科研成果与进展.pdf

科研成果与进展Scientific Achievements and Advances4灾害天气 Severe Weather灾害天气研究进展Advances in Research on Severe Weather1 灾害天气监测1 Severe weather monitoring technology1.1 A bayes-based approach against sample imbalance to improving the potential forecasts of galeSample imbalance prevents the Bayesian model from making effective potential forecasts of gale.An approach is thus proposed to modify the Bayesian model to deal with sample imbalance and verified using the 20152019 of reanalysis and radar data.The approach reduces sample imbalance by the resampling based on the value ranges of environmental parameters and the application of multi-layer conditional probabilities.The samples that are insensitive to gale forecasts are excluded according to the value ranges of environmental parameters for gale occurrence.This resampling greatly improves gale occurrence hits and suppresses its false alarms,leading to significant improvement in gale forecasting skill.On the basis of the resampling,the application of the multi-layer conditional probabilities that forecast gale occurrence balances the samples to an equivalent magnitude.Consequently,the false alarms are further suppressed although some hits are reduced,resulting in a higher forecasting skill of gale.(Liang Zhaoming,Hu Zhiqun)1.2 Combined radar quality index for quantitative precipitation estimation of heavy rainfall eventsFor quantitative precipitation estimation(QPE)based on polarimetric radar(PR)and rain gauges(RGs),the quality of the radar data is crucial for estimation accuracy.This paper proposes a combined radar quality index(CRQI)to represent the quality of the radar data used for QPE and an algorithm that uses CRQI to improve the QPE performance.Nine heavy rainfall events that occurred in Guangdong Province,China,were used to evaluate the QPE performance in five contrast tests.The QPE performance was evaluated in terms of the overall statistics,spatial distribution,near real-time statistics,and microphysics.CRQI was used to identify good-quality data pairs(i.e.,PR-based QPE and RG observation)for correcting estimators(i.e.,relationships between the rainfall rate and the PR parameters)in real-time.The PR-based QPE performance was improved because estimators were corrected according to variations in the drop size distribution,especially for data corresponding to 1.1 mm average Dm 1.4 mm,and 4 average lgNw 0.6(D 0.6)mm are higher(lower)in Nagqu than in Medog.The fitted normalized gamma distributions of the averaged DSDs for the five rainfall rate categories show that Nagqu has a larger(lower)mass-weighted mean diameter Dm(normalized intercept parameter,lgNw)than does Medog.The difference in Dm between Nagqu and Medog increases with the rainfall rate.Convective clusters in Nagqu could be identified as continental-like,while convective precipitation in Medog could be classified as maritime-like.The relationships between the shape factor and slope parameter of the gamma distribution model,the radar reflectivity Z,and the rainfall rate R are also derived.Furthermore,the possible causative mechanism for the notable DSD variation between the two regions during the rainy season is illustrated using reanalysis data and automated weather station observations.Cold rain processes are mainly responsible for the lower concentrations of larger drops observed in Nagqu,whereas warm rain prevails in Medog,producing abundant small drops.(Wang Gaili,Li Ran,Sun Jisong)1.4 Identification of convective and stratiform clouds based on the improved DBSCAN clustering algorithmA convective and stratiform cloud classification method for weather radar is proposed based on the density-based spatial clustering of applications with noise(DBSCAN)algorithm.To identify convective and stratiform clouds in different developmental phases,two-dimensional(2D)and three-dimensional(3D)models are proposed by applying reflectivity factors at 0.5 and at 0.5,1.5,and 2.4 elevation angles,respectively.According to the thresholds of the algorithm,which include echo intensity,the echo top height of 35 dBz(ET),density threshold,and epsilon neighborhood,cloud clusters can be marked into four types:deep-convective cloud(DCC),shallow-convective cloud(SCC),hybrid convective-stratiform cloud(HCS),and stratiform cloud(SFC)types.Each cloud cluster type is further identified as a core area and boundary area,which can provide more abundant cloud structure information.The algorithm is verified using the volume scan data observed with new-generation S-band weather radars in Nanjing,Xuzhou,and Qingdao.The results show that cloud clusters can be intuitively identified as core and boundary points,which change in area continuously during the process of convective evolution,by the improved DBSCAN algorithm.Therefore,the occurrence and disappearance of convective weather can be estimated in advance by observing the changes of the classification.Because density thresholds are different and multiple elevations are utilized in the 3D model,the identified echo types and areas are dissimilar between the 2D and 3D models.The 3D model identifies larger convective and stratiform clouds than the 2D model.However,the developing convective clouds of small areas at lower heights cannot be identified with the 3D model because they are covered by thick stratiform clouds.In addition,the 3D model can avoid the influence of the melting layer and better suggest convective clouds in the developmental stage.(Zuo Yuanyuan,Hu Zhiqun,Yuan Shujie)科研成果与进展Scientific Achievements and Advances61.5 Improved spectral processing for a multi-mode pulse compression Ka-Ku-band cloud radar systemCloud radars are widely used in observing clouds and precipitation.However,the raw data products of cloud radars are usually affected by multiple factors,which may lead to misinterpretation of cloud and precipitation processes.In this study,we present a Doppler-spectra-based data processing framework to improve the data quality of a multi-mode pulse-compressed Ka-Ku radar system.Firstly,non-meteorological signal close to the ground was identified with enhanced Doppler spectral ratios between different observing modes.Then,for the Doppler spectrum affected by the range sidelobe due to the implementation of the pulse compression technique,the characteristics of the probability density distribution of the spectral power were used to identify the sidelobe artifacts.Finally,the Doppler spectra observations from different modes were merged via the shift-then-average approach.The new radar moment products were generated based on the merged Doppler spectrum data.The presented spectral processing framework was applied to radar observations of a stratiform precipitation event,and the quantitative evaluation shows good performance of clutter or sidelobe suppression and spectral merging.(Ding Han,Li Haoran,Liu Liping)1.6 Raindrop size distribution prediction by an improved long short-term memory networkThe observation and research on raindrop size distribution(DSD)are important for mastering and understanding the mutual restriction relationship between cloud dynamics and cloud microphysics in a process of precipitation;it also plays an irreplaceable role in many fields,such as radar meteorology,weather modification,boundary layer land surface processes,aerosols,etc.Using more than 1.7 million minutes of raindrop data observed with 17 laser disdrometers at 17 stations in Anhui Province,China,from 7 August 2009 to 30 April 2020,a DSD training dataset was constructed.Furthermore,the data are fitted to a normalized Gamma function and used to obtain its three parameters,i.e.,the normalized intercept Nw,the mass weighted average diameter Dm,and the shape factor.Based on the long short-term memory network(LSTM),a DSD Gamma distribution prediction network(DSDnet)was designed.In the process of modeling based on the DSDnet,a self-defined loss function(SLF)was proposed in order to improve the DSD prediction by increasing the weight values in the poor fitting regions according to the common mean square error loss function(MLF).By means of the training dataset,a DSDnet-based model was trained to realize the prediction of Nw,Dm,and minute-to-minute over the course of 30 min,and then was evaluated by the test dataset according to three indicators,namely,mean relative error(MRE),mean absolute error(MAE),and correlation coefficient(CC).The CC of lgNw,Dm,and can reach 0.93403,0.90934,and 0.89741 for 12-min predictions,and 0.87559,0.85261,and 0.84564 for 30-min predictions,respectively,which means that the DSD prediction accuracy within 30 min can basically reach the application level.Furthermore,the 12-and 30-min predictions of 3 precipitation processes were taken as examples to fully demonstrate the application effect of model.The prediction effects of Nw and Dm are better than that of,and the stratiform precipitation is better than the convective and convective-stratiform mixed cloud precipitation.(Zhu Yongjie,Hu Zhiqun,Yuan Shujie)1.7 Seasonal variation in microphysical characteristics of precipitation at the entrance of water vapor channel in Yarlung Zangbo Grand CanyonMedog is located at the entrance of the water vapor channel in the Yarlung Zangbo Grand Canyon(YGC).This area has the largest annual accumulated rainfall totals and precipitation frequency on the Tibetan Plateau(TP).This paper investigates the seasonal variation in raindrop size distribution(DSD)characteristics in Medog based on disdrometer observations from 1 July 2019 to 30 June 2020.The DSD characteristics are examined under six rain rate classes and two rainfall types(stratiform and convective)in the winter,2022 CAMSAnnual Report7premonsoon,monsoon and postmonsoon periods.The highest(lowest)concentration of small raindrops is observed in monsoon(winter)precipitation,whereas large raindrops predominate in premonsoon precipitation.For stratiform rainfall,the mean mass-weighted mean diameter(Dm)exhibits overlooked differences in the four periods,while the mean normalized intercept parameter(Nw)is significantly higher in the monsoon period than in the other three periods.The convective rainfall in the monsoon and postmonsoon periods is characterized by a high concentration of limited-size drops and can be classified as maritime-like.This is probably attributed to abundant warm and humid airflow transported by the Indian Ocean monsoon into Medog.The westerly winds prevail over the TP during the premonsoon period,and thereby the premonsoon convective rainfall in Medog has a larger mean Dm and a lower mean Nw.In addition,the relationships of radar reflectivity Z and rain rate R for different precipitation types in different periods are also derived.A better understanding of the seasonal variation in the microphysical characteristics of precipitation in Medog is important for improving the microphysical parameterization scheme and the precipitation forecast of models on the TP.(Li Ran,Wang Gaili,Zhou Renran)1.8 S波段双偏振雷达和X波段相控阵天气雷达中气旋识别结果对比为了比较S波段双偏振雷达资料和X波段相控阵天气雷达资料识别中气旋的差异，结合X波段相控阵天气雷达（XPAR）和S波段双偏振天气雷达（SPOL）及地面观测资料，对比分析了2019年4月19日发生在广州的一次中小尺度天气过程。结果显示：使用的识别算法可以正确识别出中气旋；XPAR的高时空分辨率数据可以弥补SPOL仰角层不足的缺陷，观测到更加完整的中气旋垂直结构，识别结果中的参数也比SPOL更加细致，更精准地揭示了中气旋的短时演变。研究结果表明XPAR对于强天气回波的观测识别性能相比SPOL具有持续时间更长、垂直结构更加精细、正负速度对差值更大、随整个天气过程演变更加细致等优势，有利于对中小尺度天气系统的快速发展、演变开展细致深入的研究。（苏永彦，刘黎平）1.9 对流降水云中大气垂直运动反演及个例试验为深入认识对流降水云结构及动力特征，基于降水频段调频连续波5520 MHz垂直指向雷达（VPR-CFMCW），使用地面至15 km高度的反射率因子及径向速度，建立对流降水云中大气垂直运动的反演方法，分析对流垂直结构及大气垂直运动随高度分布的演变特征。对在广东龙门测站探测的2019年4月2022日前汛期4次对流降水进行反演试验发现：对流降水前大气上升运动对降水云反射率因子及地面降水有正贡献，深厚对流具有倾斜性，会导致垂直剖面在某些时刻呈分层结构；对流降水整层以下沉运动为主导，高层大气上升运动与下沉运动交替出现，低层大气下沉运动占比最高，大气上升运动在6 km高度以上占比有所增加；大气垂直速度在高层较大、在低层较小，超过10 m/s的强上升运动与下沉运动多出现在6 km高度以上，4 6 km高度垂直运动变化较大，4 km高度以下的平均下沉运动小于5 m/s，上升运动约为2 m/s。（董佳阳，崔晔，阮征）1.10 华北地区一次长生命史超级单体雹暴特征观测对2020年6月25日河北南部从保定涞源到沧州东光的超级单体雹暴环境条件、雷达观测及风场反演特征进行了分析。结果表明：（1）本次超级单体雹暴是在水平对流卷上触发起来并经过不断加强形成的。（2）超级单体雹暴长时间处于中等到强的基本顺时针旋转环境风的垂直风切变环境中、中气旋维持时间超过3 h以及较强冷池作用，是本次超级单体雹暴维持长生命史的可能原因。（3）对流单体VIL持续7 h基本维持在55 80 kg/m2区间，与单体移动路径上降雹时段一致，超过4 h持续出现三体散射长钉或旁瓣回波特征，与本次过程中灾情调查和气象站观测到的大冰雹时间段基本一致，对大冰雹有较好指示意义。（4）超级单体雹暴出现钩状回波（低层）、回波悬垂“穹隆”结构、风暴顶辐散、科研成果与进展Scientific Achievements and Advances8ZDR柱和KDP柱等特征。“穹隆”顶部为65 dBz以上反射率因子，该处风场以垂直上升为主，水平风分量较弱，呈现出“穴道零域”结构，利于大冰雹增长。超级单体垂直速度随高度增大，利于中气旋形成和维持。（刘瑾，王丛梅，胡志群）1.11 利用深度学习开展偏振雷达定量降水估测研究利用20182020年经偏振升级改造后的广州S波段双偏振雷达（CINRAD/SAD）82892个体扫的0.5 仰角数据，以及雷达100 km探测范围内1109个雨量站共计538560个分钟雨量数据，分别构建了单参量、三参量雷达定量降水估测（QPE）深度学习网络架构（Z-Rnet、KDP-Rnet、Pol-Rnet），并以KDP=0.5/km为阈值分别训练得到大雨、小雨、总体等9个定量降水估测模型。在常用的均方误差作为损失函数的基础上，对不同降水强度采用不同权重提出了自定义损失函数，并利用比率偏差、相对偏差、均方差、平均绝对误差和平均相对误差作为评价指标对模型进行评估。通过对以积层混合云为主、以对流云为主和以层状云为主的3次降水过程的模型验证结果表明，利用深度学习训练的模型有较好的定量降水估测效果，区分雨强的小雨、大雨模型比不区分雨强的总体模型的效果要好。采用自定义损失函数模型效果更好，其均方差、平均绝对误差和平均相对误差分别较采用传统均方误差损失函数提升了8.62%、12.52%、16.34%。自定义损失函数中，采用ZH-ZDR-KDP3参量网络架构训练得到的定量降水估测模型效果最好，其均方差、平均绝对误差和平均相对误差分别较采用ZH的单参量Z-Rnet架构提升6.82%、8.43%、7.22%；较采用KDP的单参量KDP-Rnet架构提升12.33%、17.61%、17.26%。（皇甫江，胡志群，郑佳锋）1.12 利用深度学习填补双偏振雷达回波遮挡广州S波段双偏振天气雷达低仰角多方位存在遮挡，高仰角也存在部分遮挡。基于卷积神经网络等深度学习方法，构建垂直填补（VEF）和水平填补（HEF）网络架构，基于两种架构，利用无遮挡区的反射率因子ZH、差分反射率ZDR，差传播相移率KDP构建训练集，填补遮挡区的ZH和ZDR。针对仅0.5 仰角存在遮挡的区域，基于VEF架构，利用上层多个仰角、径向、距离库的三维数据，分距离段训练垂直填补模型。针对遮挡仰角较高的区域，则基于HEF架构，利用同一仰角左右相邻的多个径向、距离库的数据，分遮挡径向训练水平填补模型。根据解释方差、平均绝对偏差和相关系数3个指标和3个个例，对模型效果进行评估。结果表明：ZH填补模型的解释方差最大为0.92，平均绝对偏差最小为1.69 dB，相关系数最高为0.96；ZDR填补模型的解释方差最大为0.92，平均绝对偏差最小为0.12 dB，相关系数最高为0.96。利用该研究构建的深度学习填补架构，可有效填补偏振雷达遮挡区域回波，提高雷达数据质量。（尹晓燕，胡志群，郑佳锋）1.13 青藏高原东南部墨脱地区弱降水微物理特征的Ka波段云雷达观测研究藏东南地区的墨脱县位于雅鲁藏布江下游的河谷区域，是印度洋水汽进入高原的最主要水汽通道。墨脱作为西藏年平均降水量最多的地区，是青藏高原云降水系统的重要组成部分。本文以2020年墨脱地区的Ka波段云雷达观测数据为基础，首先对云雷达功率谱数据进行预处理，并采用降水现象仪对云雷达观测进行验证。在此基础上，选取了2020年3月6日和8月24日具有层状云降水特性的两次弱降水过程，利用云雷达功率谱数据反演了雨滴谱，探究墨脱地区旱季和雨季弱降水的微物理特征。结果表明：云雷达观测与降水现象仪雨滴谱数据计算的Ka波段云雷达回波强度理论值存在大约12 dB的系统误差，订正之后二者随时间变化一致性较好，云雷达反演的近地面雨滴谱特征与降水现象仪观测接近。墨脱地区零度层高度随季节变化明显，旱季零度层高度较低（例如地面上1.5 km左右），而雨季零度层高度较高（例如地面上4 km左右）。墨脱层状云雨滴谱的宽度较窄，降水粒子直径不超过3 mm。在零度层以上，根据谱偏度和峰度的垂直变化可以推测冰晶粒子直径随高度下降缓慢增长，但旱季冰2022 CAMSAnnual Report9晶粒子增长比雨季更为明显。经过零度层后，冰晶粒子转化为雨滴，雨滴在下落过程中由于碰并及蒸发作用造成浓度减小，直径越小的粒子浓度减小越快。在近地面，由于蒸发作用的加强导致随高度降低雨滴浓度明显减小。（张静怡，王改利，郑佳锋）1.14 青藏高原那曲对流云中过冷水的毫米波雷达反演研究对流云中过冷水的识别一直是气象探测的难点。基于Ka波段毫米波雷达功率谱数据，结合探空资料，提出了高原对流云内过冷水的识别和反演算法；利用那曲两个个例对算法效果进行了分析，并结合同址的微波辐射计资料对雷达结果进行了初步验证；最后探讨了算法与以往方法的差异。结果表明：高原层积云、浓积云和高积云内由上升气流主导，云内粒子相态变化快，过冷水粒子的回波强度、粒径和含水量分布较广。对于不同云类，过冷水的空间分布存在一定差异。过冷水的回波强度、粒径和含水量都与上升气流速度呈正相关，在时间变化趋势和空间分布上都有很好的对应。微波辐射计和雷达的液态水路径在时间变化趋势和峰值大小上都较为一致，相关系数为0.63 0.79。与以往方法相比，算法对过冷水位置和参数反演的结果更为合理。（任涛，郑佳锋，刘黎平）1.15 双偏振相控阵雷达误差评估与相态识别方法选取2020年39月深圳求雨坛的X波段双偏振相控阵雷达探测数据，与同位置的S波段双偏振雷达进行对比。通过一定限制条件定量分析引入误差的原因，发现反射率因子ZH和差分反射率ZDR的标定误差和随机误差较大，其中ZH误差变化范围为-0.5 4.5 dB，ZDR误差变化范围为-0.7 0.2 dB。在上述较大误差影响下，传统模糊逻辑相态识别方法的水凝物相态识别结果不可靠，因此根据不同相态的雷达参量特征范围以及融化层高度建立基本结构为二叉树的决策树相态识别方法。针对上述方法的实际应用效果，分别从水凝物相态识别结果对误差的敏感性和空间分布的合理性进行评估。结果表明：决策树相态识别方法的水凝物相态识别结果稳定性高于模糊逻辑相态识别方法，且在对流云中的水凝物相态分布更加合理，能够发挥X波段双偏振相控阵雷达在研究云内水凝物相态演变的优势。（李哲，吴翀，刘黎平）2 青藏高原天气研究2 Research on weather over the Tibetan Plateau2.1 A vertical transport window of water vapor in the troposphere over the Tibetan Plateau with implications for global climate changeBy using the multi-sourc