May 26 - July 31
Gust Front Detection Using Neuro-Fuzzy Algorithm With Polarimetric WSR-88D
Amber Liggett, Tian-You Yu, and Yunsung Hwang
What is already known:
- Studying polarimetric gust front signatures is timely as the upgrade of National Network Weather Surveillance Radar-1988 Doppler (WSR-88D) to polarimetric capabilities was recently completed in 2013.
- Neuro-Fuzzy Gust-front Detection Algorithm (NFGDA) is an artificial intelligence algorithm that was developed at University of Oklahoma Advanced Radar Research Center (ARRC) in 2013 to emulate human decision-making in order to detect polarimetric gust fronts.
- NFGDA preliminary results yielded a higher performance rate than the current machine intelligence algorithm, motivating the investigation of additional gust front cases to confirm the neuro-fuzzy algorithm is more accurate.
What this study adds:
- Eight more polarimetric gust front cases are evaluated under the NFGDA, yielding similar signatures results as preliminary cases.
- The new cases further verify gust front polarimetric signatures.
- A NFGDA performance evaluation led to proposed refinements to the algorithm in order to yield a more realistic performance evaluation.
Abstract:
The strong wind, shear, and turbulence associated with gust fronts can negatively impact aircraft operations at terminals, vegetation and other structures. Currently, the Machine Intelligence Gust-Front detection Algorithm (MIGFA) identifies gust front based on signatures from Doppler radar measurements. The upgrade of the Weather Surveillance Radar-1988 Doppler (WSR-88D) network to polarimetric capabilities was recently completed in 2013. Therefore it is timely to exploit the additional polarimetric measurements to improve gust front detection. The Neuro-Fuzzy Gust-front Detection Algorithm (NFGDA) was developed for this task. NFGDA preliminary results yielded a higher performance than MIGFA, motivating this study to investigate more gust front cases to confirm polarimetric signatures of gust fronts and verify the performance of NFGDA. In this study, eight gust front cases are identified and analyzed using the NFGDA. Findings included similarities between these and preliminary polarimetric gust front signatures. Additionally, the performance results yielded suggested refinements based on the statistical analysis of the algorithm. More specific guidelines can be placed in defining a gust front, as it is not a well-defined storm feature. Overall, there is room promising algorithm.