Research is ongoing to utilize the recent dual-polarimetric upgrade to the National Weather Service operational radar network (WSR-88D) for remote detection of in-flight aircraft icing. It has been well documented that the S-band dual-polarimetric radar signatures at individual range gates of super-cooled liquid water and ice crystals overlap significantly, complicating the identification of icing conditions using individual radar measurements. Recently several investigators have found that the aggregate characteristics of dual-polarimetric radar measurements over regions on the order of several kilometers show distinguishing features between regions containing super-cooled liquid and those with ice only. In NCAR's ongoing study, the features found in the literature are combined using a fuzzy-logic framework to provide an icing threat likelihood. The results of this algorithm are currently being output in realtime for two operational NEXRADs - KCLE in Cleveland, Ohio and KFTG in Denver, CO. During 2016, RadIA will be implemented into NSSL's MRMS platform. The ultimate goal is to produce an end-to-end algorithm which outputs an accurate and reliable icing threat product that can then be combined with existing icing detection systems to improve their performance.

IHL Final Output

The inputs to RadIA are radar and Numerical Weather Prediction (NWP) model data from the Weather Research and Forecast Rapid Refresh (WRF-RR) model. A simplified version of the RadIA algorithm is shown in the flow chart (at right). Membership functions for specific radar moment fields are utilized to create 'Freezing Drizzle', 'SLW', 'Mixed Phase' and 'Plate-shaped Crystal' Detection Interest maps, and those maps are combined to create the final RadIA product.

Future Improvements

• Redesign the freezing level routine so that the freezing level becomes a function of azimuth angle. An azimuth angle increment of 30 degrees may be fine enough to capture the relevant freezing level variability of the environment.
• The freezing level algorithm can effectively identify bands of elevated Zdr that are indicative of icing hazard. This ability can be added to the elevated Zdr icing detection routine.
• The freezing level algorithm along with the PID can be used to identify multiple freezing levels. Output from the NWP model may also be used. Data cases are needed that exhibit the multiple freezing level polarimetric signatures.
• Incorporate more inference of microphysical conditions via polarization variables.
• Test the dependence of the feature fields with various operating configurations of the WSR-88D radars. The measurement variance of the radar data depends on the number of samples used, the scan rate, etc. Thus it is recommended to investigate the feature fields utilizing the local spatial standard deviation of radar measurements for the various Volume Coverage Patterns (VCPs). It may be necessary to define separate membership functions for different VCPs.
• Account for radar-by-radar biases in RadIA's internal algorithms.

Other applications

• Combine RadIA output with NCAR's CIP and MICRO products
• Combine RadIA output with TAIWIN – Terminal Area Icing Weather Information Network – which will develop improved detection and forecasting of freezing precipitation in the airport terminal area.