Detection and analysis of cloud boundary in Xi'an, China, employing 35 GHz cloud radar aided by 1064 nm lidar

Abstract

Abstract. Lidar at 1064 nm and Ka-band millimetre-wave cloud radar (MMCR) are powerful tools for detecting the height distribution of cloud boundaries and can monitor the entire life cycle of cloud layers. In this study, lidar and MMCR are employed to jointly detect cloud boundaries under different conditions. By enhancing the echo signal of lidar at 1064 nm and combining its signal-to-noise ratio (SNR), the cloud signal can be accurately extracted from the aerosol signals and background noise. The interference signal is eliminated from Doppler spectra of the MMCR by using the noise ratio of the smallest measurable cloud signal (SNRmin⁡) and the spectral point continuous threshold (Nts). Moreover, the quality control of the reflectivity factor of MMCR obtained by the inversion is conducted, which improves the detection accuracy of the cloud signal. We analysed three typical cases studies; case one presents two interesting phenomena: (a) at 19:00–20:00 CST (China standard time), the ice crystal particles at the cloud top boundary are too small to be detected by MMCR, but they are well detected by lidar. (b) At 19:00–00:00 CST, the cirrus cloud changes to altostratus where the cloud particles eventually grow into large sizes, producing precipitation. Further, MMCR has more advantages than lidar in detection of the cloud top boundary within this period. Considering the advantages of the two devices, the change characteristics of the cloud boundary in Xi'an from December 2020 to November 2021 were analysed, with MMCR detection data as the main data and lidar data as the assistant data. The seasonal variation characteristics of clouds show that, in most cases, high clouds often occur in summer and autumn, and the low clouds are usually in winter. The normalized cloud cover shows that the maximum and minimum cloud cover occur in summer and winter, respectively. Furthermore, the cloud boundary frequency distribution results for the whole of the observation period show that the cloud bottom boundary below 1.5 km is more than 1 %, the frequency within the height range of 3.06–3.6 km is approximately 0.38 %, and the frequency above 8 km is less than 0.2 %. The cloud top boundary frequency distribution exhibits the characteristics of a bimodal distribution. The first narrow peak lies at approximately 1.0–3.1 km, and the second peak appears at 6.4–9.8 km.

A Cloud Detection Method for Vertically Pointing Millimeter-Wavelength Cloud Radar

“…Except for the simplest single-point threshold determination in the early days, all cloud signal detection methods use the principle of cloud continuity to improve detection performance and reduce FAR and MDR. A common method is to use the continuity of Doppler velocity spectral width of cloud signals [17,18]. Only the spectral segment with N = 2N h + 1 continuous bins all over the threshold N s can be masked as the cloud signal.…”

“…To decrease both the FAR and MDR, most cloud detection methods utilize the spatiotemporal continuity of the cloud to discriminate the cloud from noise to achieve better FAR and MDR. Some of them perform the detection method at the Doppler power spectrum stage [17,18]. For example, Shupe [17] thought that spectral peaks of cloud signal should have a minimum width (in his use case, seven points) over the noise level.…”

Doppler power spectrum processing methods for a vertical sensing 94 GHz millimeter-wavelength cloud radar

“…After getting the noise level, it is time to distinguish between noise and signal. A common method is to use the continuity of velocity spectral width of cloud signals (Shupe et al, 2004;Yuan et al, 2022). Only spectral segment with N = 2N h + 1 continuous bins all over the threshold N s can be masked as cloud signal.…”

A 94 GHz Pulse Doppler Solid-State Millimeter-Wave Cloud Radar

“…Due to the fluctuation of both the cloud signal and noise, simply using a threshold method will lead to an increased false alarm rate (FAR) and missed detection rate (MDR). One common method [15,16] is to exploit the continuity of the velocity spectrum of cloud signals, masking only N continuous bins over the threshold as cloud signals. In our application, however, the continuum points method did not fit because of its poor performance for weak cloud signals.…”