Abstract: Waveform macro-anomalies refer to false single-station triggers caused by non-seismic events that can lead the processing system to misinterpret data. When multiple stations experience coupled waveform macroanomalies, they can easily result in false alarms. To enhance the operational quality and data reliability of the Shaanxi earthquake early warning network, we collected various abnormal waveforms encountered during the trial operation of the network. These anomalies are typically categorized into five types: zero drift, spike, gourd-shaped, step, and jump. By analyzing continuous waveform data from the low-noise period each early morning, combined with spectral analysis and noise characterization, we identified the typical time- and frequency-domain manifestations of each anomaly type and explored their potential causes. Based on these findings, specific recommendations for on-site inspection and operational maintenance were proposed for different anomaly types. The implementation of this research effectively improves the data quality of the Shaanxi warning network, helps prevent false triggers of the early warning system, ensures its stable and reliable operation, and provides a foundation for further research on the automated identification of abnormal waveforms.