Application of 4D-Variational data assimilation to the California Current System

The Incremental Strong constraint 4D-Variational (IS4DVAR) data assimilation system of the Regional Ocean Model System (ROMS) is used to study the controllability of a realistic, high resolution configuration of the California Current System. The configuration and results of assimilating both satellite-derived surface observations along with in situ data are presented. Results show consequent improvements in many characteristics of the model circulation, and some of the strengths of the adjoint method for data assimilation are highlighted. General issues of the sensitivity of the results to the configuration of ROMS-IS4DVAR are also discussed.     

四维变分同化和集合平方根滤波联合反演土壤湿度廓线的研究

提出将集合平方根滤波（EnSRF）估计的预报误差协方差用于四维变分（4DVAR）的同化方案（文中称混合四维变分同化方法,简称混合方法）来反演土壤湿度廓线,该方法由两个同化时段构成: 第一时段为EnSRF,第二时段为4DVAR,此种组合可以充分发挥每一同化方法的优势。通过同化表层土壤湿度观测反演土壤湿度廓线这一理想试验来验证方法的可行性,并与EnSRF和4DVAR的反演结果进行比较,结果表明, 混合方法反演的分析时刻土壤湿度廓线都优于EnSRF和4DVAR的结果。与此同时,为了克服小样本在估算背景场误差协方差矩阵时出现的虚假相关对反演的干扰, 提出在原有协方差矩阵中加入具有高斯指数函数成分来降低其影响;与修正前结果相比,反演的中下层（地下34～100 cm） 土壤湿度的均方根误差从0.036 cm3/cm3降到0.016 cm3/cm3, 降幅为55.6%, 更重要的是大大降低了部分深度处反演土壤湿度的误差, 如地下90 cm处误差从0.085 cm3/cm3降到0.024 cm3/cm3, 降幅达71.8%。     

多普勒激光雷达风场反演方法研究

采用三维变分同化反演(3DVAR)、 四维变分同化反演(4DVAR) 对多普勒激光雷达资料反演风场的方法进行了研究, 利用车载多普勒激光雷达在2008年残奥会测试赛期间外场试验取得的数据, 反演了海面10 m高度处的风场, 并将风场反演结果与浮标资料进行了对比分析, 结果表明: 3DVAR、4DVAR风场反演方法均能实现近海面风的精细化风场反演, 并能反映出风向的变化, 反演风场与浮标数据基本一致, 在风速较大的天气情况, 3DVAR与4DVAR反演风场的一致性要好于风速较小的天气情况; 4DVAR反演方法中以浮标资料作为背景场, 使得其与浮标的符合程度要好于3DVAR方法反演风场; 反演风场的风向与浮标风向具有很好的相关关系, 反演风场的风速与浮标风速具有一定的相关关系, 反演风场的风向、风速与浮标的风向、风速之间平均均方根误差和平均绝对误差表明, 这两序列之间具有一定差别, 在风速较小的天气情况下使用时需要注意。     

GRAPES伴随模式底层数据栈优化

GRAPES伴随模式是其四维变分同化系统的核心组成部分。由于其计算过程复杂,临时数据较多,实现中采用断点存储策略可以有效减少伴随模式的计算时间和存储空间。极限断点存储策略是在单积分步内以全存储策略实现为基础,将其中部分基态以计算代替的一种类断点存储策略。在该策略的支持下,需要一种新的数据管理结构,来保证程序的正确运行。文章提出了在已有栈基础上优化的新数据存储管理方式——嵌套多链栈,这种结构可有效满足使用极限断点存储技术实现GRAPES伴随模式的初态管理需求。试验表明:相比断点存储技术,在总内存增加不超过30%的情况下可使GRAPES的运行效率提高1倍。     

Variational Assimilation of GPS Precipitable Water Vapor and Hourly Rainfall Observations for a Meso-β Scale Heavy Precipitation Event During the 2002 Mei-Yu Season

Recent advances in Global Positioning System （GPS） remote sensing technology allow for a direct estimation of the precipitable water vapor （PWV） from delayed signals transmitted by GPS satellites, which can be assimilated into numerical models with four-dimensional variational （4DVAR） data assimilation. A mesoscale model and its 4DVAR system are used to access the impacts of assimilating GPS-PWV and hourly rainfall observations on the short-range prediction of a heavy rainfall event on 20 June 2002. The heavy precipitation was induced by a sequence of meso-β-scale convective systems （MCS） along the mei-yu front in China. The experiments with GPS-PWV assimilation cluster and also eliminated the erroneous rainfall successfully simulated the evolution of the observed MCS systems found in the experiment without 4DVAR assimilation. Experiments with hourly rainfall assimilation performed similarly both on the prediction of MCS initiation and the elimination of erroneous systems, however the MCS dissipated much sooner than it did in observations. It is found that the assimilation-induced moisture perturbation and mesoscale low-level jet are helpful for the MCS generation and development. It is also discovered that spurious gravity waves may post serious limitations for the current 4DVAR algorithm, which would degrade the assimilation efficiency, especially for rainfall data. Sensitivity experiments with different observations, assimilation windows and observation weightings suggest that assimilating GPS-PWV can be quite effective, even with the assimilation window as short as 1 h. On the other hand, assimilating rainfall observations requires extreme cautions on the selection of observation weightings and the control of spurious gravity waves.     

Coupling Ensemble Kalman Filter with Four-dimensional Variational Data Assimilation

This study examines the performance of coupling the deterministic four-dimensional variational assimilation system (4DVAR) with an ensemble Kalman filter (EnKF) to produce a superior hybrid approach for data assimilation. The coupled assimilation scheme (E4DVAR) benefits from using the state-dependent uncertainty provided by EnKF while taking advantage of 4DVAR in preventing filter divergence: the 4DVAR analysis produces posterior maximum likelihood solutions through minimization of a cost function about which the ensemble perturbations are transformed, and the resulting ensemble analysis can be propagated forward both for the next assimilation cycle and as a basis for ensemble forecasting. The feasibility and effectiveness of this coupled approach are demonstrated in an idealized model with simulated observations. It is found that the E4DVAR is capable of outperforming both 4DVAR and the EnKF under both perfect- and imperfect-model scenarios. The performance of the coupled scheme is also less sensitive to either the ensemble size or the assimilation window length than those for standard EnKF or 4DVAR implementations.     

Variational Assimilation of GPS Precipitable Water Vapor and Hourly Rainfall Observations for a Meso-β Scale Heavy Precipitation Event During the 2002 Mei-Yu Season

Recent advances in Global Positioning System (GPS) remote sensing technology allow for a direct estimation of the precipitable water vapor (PWV) from delayed signals transmitted by GPS satellites, which can be assimilated into numerical models with four-dimensional variational (4DVAR) data assimilation. A mesoscale model and its 4DVAR system are used to access the impacts of assimilating GPS-PWV and hourly rainfall observations on the short-range prediction of a heavy rainfall event on 20 June 2002. The heavy precipitation was induced by a sequence of meso-β-scale convective systems (MCS) along the mei-yu front in China.The experiments with GPS-PWV assimilation successfully simulated the evolution of the observed MCS cluster and also eliminated the erroneous rainfall systems found in the experiment without 4DVAR assimilation. Experiments with hourly rainfall assimilation performed similarly both on the prediction of MCS initiation and the elimination of erroneous systems, however the MCS dissipated much sooner than it did in observations. It is found that the assimilation-induced moisture perturbation and mesoscale low-level jet are helpful for the MCS generation and development. It is also discovered that spurious gravity waves may post serious limitations for the current 4DVAR algorithm, which would degrade the assimilation efficiency, especially for rainfall data. Sensitivity experiments with different observations, assimilation windows and observation weightings suggest that assimilating GPS-PWV can be quite effective, even with the assimilation window as short as 1 h. On the other hand, assimilating rainfall observations requires extreme cautions on the selection of observation weightings and the control of spurious gravity waves.     

降水资料同化在GRAPES-MESO模式中应用试验研究

利用国家气象中心中尺度业务数值预报模式GRAPES-MESO v3.0,以2010年6月1～30日为例,开展地面降水率1DVAR(one-dimensional variational assimilation)同化方案在GRAPES-3DVAR(three-dimensional variational assimilation)同化系统中的应用试验研究(ASSI试验),并以未加降水资料同化的试验为对照试验(CNTL试验),以评估全国1h加密雨量资料在模式中同化应用的效果。结果表明:1)在相对湿度背景误差和降水率观测误差范围内,1DVAR同化方案能够对湿度廓线进行有意义的调整,使分析降水向观测降水靠近;ASSI试验对初始温、压、湿、风场的修正主要为正效果;2)对2010年6月17～21日江南、华南连续性降水过程进行了分析,整体而言ASSI试验对逐日及逐时降水强度的预报普遍强于CNTL试验,与实况更加接近;3)ASSI试验对2010年6月1～30日08时起报的0～24 h模式预报的小雨、中雨、大雨、暴雨、大暴雨各个降水量级TS评分及ETS评分相比CNTL试验均有较明显提高,预报偏差也更接近于1;4)ASSI试验较CNTL试验能更好地模拟雨带的分布、雨带演变特征和降水强度的变化;5)对降水所做的典型个例和统计检验分析从不同角度说明了地面降水资料1DVAR同化方案在GRAPES-3DVAR系统中的应用改善了GRAPES-MESO v3.0的降水模拟效果。     

Experiments of the WRF three-/four-dimensional variational (3/4DVAR) data assimilation in the forecasting of Antarctic cyclones

The three-/four-dimensional variational data assimilation systems (3/4DVAR) of the Weather Research and Forecasting (WRF) model were explored in the forecasting of two Antarctic synoptic cyclones, which had large influence on the Ross Sea/Ross Ice Shelf region in October 2007. A suite of variational data assimilation experiments, including regular 3DVAR, high-resolution 3DVAR, and 4DVAR experiments, were designed to evaluate their performances in weather analysis and forecasting in Antarctica. In general, both 4DVAR and high-resolution 3DVAR experiments showed better forecasting skill than regular 3DVAR experiments. High-resolution 3DVAR experiments were the most efficient in reducing the analysis errors of surface winds and temperature, and had the best performance during the first 24 h of forecasting. However, during the following forecast period, 4DVAR experiments showed either better or about comparable performance to high-resolution 3DVAR experiments. These results indicate that increasing the spatial resolution during 3DVAR is an economical approach to improving the weather analysis and forecasting over Antarctica. At the same time, the 4DVAR approach had a longer impact on forecasting than the high-resolution 3DVAR approach. Understandably, both of the variational assimilation approaches are promising techniques toward improving the regional analysis and forecasting over Antarctica.     

Assimilation of Ocean Surface Wind Data by the HY-2B Satellite in GRAPES: Impacts on Analyses and Forecasts

The ocean surface wind(OSW) data retrieved from microwave scatterometers have high spatial accuracy and represent the only wind data assimilated by global numerical models on the ocean surface, thus playing an important role in improving the forecast skills of global medium-range weather prediction models. To improve the forecast skills of the Global/Regional Assimilation and Prediction System Global Forecast System(GRAPES＿GFS), the HY-2B OSW data is assimilated into the GRAPES＿GFS four-dimensio...     

Four-dimensional variational data assimilation for mesoscale and storm-scale applications

Summary ?The status and progress of the four-dimensional variational data assimilation (4DVAR) are briefly reviewed focusing on application to prediction of mesoscale/storm-scale atmospheric phenomena. Theoretical background is provided for each important component of the 4DVAR system – forecast and adjoint models, observations, background, cost function, preconditioning, and minimization. An overview of practical issues specific for mesoscale/storm-scale 4DVAR is then presented in terms of high-resolution observations, nonlinearity and discontinuity problem, model error, errors from lateral boundary condition, and precipitation assimilation. Practical strategies for efficient and simplified 4DVAR are also introduced, e.g., incremental 4DVAR, poor man’s 4DVAR, and inverse 3DVAR. A new concept on hybrid approach is proposed to combine an efficient 4DVAR scheme and the standard 4DVAR scheme aiming at reducing computational demand required by the standard 4DVAR while improving the accuracy of the simplified 4DVAR. Applications to both hydrostatic and nonhydrostatic models are illustrated and our vision on opportunities and directions for future research is provided. Received March 12, 2001; revised July 24, 2001; accepted September 5, 2001     

GRAPES混合同化方法在青藏高原区域的初步试验

针对青藏高原地区气象观测站点稀少和模式同化分析质量较低的问题,将GRAPES区域集合变分(En-3DVAR)混合同化系统应用于青藏高原地区,进行了单点理想试验和真实观测资料同化分析预报试验,分析评估青藏高原混合同化分析增量及预报误差的水平垂直结构特征及其合理性,并与中国东部平原地区进行比对。单点理想试验表明,En-3DVAR混合同化系统中背景误差协方差具备流依赖属性。真实资料混合同化试验结果表明,基于集合预报估计的分析增量具有流依赖的特征,特别是在高原地区和槽脊系统附近;青藏高原地区分析场的绝对误差总体低于3DVAR系统,其平均绝对误差在中层和高层低于平原地区,说明在青藏高原地区的改进效果略优于平原地区。需要关注的是,青藏高原地区En-3DVAR混合同化分析增量总体大于3DVAR,特别是近地面层u风分量分析增量明显偏大,这可能与青藏高原复杂地形有关。     

雷达资料同化对一次飑线过程的模拟影响

雷达资料同化能够改善强对流天气的预报,但是不同的模式方案配置会得到不同的结果。本文针对中国南部2018年3月4日一次飑线过程,以全球预报模式GFS分析场和预报场为背景场,采用中尺度区域气象预报模式ARPS 3DVAR系统同化多普勒雷达径向速度,用云分析处理反射率数据,考虑同化间隔、频次、云分析中不同参数调整,采用1 h同化窗口,设计不同同化方案,最后用WRF模式进行预报,研究雷达资料同化对飑线系统触发及发展机制的影响。结果表明,同化间隔过短时,由于模式热动力变量没有平衡产生虚假回波,同化间隔过长时,系统触发和发展的特征普遍偏弱;采用12 min间隔同化得到了最好的初始场,并且同化频次越高得到的降水预报结果越好。此外,ARPS云分析能大大改善初始场,减少模式自调整时间,其中湿度调整、温度调整、雨水调整及水汽调整对系统动力过程和水凝物初始场分布都有较大的影响,而垂直速度相关参数调整影响较小。     

中尺度模式遗传算法同化系统及同化试验研究

设计了一个以MM5模式为基础的遗传算法同化系统,并对一次暴雨过程进行了实际降水的模拟,通过对比遗传同化和伴随同化的降水预报效果,对遗传算法同化系统的同化性能进行验证。试验结果表明,遗传算法与四维变分相结合的同化系统能有效地改善模式的初始场,使MM5模式要素预报和降水预报的准确率得到提高。     

Assimilating surface observations in a four-dimensional variational Doppler radar data assimilation system to improve the analysis and forecast of a squall line case

This paper examines how assimilating surface observations can improve the analysis and forecast ability of a fourdimensional Variational Doppler Radar Analysis System(VDRAS).Observed surface temperature and winds are assimilated together with radar radial velocity and reflectivity into a convection-permitting model using the VDRAS four-dimensional variational(4DVAR) data assimilation system.A squall-line case observed during a field campaign is selected to investigate the performance of the technique.A single observation experiment shows that assimilating surface observations can influence the analyzed fields in both the horizontal and vertical directions.The surface-based cold pool,divergence and gust front of the squall line are all strengthened through the assimilation of the single surface observation.Three experiments—assimilating radar data only,assimilating radar data with surface data blended in a mesoscale background,and assimilating both radar and surface observations with a 4DVAR cost function—are conducted to examine the impact of the surface data assimilation.Independent surface and wind profiler observations are used for verification.The result shows that the analysis and forecast are improved when surface observations are assimilated in addition to radar observations.It is also shown that the additional surface data can help improve the analysis and forecast at low levels.Surface and low-level features of the squall line—including the surface warm inflow,cold pool,gust front,and low-level wind—are much closer to the observations after assimilating the surface data in VDRAS.     

Variational data assimilation experiments of mei-yu front rainstorms in China

The numerical forecasts of mei-yu front rainstorms in China has been an important issue. The intensity and pattern of the frontal rainfall are greatly influenced by the initial fields of the numerical model. The 4-dimensional variational data assimilation technology (4DVAR) can effectively assimilate all kinds of observed data, including rainfall data at the observed stations, so that the initial fields and the precipitation forecast can both be greatly improved. The non-hydrostatic meso-scale model (MM5) and its adjoint model are used to study the development of the mei-yu front rainstorm from 1200 UTC 25 June to 0600 UTC 26 June 1999. By numerical simulation experiments and assimilation experiments, the T106 data and the observed 6-hour rainfall data are assimilated. The influences of many factors, such as the choice of the assimilated variables and the weighting coefficient, on the precipitation forecast results are studied. The numerical results show that 4DVAR is valuable and important to mei-yu front rainfall prediction.     

ANALYSIS OF THE EFFECT OF 3DVAR AND ENSRF DIRECT ASSIMILATION OF RADAR DATA ON THE FORECAST OF A HEAVY RAINFALL EVENT

The present study designs experiments on the direct assimilation of radial velocity and reflectivity data collected by an S-band Doppler weather radar (CINRAD WSR-98D) at the Hefei Station and the reanalysis data produced by the United States National Centers for Environmental Prediction using the Weather Research and Forecasting (WRF) model, the WRF model with a three-dimensional variational (3DVAR) data assimilation system and the WRF model with an ensemble square root filter (EnSRF) data assimilation system. In addition, the present study analyzes a Meiyu front heavy rainfall process that occurred in the Yangtze -Huaihe River Basin from July 4 to July 5, 2003, through numerical simulation. The results show the following. (1) The assimilation of the radar radial velocity data can increase the perturbations in the low-altitude atmosphere over the heavy rainfall region, enhance the convective activities and reduce excessive simulated precipitation. (2) The 3DVAR assimilation method significantly adjusts the horizontal wind field. The assimilation of the reflectivity data improves the microphysical quantities and dynamic fields in the model. In addition, the assimilation of the radial velocity and reflectivity data can better adjust the wind fields and improve the intensity and location of the simulated radar echo bands. (3) The EnSRF assimilation method can assimilate more small-scale wind field information into the model. The assimilation of the reflectivity data alone can relatively accurately forecast the rainfall centers. In addition, the assimilation of the radial velocity and reflectivity data can improve the location of the simulated radar echo bands. (4) The use of the 3DVAR and EnSRF assimilation methods to assimilate the radar radial velocity and reflectivity data can improve the forecast of precipitation, rain-band areal coverage and the center location and intensity of precipitation.     

Simulation of heavy rainfall events over Indian monsoon region using WRF-3DVAR data assimilation system

We present the results of the impact of the 3D variational data assimilation (3DVAR) system within the Weather Research and Forecasting (WRF) model to simulate three heavy rainfall events (25–28 June 2005, 29–31 July 2004, and 7–9 August 2002) over the Indian monsoon region. For each event, two numerical experiments were performed. In the first experiment, namely the control simulation (CNTL), the low-resolution global analyses are used as the initial and boundary conditions of the model. In the second experiment (3DV-ANA), the model integration was carried out by inserting additional observations in the model’s initial conditions using the 3DVAR scheme. The 3DVAR used surface weather stations, buoy, ship, radiosonde/rawinsonde, and satellite (oceanic surface wind, cloud motion wind, and cloud top temperature) observations obtained from the India Meteorological Department (IMD). After the successful inclusion of additional observational data using the 3DVAR data assimilation technique, the resulting reanalysis was able to successfully reproduce the structure of convective organization as well as prominent synoptic features associated with the mid-tropospheric cyclones (MTC). The location and intensity of the MTC were better simulated in the 3DV-ANA as compared to the CNTL. The results demonstrate that the improved initial conditions of the mesoscale model using 3DVAR enhanced the location and amount of rainfall over the Indian monsoon region. Model verification and statistical skill were assessed with the help of available upper-air sounding data. The objective verification further highlighted the efficiency of the data assimilation system. The improvements in the 3DVAR run are uniformly better as compared to the CNTL run for all the three cases. The mesoscale 3DVAR data assimilation system is not operational in the weather forecasting centers in India and a significant finding in this study is that the assimilation of Indian conventional and non-conventional observation datasets into numerical weather forecast models can help improve the simulation accuracy of meso-convective activities over the Indian monsoon region. Results from the control experiments also highlight that weather and regional climate model simulations with coarse analysis have high uncertainty in simulating heavy rain events over the Indian monsoon region and assimilation approaches, such as the 3DVAR can help reduce this uncertainty.     

TOPOGRAPHIC DEPENDENT HORIZONTAL LOCALIZATION SCALE SCHEME IN GRAPES-MESO HYBRID EN-3DVAR ASSIMILATION SYSTEM

Based on the GRAPES-MESO hybrid En-3DVAR (Ensemble three-dimension hybrid data assimilation for Global/Regional Assimilation and Prediction system) constructed by China Meteorological Administration, a 7-day simulation (from 10 July 2015 to 16 July 2015) is conducted for horizontal localization scales. 48h forecasts have been designed for each test, and seven different horizontal localization scales of 250, 500, 750, 1000, 1250, 1500 and 1750 km are set. The 7-day simulation results show that the optimal horizontal localization scales over the Tibetan Plateau and the plain area are 1500 km and 1000 km, respectively. As a result, based on the GRAPES-MESO hybrid En-3DVAR, a topography-dependent horizontal localization scale scheme (hereinafter referred to as GRAPES-MESO hybrid En-3DVAR-TD-HLS) has been constructed. The data assimilation and forecast experiments have been implemented by GRAPES-MESO hybrid En-3DVAR, 3DVAR and GRAPES-MESO hybrid En-3DVAR-TD-HLS, and then the analysis and forecast field of these three systems are compared. The results show that the analysis field and forecast field within 30h of GRAPES-MESO hybrid En-3DVAR-TD-HLS are better than those of the other two data assimilation systems. Particularly in the analysis field, the root mean square error (RMSE) of u_wind and v_wind in the entire vertical levels is significantly less than that of the other two systems. The time series of total RMSE indicate, in the 6-30h forecast range, that the forecast result of En-3DVAR-TD-HLS is better than that of the other two systems, but the En-3DVAR and 3DVAR are equivalent in terms of their forecast skills. The 36-48h forecasts of three data assimilation systems have similar forecast skill.     

Hybrid ETKF-3DVAR方法同化多普勒雷达速度观测资料Ⅰ:模拟资料试验

利用WRF(Weather research and forecasting)模式及模式模拟的资料,采用Hybrid ETKF-3DVAR(ensemble transform Kalman filter-three-dimensional variational data assimilation)方法同化模拟雷达观测资料。该混合同化方法将集合转换卡尔曼滤波(ensemble transform Kalman filter)得到的集合样本扰动通过转换矩阵直接作用到背景场上,利用顺序滤波的思想得到分析扰动场;然后通过增加额外控制变量的方式把"流依赖"的集合协方差信息引入到变分目标函数中去,在3DVAR框架基础下与观测数据进行融合,从而给出分析场的最优估计。试验结果表明,Hybrid ETKF-3DVAR同化方法相比传统3DVAR可以提供更为准确的分析场,Hybrid方法雷达资料初始化模拟的台风涡旋结构与位置比3DVAR更加接近"真实场",对台风路径预报也有明显改进。通过对比Hybrid S试验与Hybrid F试验发现,Hybrid的正效果主要来源于混合背景误差协方差中的"流依赖"信息,集合平均场代替确定性背景场带来的效果并不显著。