This study investigates the application of an improved multi-threshold method based on the K-means algorithm for clutter filtering in W-band cloud and fog radar observations. Utilizing W-band millimeter-wave cloud and fog radar data collected from March to July 2023 in the Qingdao area, a dataset of cloud and fog echo of different types was constructed and statistically analyzed. Subsequently, a multi-threshold clutter filtering method was proposed to identify and eliminate abnormal interferences such as noise spikes, radial interference, and suspended matter clutter. This method employs the basic data and spatiotemporal information from the cloud radar as feature variables for K-means clustering and dynamically adjusts thresholds based on the clustering results. The clutter-filtered data were further used for the verification analysis of cloud and fog identification. The results demonstrate that the proposed multi-threshold method effectively removes clutter and significantly reduces its impact on cloud and fog echo under weather conditions of clouds, fog, and coexisting cloud–fog, while controlling the loss of cloud and fog echo within the required accuracy range.

This study investigates the application of an improved multi-threshold method based on the K-means algorithm for data quality control in W-band cloud and fog radar observations. Utilizing W-band millimeter-wave cloud and fog radar data collected from March to July 2023 in the Qingdao area, a dataset of cloud and fog echo of different types was constructed and statistically analyzed. Subsequently, a multi-threshold quality control method was proposed to identify and eliminate abnormal interferences such as noise spikes, radial interference, and suspended matter clutter. This method employs the basic data and spatio-temporal information from the cloud radar as feature variables for K-means clustering and dynamically adjusts thresholds based on the clustering results. The quality-controlled data were further used for the verification analysis of cloud and fog identification . The results demonstrate that the proposed multi-threshold method effectively removes clutter and significantly reduces the impact of clutter on cloud and fog echo under weather conditions where clouds, fog, and cloud-fog coexist, while controlling the loss of cloud and fog echo within the required accuracy range.