This study presents a method to identify and distinguish insects, clouds, and precipitation in 35 GHz (Ka-band) vertically pointing polarimetric radar Doppler velocity power spectra and then produce masks indicating the occurrence of hydrometeors (i.e., clouds or precipitation) and insects at each range gate. The polarimetric radar used in this study transmits a linear polarized wave and receives signals in collinear (CoPol) and cross-linear (XPol) polarized channels. The measured CoPol and XPol Doppler velocity spectra are used to calculate linear depolarization ratio (LDR) spectra. The insect–hydrometeor discrimination method uses CoPol and XPol spectral information in two separate algorithms with their spectral results merged and then filtered into single value products at each range gate. The first algorithm discriminates between insects and clouds in the CoPol Doppler velocity power spectra based on the spectra texture, or spectra roughness, which varies due to the scattering characteristics of insects vs. cloud particles. The second algorithm distinguishes insects from raindrops and ice particles by exploiting the larger Doppler velocity spectra LDR produced by asymmetric insects. Since XPol power return is always less than CoPol power return for the same target (i.e., insect or hydrometeor), fewer insects and hydrometeors are detected in the LDR algorithm than the CoPol algorithm, which drives the need for a CoPol based algorithm. After performing both CoPol and LDR detection algorithms, regions of insect and hydrometeor scattering from both algorithms are combined in the Doppler velocity spectra domain and then filtered to produce a binary hydrometeor mask indicating the occurrence of cloud, raindrops, or ice particles at each range gate. Forty-seven summertime days were processed with the insect–hydrometeor discrimination method using US Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) program Ka-band zenith pointing radar observations in northern Oklahoma, USA. For these 47 d, over 70 % of the hydrometeor mask column bottoms were within ±100 m of simultaneous ceilometer cloud base heights. All datasets and images are available to the public on the DOE ARM repository.

中文翻译：

---

使用垂直指向的极化雷达多普勒速度谱识别昆虫、云和降水

本研究提出了一种在 35 GHz（Ka 波段）垂直指向极化雷达多普勒速度功率谱中识别和区分昆虫、云和降水的方法，然后生成指示水凝物（即云或降水）和昆虫出现的掩码每个距离门。本研究中使用的极化雷达发射线性极化波，并在共线 (CoPol) 和交叉线性 (XPol) 极化通道中接收信号。测量的 CoPol 和 XPol 多普勒速度谱用于计算线性去极化比 (LDR) 谱。昆虫-水凝物鉴别方法在两个单独的算法中使用 CoPol 和 XPol 光谱信息，将它们的光谱结果合并，然后在每个距离门处过滤成单值产品。第一种算法基于光谱纹理或光谱粗糙度区分 CoPol 多普勒速度功率谱中的昆虫和云，由于昆虫与云粒子的散射特性而异。第二种算法通过利用非对称昆虫产生的较大的多普勒速度谱 LDR，将昆虫与雨滴和冰粒区分开来。由于对于同一目标（即昆虫或水凝物），XPol 功率返回总是小于 CoPol 功率返回，因此在 LDR 算法中检测到的昆虫和水凝物比 CoPol 算法少，这推动了对基于 CoPol 的算法的需求。在执行 CoPol 和 LDR 检测算法后，来自两种算法的昆虫和水凝物散射区域在多普勒速度谱域中组合，然后过滤以产生二进制水凝物掩模，指示每个距离门处出现云、雨滴或冰粒。使用美国能源部 (DOE) 大气辐射测量 (ARM) 计划在美国俄克拉荷马州北部的 Ka 波段天顶指向雷达观测，使用昆虫-水凝物鉴别方法处理了 47 个夏季天。在这 47 天中，超过 70% 的水凝物掩膜塔底物在 使用美国能源部 (DOE) 大气辐射测量 (ARM) 计划在美国俄克拉荷马州北部的 Ka 波段天顶指向雷达观测，使用昆虫-水凝物鉴别方法处理了 47 个夏季天。在这 47 天中，超过 70% 的水凝物掩膜塔底物在 使用美国能源部 (DOE) 大气辐射测量 (ARM) 计划在美国俄克拉荷马州北部的 Ka 波段天顶指向雷达观测，使用昆虫-水凝物鉴别方法处理了 47 个夏季天。在这 47 天中，超过 70% 的水凝物掩膜塔底物在± 100 m 的同步云高仪云底高度。所有数据集和图像都可以在 DOE ARM 存储库上向公众开放。