A Prototype Regional GSI-based EnKF-Variational Hybrid Data Assimilation System for the Rapid Refresh Forecasting System：Dual-Resolution Implementation and Testing Results

A dual-resolution（DR） version of a regional ensemble Kalman filter（EnKF）-3D ensemble variational（3DEnVar） coupled hybrid data assimilation system is implemented as a prototype for the operational Rapid Refresh forecasting system. The DR 3DEnVar system combines a high-resolution（HR） deterministic background forecast with lower-resolution（LR） EnKF ensemble perturbations used for flow-dependent background error covariance to produce a HR analysis. The computational cost is substantially reduced by running the ensemble forecasts and EnKF analyses at LR. The DR 3DEnVar system is tested with 3-h cycles over a 9-day period using a 40/13-km grid spacing combination. The HR forecasts from the DR hybrid analyses are compared with forecasts launched from HR Gridpoint Statistical Interpolation（GSI） 3D variational（3DVar）analyses, and single LR hybrid analyses interpolated to the HR grid. With the DR 3DEnVar system, a 90% weight for the ensemble covariance yields the lowest forecast errors and the DR hybrid system clearly outperforms the HR GSI 3DVar.Humidity and wind forecasts are also better than those launched from interpolated LR hybrid analyses, but the temperature forecasts are slightly worse. The humidity forecasts are improved most. For precipitation forecasts, the DR 3DEnVar always outperforms HR GSI 3DVar. It also outperforms the LR 3DEnVar, except for the initial forecast period and lower thresholds.

Assimilation of Total Lightning Data Using the Three-Dimensional Variational Method at Convection-Allowing Resolution

A large number of observational analyses have shown that lightning data can be used to indicate areas of deep convection. It is important to assimilate observed lightning data into numerical models, so that more small-scale information can be incorporated to improve the quality of the initial condition and the subsequent forecasts. In this study, the empirical relationship between flash rate, water vapor mixing ratio, and graupel mixing ratio was used to adjust the model relative humidity, which was then assimilated by using the three-dimensional variational data assimilation system of the Weather Research and Forecasting model in cycling mode at 10-min intervals. To find the appropriate assimilation time-window length that yielded significant improvement in both the initial conditions and subsequent forecasts, four experiments with different assimilation time-window lengths were conducted for a squall line case that occurred on 10 July 2007 in North China. It was found that 60 min was the appropriate assimilation time-window length for this case, and longer assimilation window length was unnecessary since no further improvement was present. Forecasts of 1-h accumulated precipitation during the assimilation period and the subsequent 3-h accumulated precipitation were significantly improved compared with the control experiment without lightning data assimilation. The simulated reflectivity was optimal after 30 min of the forecast, it remained optimal during the following 42 min, and the positive effect from lightning data assimilation began to diminish after 72 min of the forecast. Overall,the improvement from lightning data assimilation can be maintained for about 3 h.

Application of Lightning Data Assimilation to Numerical Forecast of Super Typhoon Haiyan (2013)

Previous observations from World Wide Lightning Location Network(WWLLN) and satellites have shown that typhoon-related lightning data have a potential to improve the forecast of typhoon intensity. The current study was aimed at investigating whether assimilating TC lightning data in numerical models can play such a role. For the case of Super Typhoon Haiyan in 2013, the lightning data assimilation(LDA) was realized in the Weather Research and Forecasting(WRF) model, and the impact of LDA on numerical prediction of Haiyan’s intensity was evaluated.Lightning data from WWLLN were used to adjust the model’s relative humidity(RH) based on the method developed by Dixon et al.(2016). The adjusted RH was output as a pseudo sounding observation, which was then assimilated into the WRF system by using the three-dimensional variational(3DVAR) method in the cycling mode at 1-h intervals. Sensitivity experiments showed that, for Super Typhoon Haiyan(2013), which was characterized by a high proportion of the inner-core(within 100 km from the typhoon center) lightning, assimilation of the inner-core lightning data significantly improved its intensity forecast, while assimilation of the lightning data in the rainbands(100–500 km from the typhoon center) led to no obvious improvement. The improvement became more evident with the increase in LDA cycles, and at least three or four LDA cycles were needed to achieve obvious intensity forecast improvement. Overall, the improvement in the intensity forecast by assimilation of the inner-core lightning data could be maintained for about 48 h. However, it should be noted that the LDA method in this study may have a negative effect when the simulated typhoon is stronger than the observed, since the LDA method cannot suppress the spurious convection.

Doppler Radar Data Assimilation with a Local SVD-En3DVar Method

An observation localization scheme is introduced into an ensemble-based three-dimensional variational （3DVar） assimilation method based on the singular value decomposition technique （SVD-En3DVar） to im- prove assimilation skill. A point-by-point analysis technique is adopted in which the weight of each obser- vation decreases with increasing distance between the analysis point and the observation point. A set of numerical experiments, in which simulated Doppler radar data are assimilated into the Weather Research and Forecasting （WRF） model, is designed to test the scheme. The results are compared with those ob- tained using the original global and local patch schemes in SVD-En3DVar, neither of which includes this type of observation localization. The observation localization scheme not only eliminates spurious analysis increments in areas of missing data, but also avoids the discontinuous analysis fields that arise from the local patch scheme. The new scheme provides better analysis fields and a more reasonable short-range rainfall forecast than the original schemes. Additional forecast experiments that assimilate real data from i0 radars indicate that the short-term precipitation forecast skill can be improved by assimilating radar data and the observation localization scheme provides a better forecast than the other two schemes.

Impact of FY-3D MWRI Radiance Assimilation in GRAPES 4DVar on Forecasts of Typhoon Shanshan

In this study, Fengyun-3 D(FY-3 D) Micro Wave Radiation Imager(MWRI) radiance data were directly assimilated into the Global/Regional Assimilation and Pr Ediction System(GRAPES) four-dimensional variational(4 DVar) system. Quality control procedures were developed for MWRI applications by using algorithms from similar microwave instruments. Compared with the FY-3 C MWRI, the bias of FY-3 D MWRI observations did not show a clear node-dependent difference from the numerical weather prediction background simulation. A conventional bias correction approach can therefore be used to remove systematic biases before the assimilation of data. After assimilating the MWRI radiance data into GRAPES, the geopotential height and humidity analysis fields were improved relative to the control experiment. There was a positive impact on the location of the subtropical high, which led to improvements in forecasts of the track of Typhoon Shanshan.

Effect of 2-m Temperature Data Assimilation in the CMA-MESO 3DVAR System

Assimilation of surface observations including 2-m temperature(T_(2m))in numerical weather prediction(NWP)models remains a challenging problem owing to differences between the elevation of model terrain and that of actual observation stations.NWP results can be improved only if surface observations are assimilated appropriately.In this study,a T_(2m)data assimilation scheme that carefully considers misrepresentation of model and station terrain was established by using the three-dimensional variational data assimilation(3DVAR)system of the China Meteorological Administration mesoscale model(CMA-MESO).The corresponding forward observation operator,tangent linear operator,and adjoint operator for the T_(2m)observations under three terrain mismatch treatments were developed.The T_(2m)data were assimilated in the same method as that adopted for temperature sounding data with additional representative errors,when station terrain was 100 m higher than model terrain;otherwise,the T_(2m)data were assimilated by using the surface similarity theory assimilation operator.Furthermore,if station terrain was lower than model terrain,additional representative errors were stipulated and corrected.Test of a rainfall case showed that the observation innovation and analysis residuals both exhibited Gaussian distribution and that the analysis increment was reasonable.Moreover,it was found that on completion of the data assimilation cycle,T_(2m)data assimilation obviously influenced the temperature,wind,and relative humidity fields throughout the troposphere,with the greatest impact evident in the lower layers,and that both the area and the intensity of rainfall were better forecasted,especially for the first 12hours.Long-term continuous experiments for 2–28 February and 5–20 July 2020,further verified that T_(2m)data assimilation reduced deviations not only in T_(2m)but also in 10-m wind forecasts.More importantly,the precipitation equitable threat scores were improved over the two experimental periods.In summary,this study confirmed that the T_(2m)data assimilation scheme that we implemented in the kilometer-scale CMA-MESO 3DVAR system is effective.

Assimilation of All-Sky Radiance from the FY-3 MWHS-2 with the Yinhe 4D-Var System

Compared with traditional microwave humidity sounding capabilities at 183 GHz,new channels at 118 GHz have been mounted on the second generation of the Microwave Humidity Sounder(MWHS-2)onboard the Chinese FY-3C and FY-3D polar orbit meteorological satellites,which helps to perform moisture sounding.In this study,as the allsky approach can manage non-linear and non-Gaussian behavior in cloud-and precipitation-affected satellite radiances,the MWHS-2 radiances in all-sky conditions were first assimilated in the Yinhe four-dimensional variational data assimilation(YH4DVAR)system.The data quality from MWHS-2 was evaluated based on observation minus background statistics.It is found that the MWHS-2 data of both FY-3C and FY-3D are of good quality in general.Six months of MWHS-2 radiances in all-sky conditions were then assimilated in the YH4DVAR system.Based on the forecast scores and observation fits,we conclude that the all-sky assimilation of the MWHS-2 at 118-and 183-GHz channels on FY-3C/D is beneficial to the analysis and forecast fields of the temperature and humidity,and the impact on the forecast skill scores is neutral to positive.Additionally,we compared the impacts of assimilating the 118-GHz channels and the equivalent Advanced Microwave Sounding Unit-A(AMSUA)channels on global forecast accuracy in the absence of other satellite observations.Overall,the impact of the 118-GHz channels on the forecast accuracy is not as large as that for the equivalent AMSUA channels.Nevertheless,all-sky radiance assimilation of MWHS-2 in the YH4DVAR system has indeed benefited from the 118-GHz channels.

Assimilation of Radar and Cloud-to-Ground Lightning Data Using WRF-3DVar Combined with the Physical Initialization Method——A Case Study of a Mesoscale Convective System

Radar data, which have incomparably high temporal and spatial resolution, and lightning data, which are great indicators of severe convection, have been used to improve the initial field and increase the accuracies of nowcasting and short-term forecasting. Physical initialization combined with the three-dimensional variational data assimilation method(PI3 DVarrh) is used in this study to assimilate two kinds of observation data simultaneously, in which radar data are dominant and lightning data are introduced as constraint conditions. In this way, the advantages of dual observations are adopted. To verify the effect of assimilating radar and lightning data using the PI3 DVarrh method, a severe convective activity that occurred on 5 June 2009 is utilized, and five assimilation experiments are designed based on the Weather Research and Forecasting(WRF) model. The assimilation of radar and lightning data results in moister conditions below cloud top, where severe convection occurs;thus, wet forecasts are generated in this study.The results show that the control experiment has poor prediction accuracy. Radar data assimilation using the PI3 DVarrh method improves the location prediction of reflectivity and precipitation, especially in the last 3-h prediction, although the reflectivity and precipitation are notably overestimated. The introduction of lightning data effectively thins the radar data, reduces the overestimates in radar data assimilation, and results in better spatial pattern and intensity predictions. The predicted graupel mixing ratio is closer to the distribution of the observed lightning,which can provide more accurate lightning warning information.

不同初始场条件对GRAPES模式数值预报的影响

利用中国新一代数值预报模式GRAPES(Global/Regional Assimilation and Prediction Enhanced System),分别使用T213 L31及NCEP FNL 6 h间隔的资料作为初始和边界条件,对2005年7月9—10日淮河流域一次致洪暴雨过程进行了个例试验,初步探讨了GRAPES模式的数值预报产品对不同初始场的敏感程度,以及三维变分同化对数值预报结果的可能影响。结果表明,T213和NCEP初始场中存在着差异较大的次天气尺度特征,并由此造成了此后GRAPES模式预报场的差异,且此差异不会随时间消失;同化对GRAPES模式积分结果的影响主要表现在最初的24 h内;模式对此次致洪暴雨过程的暴雨区分布、强度均有一定的模拟能力,但模拟的强暴雨区与实况仍存在着较大差异。由此可知,GRAPES模式的数值预报能力对不同的初始场和侧边界条件存在不同程度的依赖性,初始场的差异决定了模拟结果的差异。

中尺度WRF数值模式系统本地化业务试验

利用中尺度WRF数值模式及WRF三维变分同化系统,在对比试验的基础上,选取了适合本地的积云过程、微物理过程和辐射过程的方案组合;选择了NCEP/GFS作为模式的背景场;统计计算了以云南为中心的区域背景误差协方差并替换了三维变分同化系统中原有的背景误差协方差;同时,考虑模式底层高度与地面观测站高度的差异,进行了地面资料地形订正。通过上述试验研究,建立了本地化的中尺度WRF数值预报业务系统,该系统能较好地刻画本地下垫面的动力和热力状况,预报能力有明显改善。

bbGPS/PWV资料三维变分同化改进MM5降水预报连续试验的评估

利用区域地基GPS网反演的高时空密度的大气垂直方向水汽总量,也称为可降水量(PWV),可大大弥补常规探空探测水汽资料的不足。为了全面评估区域GPS网PWV资料同化对业务数值天气预报改进程度的目的,在个例研究分析的基础上,进行了连续38天的GPS/PWV资料三维同化(3D-Var)改进数值业务预报的试验。研究方法是根据长江三角洲地区GPS气象网在2002年梅雨和盛夏季节观测的PWV资料,通过三维变分同化建立中尺度数值预报模式MM5的初始场,逐日作出长江三角洲地区24小时的降水量预报。以6小时累积雨量为对象,与未同化GPS/PWV资料的MM5的相应预报比较,通过多种评分方法,评估了GPS/PWV资料改进MM5降水预报的效果。结果表明GPS/PWV资料同化后的MM5降水预报能力在大部分时间和大部分地区都有所提高,主要是伪击率有较明显的下降,对小范围降水预报的改进更为明显。预报明显改进的区域恰好位于GPS站填补常规探空站间距较大的地区。

GPS-PW资料在川西暴雨中的应用研究

利用WRF-3DVAR系统同化常规探空资料及成都4个GPS测站的可降水资料,对2008年9月23～25日一次川西暴雨进行一系列同化试验。结果表明,选用合理的物理过程、积分步长和初始条件,WRF模式可以较好地模拟此次暴雨过程;同化可以改善暴雨落区和强度的预报,同化GPS可降水资料可有效地调节局地及水汽输送下游地区暴雨预报,对北川地区的强降水中心强度及位置预报较好,同化探空资料对雨带形状描述接近实况;同化GPS可降水资料对初始场的湿度场影响明显,而对其他变量场的影响相对较弱;2种资料的单独及同时同化,都改善了预报场的动力结构(垂直速度),有效地减少了spin-up时间,使得模式在积分初期就能模拟出与实况相近的动力结构;同化试验的温度预报场随时间误差增大,这可能与降水的触发有关。

一个用于气候模式的三维变分资料同化系统——方案设计及检验

介绍了一种适合格点模式的三维变分同化方案,并应用于中国科学院大气物理研究所的9层大气环流格点模式(IAP9L2°×2.5°-AGCM)中。同化方案考虑质量场和风场的动力约束关系,并通过变量变换将模式变量转化为互不相关的控制变量:流函数、非平衡速度势、非平衡位势和比湿;同时采用共轭梯度法求解代价函数避免了直接计算背景误差协方差矩阵逆的困难。通过理想试验、模式连续同化积分试验以及对东亚地区特殊寒潮天气的同化试验,结果表明:该系统能够正确反应多变量之间的相互关系,并且长时间运行稳定,更重要的是同化不仅可以明显改善直接参与同化的物理量,而且经过模式的动力调整也改善了未参与同化的物理量(如海平面气压、近地面气温、降水等),这为大气的持续性信息存储于陆面物理量提供了可能,从而可为IAP9L2°×2.5°-AGCM提供陆—气相协调的初始场。

国家气候中心全球海洋资料四维同化系统在热带太平洋的结果初步分析

介绍了国家气候中心(NCC)“全球海洋资料四维同化系统”(简称NCC-GODAS).该系统包含观测资料预处理系统、插值分析系统和所应用的动力模式.插值分析系统采用四维同化技术方案,在时间上设置一个4周的窗口,将此窗口之内的观测资料以一定的权重插入插值分析系统, 在空间上采用三维变分方案.海洋动力模式为“九五”期间LASG研制的L30T63OGCM 1 0版本.文中考察了该系统从1982年到2003年3月在热带太平洋的部分同化分析结果,并与NCEP的再分析资料和EMC/NCEP的太平洋区域海洋资料同化系统的结果进行了对比分析.结果显示,该系统的同化结果(如SST,SSTA,Nino指数,次表层海温变化等)与NCEP的同期同化结果具有很好的一致性.同时,该系统得出的海洋同化资料气候场与海洋模式气候场相比,有显著的改进.表明该系统具有较好的同化能力,其同化结果可为海-气耦合模式进行季节和跨季节的气候预测业务提供可信的海洋初始场资料,同时还可以为相关研究提供海洋分析场.

The Application of ARGO Data to the Global Ocean Data Assimilation Operational System of NCC

In this paper, we have preliminarily studied the application of ARGO (Arrayfor Real-time Geostrophic Oceanography) data to the Global Ocean Data Assimilation System ofNational Climate Center of China (NCC-GODAS), which mainly contains 4 sub-systems such as datapreprocessing, real-time wind stress calculating, variational analysis and interpolating, and oceandynamic model. For the sake of using ARGO data, the relevant adjustment and improvement have beenmade at the corresponding aspects in the subsystems. Using the observation data from 1981 to 2003including the ARGO data of 2001 to July. 2003, we have performed a series of numerical experimentson this system. Comparing with the corresponding results of NCEP, It is illustrated that using ARGOdata can improve the results of NCC-GODAS in the region of the Middle Pacific, for instance SST,SSTA (SST anomalies), Nino index, sea sub-surface temperature, etc. Furthermore, it is obtained thatNCC-GODAS benefits from ARGO data in the other regions such as Atlantic Ocean, Indian Ocean, andextratropical Pacific Ocean much more than in the tropical Pacific.

UNIFICATION AND APPLICATIONS OF MODERN OCEANIC/ATMOSPHERIC DATA ASSIMILATION ALGORITHMS

The key mathematics and applications of various modern atmospheric/oceanicdata assimilation methods including Optimal Interpolation (OI), 4-dimensional variational approach(4D-Var) and filters were systematically reviewed and classified. Based on the data assimilationphilosophy, i. e. , using model dynamics to extract the observational information, the commoncharacter of the problem, such as the probabilistic nature of the evolution of theatmospheric/oceanic system, noisy and irregularly spaced observations, and the advantages anddisadvantages of these data assimilation algorithms, were discussed. In the filtering framework, allmodern data assimilation algorithms were unified: OI/3D-Var is a stationary filter, 4D-Var is alinear (Kalman) filter and an ensemble of Kalman filters is able to construct a nonlinear filter.The nonlinear filter such as the Ensemble Kalman Filter (EN-KF), Ensemble Adjustment Kalman Filter(EAKF) and Ensemble Transformation Kalman Filter (ETKF) can, to some extent, account for thenon-Gaussian information of the prior distribution from the model. The flow-dependent covarianceestimated by an ensemble filter may be introduced to OI and 4D-Var to improve these traditionalalgorithms. In practice, the performance of algorithms may depend on the specific numerical modeland the choice of algorithm may depend on the specific problem. However, the unification ofalgorithms allows us to establish a unified test system to evaluate these algorithms, which providesmore insights into data assimilation philosophies and helps improve data assimilation techniques.

Statistics of the Z–R Relationship for Strong Convective Weather over the Yangtze–Huaihe River Basin and Its Application to Radar Reflectivity Data Assimilation for a Heavy Rain Event

The relationship between the radar reflectivity factor （Z） and the rainfall rate （R） is recalculated based on radar ob- servations from 10 Doppler radars and hourly rainfall measurements at 6529 automatic weather stations over the Yangtze-Huaihe River basin. The data were collected by the National 973 Project from June to July 2013 for severe convective weather events. The Z-R relationship is combined with an empirical qr-R relationship to obtain a new Z-qr relationship, which is then used to correct the observational operator for radar reflectivity in the three-dimensional variational （3DVar） data assimilation system of the Weather Research and Forecasting （WRF） model to im-prove the analysis and prediction of severe convective weather over the Yangtze--Huaihe River basin. The perform- ance of the corrected reflectivity operator used in the WRF 3DVar data assimilation system is tested with a heavy rain event that occurred over Jiangsu and Anhui provinces and the surrounding regions on 23 June 2013. It is noted that the observations for this event are not included in the calculation of the Z-R relationship. Three experiments are conducted with the WRF model and its 3DVar system, including a control run without the assimilation of reflectivity data and two assimilation experiments with the original and corrected refleetivity operators. The experimental results show that the assimilation of radar reflectivity data has a positive impact on the rainfall forecast within a few hours with either the original or corrected reflectivity operators, but the corrected reflectivity operator achieves a better per-forrnance on the rainfall forecast than the original operator. The corrected reflectivity operator extends the effective time of radar data assimilation for the prediction of strong reflectivity. The physical variables analyzed with the corrected reflectivity operator present more reasonable mesoscale structures than those obtained with the original re-flectivity operator. This suggests that the new statistical Z-R relationship is more suitable for predicting severe con- vective weather over the Yangtze-Huaihe River basin than the Z-R relationships currently in use.

风云三号D星MERSI AOD资料的同化应用研究

为了验证风云三号D星MERSI传感器的气溶胶光学厚度(AOD)数据对地面PM_(2.5)的污染过程预报的效果,本文基于WRF-Chem(Weather Research and Forecasting model coupled with Chemistry)大气化学模式和三维变分同化方法,针对2020-02-10—2020-02-12中国北方地区的一次PM_(2.5)重污染过程,进行了同化和预报试验研究。同化数据来自常规地面站点的PM_(2.5)浓度数据和风云三号D星MERSI传感器的气溶胶光学厚度(AOD)数据。控制试验不同化任何资料,3组同化试验分别为仅同化地面PM_(2.5),仅同化卫星AOD,以及同时同化PM_(2.5)和卫星AOD两种资料。结果表明,3组同化试验都可以有效提高初始场准确率,以地面PM_(2.5)作为检验标准,仅同化PM_(2.5)、仅同化AOD、同时同化两种资料相对于控制试验,初始场的平均偏差分别降低54.9%、21.9%和49.0%,平均相关系数分别提升51.4%、16.0%和34.0%,平均均方根误差分别降低50.6%、17.2%和42.3%。以卫星AOD作为检验标准,3组同化试验相对于控制试验,初始场的平均偏差分别降低37.6%、78.4%和83%,平均均方根误差分别降低31.6%、62.2%和65.2%。同化后的初始场对预报有显著的改进,改进持续时间达24 h,以地面PM_(2.5)作为检验标准,同时同化两种资料的试验对24 h预报的平均偏差减少19.7%,相关系数提升8.8%,均方根误差减少17.2%;以卫星AOD作为检验标准,24 h预报的平均偏差减少40.1%,相关系数提升25.9%,均方根误差降低34.7%。试验结论为,相对于仅同化地面PM_(2.5)资料,同化风云卫星AOD资料可以提升后期预报效果。