Ground Clutter Impacts on S- and X-Band Weather Radars across Three Chinese Megacities: A 75-Radar Census of Sources, Transmitter-Dependent Discrepancies, and Network Mitigation

The operational utility of S-band weather radars—the backbone of the China New Generation Weather Radar (CINRAD) network—is severely compromised by ground clutter from dense high-rise buildings and surrounding mountainous terrain in megacities. Although over 170 units have been upgraded to dual-polarization, the extent to which such clutter systematically degrades data quality, hydrometeor classification, and quantitative precipitation estimation (QPE) has not been quantified across diverse urban environments. Meanwhile, X-band radars are being deployed extensively as gap-fillers, yet a systematic comparison of their clutter susceptibility relative to S-band systems—and whether dense X-band networking can actively compensate for S-band observational deficits—remains absent. Addressing this knowledge gap is critical for optimizing multi-band collaborative observation strategies and improving severe weather nowcasting in densely populated metropolitan areas. In this study, observations from 25 S-band and 50 X-band radars during the 2024 flood season in Beijing, Hangzhou, and Guangzhou are analyzed. A long-term statistical averaging method is applied to accumulated data from large-scale precipitation events to isolate systematic clutter signatures from random precipitation variability. The study systematically compares the ground clutter impact characteristics between S-band and X-band radars deployed across three Chinese megacities, and quantitatively evaluates the mitigation efficacy of dense X-band radar networking in clutter-affected regions of S-band radars. It is found that for S-band radars, dual-polarization anomalies at low elevations account for 35%–68%, decreasing to 6%–18% aloft, while beam blockage remains below 10% at low elevations. In contrast, X-band radars exhibit only 1%–25% low-elevation anomalies and are nearly clutter-free at high elevations. Source attribution reveals that S-band clutter originates predominantly from buildings, whereas X-band clutter sources are environment-dependent—mountains dominate when mean beam blockage is high, and buildings dominate in densely populated areas. Notably, for solid-state X-band radars, pulse-compression sidelobe effects expand the correlation coefficient (ρ_HV) impacted area far beyond that of the texture parameter of differential phase SD(Φ_DP)—a discrepancy absent in klystron-based systems. After strategic X-band networking, the combined mainlobe and sidelobe clutter proportion at 0.5 km altitude drops from 66%–94% to 5%–25%, light-to-moderate rain classification accuracy exceeds 95%, and all QPE algorithms show improved correlation with rain gauges. In conclusion, X-band radars exhibit substantially lower ground clutter susceptibility than S-band radars, and dense X-band networks effectively compensate for S-band observational gaps in complex megacity environments. These findings provide a quantitative foundation for optimizing multi-band radar network deployment and data quality control in China.