While the assimilation of dual-polarimetric radar observations in weather forecast models is promising especially for short-term forecasts of precipitation, the direct assimilation of polarimetric variables is still challenging because of the rather rudimentary appreciation of particle size and shape distributions by the models. Thus, current studies recede to assimilating model state variables derived from dual-polarimetric observables, such as hydrometeor mixing ratios. This study evaluates, improves and adapts estimators for liquid water content for their application to observations of the polarimetric C‑band radar network of the German national meteorological service as a first step towards their assimilation in the ICON model. T‑matrix simulations are used to derive polarimetric observables from a large data set of drop size distributions (DSDs) observed over Germany. Existing estimators based on reflectivity (Z), differential reflectivity (ZDR), specific attenuation (A) and specific differential phase (KDP) applied to this data set yield mostly unsatisfactory results and motivated the search for and development of improved estimators. The latter much better approximate the simulated data, and also mostly outperform the existing estimators when applied to real radar observations over Germany, although by a smaller degree. The new KDP-based estimator could only outperform existing algorithms, when KDP for azimuth-range intervals with negative KDPs were replaced by a Z-based KDP estimation. Further potential radar observation deficiencies in determining Z, ZDR, A and KDP motivated the development of a potentially more stable hybrid estimator based on the new LWC(Z,ZDR), LWC(A) and LWC(KDP) estimators. This hybrid estimator outperforms both the adapted and existing LWC estimators in terms of the correlation coefficient.

中文翻译：

---

德国上空液态水含量的双极化雷达估计器

虽然在天气预报模型中双极化雷达观测的同化很有前景，尤其是对于降水的短期预测，但由于模型对粒子大小和形状分布的初步评估，极化变量的直接同化仍然具有挑战性。因此，当前的研究退步到同化源自双极化观测值的模型状态变量，例如水凝物混合比。本研究评估、改进和调整液态水含量估算器，将其应用于德国国家气象局极化 C 波段雷达网络的观测，作为将其纳入 ICON 模型的第一步。T 矩阵模拟用于从在德国观察到的大量液滴尺寸分布 (DSD) 数据集导出极化观测值。基于反射率 (Z)、微分反射率 (ZDR)、特定衰减 (A) 和特定微分相位 (KDP) 的现有估计器应用于该数据集产生的结果大多不令人满意，并激发了对改进估计器的搜索和开发。后者更好地近似模拟数据，并且在应用于德国上空的真实雷达观测时也大多优于现有估计器，尽管程度较小。当具有负 KDP 的方位角间隔的 KDP 被基于 Z 的 KDP 估计取代时，新的基于 KDP 的估计器只能胜过现有算法。在确定 Z、ZDR、A 和 KDP 推动了基于新 LWC(Z,ZDR)、LWC(A) 和 LWC(KDP) 估计器的潜在更稳定混合估计器的开发。这种混合估计器在相关系数方面优于适应的和现有的 LWC 估计器。