This study presents an improved fuzzy logic-based algorithm, originally developed for the U.S. Weather Sur
veillance Radar-1988 Doppler (WSR-88D) system, to classify meteorological echoes (MS) and non-meteorological
echoes (non-MS) in S-band dual-polarization radar data from China New Generation Weather Radar (CINRAD)
with S-band of type A (SA). In the improvement process, the "true" MS and non-MS are identified firstly using the
combination of a single-polarization radar technique for distinguishing the MS and non-MS from 73,423 radar
records and then manual inspection. Subsequently, a statistical analysis of dual-polarization variables and their
derived parameters is conducted to obtain the characteristics of the MS and non-MS. Finally, the membership
function parameters in the fuzzy logic-based algorithm are refined based on these characteristics. The perfor
mance of the improved algorithm is evaluated under four weather scenarios: clear-sky, weak precipitation, heavy
precipitation and typhoon. The results demonstrate that the improved algorithm effectively distinguishes be
tween non-MS and MS, with outcomes that align well with real echo data. In practical applications, the improved
algorithm markedly reduces residual non-MS contamination while preserving the MS. In order to assess the
improved algorithm more comprehensively, 7339 radar samples randomly collected at Nanchang radar station
from January to November 2023 are used for the statistical evaluation of the algorithm. Results reveal that the
improved algorithm eliminates the majority of the non-MS while maintaining the integrity of MS structures. In
contrast, the original algorithm has limited capability in filtering the non-MS, particularly near radar stations and
mountainous regions. Overall, the results demonstrate that the improved algorithm substantially enhances data
quality control and accuracy in the application of CINRAD/SA radar products.
发表评论