The Radar Echo Classifier: A Fuzzy Logic Algorithm for the WSR-88D
Cathy Kessinger, NCAR, Boulder, CO; and S. Ellis and J. Van Andel
Atmospheric conditions favorable for refraction of the radar beam can produce additional ground clutter return, called anomalously- propagated (AP) return. This AP return is a contaminant within the radar moment data that causes erroneous estimates of rainfall accumulation, false wind shears, and can confound operational users of the data such as air traffic controllers. Within the current Weather Surveillance Radar-1988 Doppler (WSR-88D) (also called "NEXRAD") quality control system, AP clutter return is removed by manual application of additional clutter filters. Automation of clutter filter control is desired. To achieve automation, a recognition algorithm must first determine where the AP ground clutter return is located. The algorithm that performs this task is contained within the Radar Echo Classifier (REC) that is currently being installed within the WSR-88D Open Radar Product Generator (ORPG), Build 2. The ORPG Build 2 will be deployed in September 2002.
The Radar Echo Classifier (REC) uses fuzzy-logic techniques to determine the type of scatterer measured by the WSR-88D. Currently, three algorithms have been designed and tested: the AP detection algorithm (APDA) locates anomalously-propagated (AP) ground clutter return, the precipitation detection algorithm (PDA) determines convective and stratiform precipitation regions, and the insect clear air detection algorithm (ICADA) defines return from insects in the boundary layer. These algorithms have been developed using data from WSR-88D systems located across the USA and from various field campaigns of the NCAR S-Pol polarimetric radar.
Expert users of the WSR-88D data provided the "truth" data sets used to optimize the algorithm performances. For the S-Pol data sets, the polarimetric variables are input into a fuzzy logic polarimetric identification (PID) algorithm to determine the type of radar echo return that is present. The PID output is used as the "truth" field for optimization of algorithm performance. Results will be presented and statistical estimates of performance shown.
Extended Abstract (544K)
Poster Session 1, All AI applications
Tuesday, 11 February 2003, 9:45 AM-11:00 AM
Previous paper Next paper
Browse or search entire meeting
AMS Home Page