基于多源资料的地基微波辐射计温度廓线数据质量评估

Quality Assessment of Ground-Based Microwave Radiometer Temperature Profiles Based on Multi-Source Data

摘要: 本文基于探空资料和ERA5再分析资料等多源数据，选取四个代表站点(北京站、新疆阿克苏站、吉林长春站和广东清远站)，对地基微波辐射计(MWR)温度廓线的数据质量进行了对比评估。与探空资料(RS)对比结果表明，北京站的MWR平均温度误差最小，基本维持在1℃以内；阿克苏站在2 km以上高度差异最大，最大偏差约为5℃；吉林长春站MWR资料偏差稳定性存在较大波动；广东清远站的MWR资料整体质量较好。与再分析资料对比表明，在所选时空范围内，基于ERA5再分析数据评估上述站点的MWR数据质量，与探空数据结果保持一致。因此在探空资料缺乏的情况下，ERA5再分析资料可在一定程度上，用于定量评估微波辐射计温度廓线数据质量。此外，在利用平均偏差(Bias)和均方根误差(RMSE)量化不同资料间差异的同时，本文还设计了相对加权欧式距离(RED)这一统计指标，以检验不同高度处气候态特征对温度相对偏差结果的影响。本研究提出的MWR温度廓线数据质量评估方法，既有效利用探空资料的精准性，又充分利用ERA5再分析数据的时空一致性和连续性，可推广至更多MWR观测站点，具有一定的普适性。

Abstract: This study evaluates the data quality of ground-based microwave radiometer (MWR) temperature profiles at four representative sites in China using radiosonde data (RS) and ERA5 reanalysis data. Comparisons with radiosonde observations show that the MWR data in Qingyuan generally has the best quality, with Beijing showing the smallest bias (mostly within 1˚C). The largest bias is found above 2 km in Aksu, reaching up to 5˚C, while Changchun shows significant errors below 2 km with notable fluctuations in data stability. Comparisons with ERA5 reanalysis data yield similar results, suggesting that ERA5 can be used for MWR evaluation in the absence of radiosonde data. Additionally, while using the mean bias and root mean square error (RMSE) to quantify the differences between different data, a new metric, Relative Euclidean Distance (RED), is introduced to test the influence of altitude-based climatological characteristics on temperature deviation results. The proposed MWR data quality evaluation method leverages both the accuracy of radiosonde data and the spatial-temporal continuity of ERA5, showing general applicability to other MWR observation sites.