Ensemble Based Diagnosis of the Track Errors of Super Typhoon Mangkhut(2018)

Super Typhoon Mangkhut(2018)was the most high-impact typhoon in 2018 because of its long lifespan and significant intensity.The operational track forecasts in the short-to-medium range(deterministic and probabilistic forecast)showed a great uncertainty and the forecast landing points varied with different lead times.This study applied ensembles of high-resolution ECMWF forecasts to investigate the major factors and mechanisms of the bias production of the Mangkhut forecast track.The ensembles with the largest track bias were analyzed to examine the possible bias associated factors.The results suggested that environmental steering flows were the main cause for the erroneous southward track error with a variance contribution of 72%.The tropical cyclone(TC)size difference and the interaction of the TC with the subtropical high(SH)were other two key factors that contributed to the track error.Particularly,larger TCs may have led to a stronger erosion of the southern part of the SH,and thus induced significant changes in the large-scale environment and eventually resulted in an additional northward movement of TC.     

垂直分层加密和预报区域扩大对GRAPES_TYM台风预报的影响

为提升GRAPES_TYM对西北太平洋和中国南海热带气旋路径及强度的预报能力、增加对北印度洋热带气旋的预报,2019年8月GRAPES_TYM 3.0版投入业务运行。GRAPES_TYM 3.0版的模式垂直分层由GRAPES_TYM 2.2版的50层增加到68层;预报区域由覆盖西北太平洋、中国南海扩展到覆盖北印度洋。试验结果显示:模式垂直分层增加可以改进模式对强台风及超强台风的预报能力,减小平均路径预报误差、显著减小平均强度预报误差以及强度预报负偏差;模式预报区域扩大到覆盖北印度洋对平均路径误差和平均强度误差影响不显著,但长时效预报比较敏感,如20°N以北热带气旋120 h预报路径。2016—2018年的回算结果与NCEP-GFS和ECMWF的预报结果对比分析表明:GRAPES_TYM 3.0版的平均路径误差与NCEP-GFS接近,同ECMWF相比误差较大;但24—96 h强度预报误差明显小于NCEP-GFS和ECMWF,NCEP-GFS和ECMWF对热带气旋强度预报存在明显的负偏差。综上所述,模式垂直分层由50层增加到68层对热带气旋强度预报至关重要,而长时效路径预报对模式预报区域扩大到覆盖北印度洋更为敏感。      

热带气旋的路径及登陆预报

用几个非线性数学模型制作热带气旋短期路径预报及热带气旋个数、登陆时段、地段的短期气候预报。5年多的研究和预报试验结果表明：用指数曲线模型制作热带气旋路径预报，准确率较高。24h预报，199次平均误差123km，达到国内先进水平。用多项式等非线性模型，制作登陆我国及登陆广东热带气旋的年、月个数预测，经过3年实际应用检验，准确率达到70％～90％。用非线性预测模型的逐日气压场、逐日雨量场长期预测结果进行分析，制作广东热带气旋登陆时段、地段和南海海面热带气旋出现时间的预报，准确率达到70％～80％，2002年热带气旋的预报，采用长中短期预报相结合，数值预报与统计预报相结合，预报效果较佳。     

Impacts of a Large-scale Adaptive Blending Scheme for CMA-MESO on Regional Forecasts-A Case Study of Typhoon Haima

Large-scale atmospheric information plays an important role in the regional model for the forecasts of weather such as tropical cyclone (TC). However, it is difficult to be fully represented in regional models due to domain size and a lack of observation data, particularly at sea used in regional data assimilation. Blending analysis has been developed and implemented in regional models to reintroduce large-scale information from global model to regional analysis. Research of the impact of this large-scale blending scheme for the Global / Regional Assimilation and PrEdiction System (CMA-MESO) regional model on TC forecasting is limited and this study attempts to further progress by examining the adaptivity of the blending scheme using the two-dimensional Discrete Cosine Transform (2D-DCT) filter on the model forecast of Typhoon Haima over Shenzhen, China in 2016 and considering various cut-off wavelengths. Results showed that the error of the 24-hour typhoon track forecast can be reduced to less than 25 km by applying the scale-dependent blending scheme, indicating that the blending analysis is effectively able to minimise the large-scale bias for the initial fields. The improvement of the wind forecast is more evident for u-wind component according to the reduced root mean square errors (RMSEs) by comparing the experiments with and without blending analysis. Furthermore, the higher equitable threat score (ETS) provided implications that the precipitation prediction skills were increased in the 24h forecast by improving the representation of the large-scale feature in the CMA-MESO analysis. Furthermore, significant differences of the track error forecast were found by applying the blending analysis with different cut-off wavelengths from 400 km to 1200 km and the track error can be reduced less than by 10 km with 400 km cut-off wavelength in the first 6h forecast. It highlighted that the blending scheme with dynamic cut-off wavelengths adapted to the development of different TC systems is necessary in order to optimally introduce and ingest the large-scale information from global model to the regional model for improving the TC forecast. In this paper, the methods and data applied in this study will be firstly introduced, before discussion of the results regarding the performance of the blending analysis and its impacts on the wind and precipitation forecast correspondingly, followed by the discussion of the effects of different blending scheme on TC forecasts and the conclusion section.     

STUDY OF THE MODIFICATION OF MULTI-MODEL ENSEMBLE SCHEMES FOR TROPICAL CYCLONE FORECASTS

This study investigates multi-model ensemble forecasts of track and intensity of tropical cyclones over the western Pacific, based on forecast outputs from the China Meteorological Administration, European Centre for Medium-Range Weather Forecasts, Japan Meteorological Agency and National Centers for Environmental Prediction in the THORPEX Interactive Grand Global Ensemble (TIGGE) datasets. The multi-model ensemble schemes, namely the bias-removed ensemble mean (BREM) and superensemble (SUP), are compared with the ensemble mean (EMN) and single-model forecasts. Moreover, a new model bias estimation scheme is investigated and applied to the BREM and SUP schemes. The results showed that, compared with single-model forecasts and EMN, the multi-model ensembles of the BREM and SUP schemes can have smaller errors in most cases. However, there were also circumstances where BREM was less skillful than EMN, indicating that using a time-averaged error as model bias is not optimal. A new model bias estimation scheme of the biweight mean is introduced. Through minimizing the negative influence of singular errors, this scheme can obtain a more accurate model bias estimation and improve the BREM forecast skill. The application of the biweight mean in the bias calculation of SUP also resulted in improved skill. The results indicate that the modification of multi-model ensemble schemes through this bias estimation method is feasible.     

A study on the impact of asymmetric components around tropical cyclone center on the accuracy of bogus data and the track forecast

Summary The Japan Meteorological Agency (JMA) has used a tropical cyclone bogus insertion procedure to produce correctlypositioned, cyclone-like vortices within the initial analyses and to track the vortices throughout the model forecasts. The TC bogus soundings are constructed from a standard axisymmetric vortices for well developed tropical cyclones based on a few manually-analyzed parameters such as storm position, central pressure and radius of gale force wind. Mainly because of such an axi-symmetric property of JMA TC bogus data, which is likely to remove the steering flow from the central core region of TC, all the JMA models have a noticeable slow-start bias error and also northward drifting blas error in TC movement. In order to investigate the impact of asymmetric wind components on the TC track forecast, an experimental analysis-forecast cycle is conducted using the JMA global spectral model, in which asymmetric components extracted from the first guess fields are added to the axisymmetric TC bogus. It is found from the experiment that both the slow-start bias error and northward bias error can be reduced by introducing the asymmetric components into the TC bogus. Besides the impact study, a statistical verification study of the bogus data was also made against real data such as sonde data and superiority of the preparation method of asymmetric components was proved.With 9 Figures     

台风路径数值预报实时订正技术及其集成应用

以台风路径数值预报的短时效预报偏差和目标时效（指所需订正的时效）的纬度预报为预报因子,采用多元线性回归方法建立了台风路径预报的偏差预估方程,继而对台风路径预报进行实时订正。本文以12 h为短时效,通过对欧洲中期天气预报中心确定性预报模式（ECMWF-IFS）和集合预报模式（ECMWF-EPS）的台风路径预报的应用,得到以下结论:2018年试报结果表明,24 h、36 h、48 h、60 h、72 h、84 h订正后的ECMWF-IFS台风路径预报的平均距离误差分别比订正前减小了7.3 km、9.3 km、8.9 km、6.5 km、6.9 km、2.6 km,总体来说较强台风（指12 h的台风强度实况≥32.7 m s?1）路径预报的订正效果更好。尝试了先对ECMWF-EPS各成员的台风路径预报进行订正,再进行集成预报,并对比了以下5种方式得到的台风路径预报:“订正后的确定性预报”、“所有集合预报成员集合平均”、“优选集合预报成员集合平均”、“所有集合预报成员先订正再集合平均”和“优选集合预报成员先订正再集合平均”,2018年试报结果表明,对于平均距离误差,24 h和36 h“优选集合预报成员先订正再集合平均”最小,48 h和60 h“所有集合预报成员先订正再集合平均”最小,72 h和84 h“优选集合预报成员集合平均”最小,如果在业务中有针对性地进行应用,有望获得一个在各预报时效表现都较优异的台风路径客观综合预报结果。24 h、36 h、48 h、60 h“优选集合预报成员先订正再集合平均”的平均距离误差分别比“所有集合预报成员集合平均”减小了13.3 km、11.7 km、10.0 km、7.6 km,比中央气象台官方预报（对应的时效为12 h、24 h、36 h、48 h）减小了0.7 km、2.0 km、3.9 km、2.4 km。     

Assimilation of GPM Microwave Imager Radiance for Track Prediction of Typhoon Cases with the WRF Hybrid En3DVAR System

The impact of assimilating radiance data from the advanced satellite sensor GMI(GPM microwave imager) for typhoon analyses and forecasts was investigated using both a three-dimensional variational(3DVAR) and a hybrid ensemble-3DVAR method. The interface of assimilating the radiance for the sensor GMI was established in the Weather Research and Forecasting(WRF) model. The GMI radiance data are assimilated for Typhoon Matmo(2014), Typhoon Chan-hom(2015), Typhoon Meranti(2016), and Typhoon Mangkhut(2018) in the Pacific before their landing. The results show that after assimilating the GMI radiance data under clear sky condition with the 3DVAR method, the wind,temperature, and humidity fields are effectively adjusted, leading to improved forecast skills of the typhoon track with GMI radiance assimilation. The hybrid DA method is able to further adjust the location of the typhoon systematically. The improvement of the track forecast is even more obvious for later forecast periods. In addition, water vapor and hydrometeors are enhanced to some extent, especially with the hybrid method.     

全球数值模式中的台风初始化Ⅱ: 业务应用

由于缺少大量有效的观测资料,台风初始化对数值天气预报业务模式而言,仍然是一个悬而末决的难题.中国国家气象中心自从1996年将台风数值预报系统投入业务运行以来,一直使用经验的人造bogus涡旋台风初始化技术.实际上,不同时期的台风有着不同的环流结构,即使同一个台风在不同的生命期也具有不同的结构特征,而这些结构特征的差异并不能依靠现有的bogus涡旋技术体现出来,这种主观方法的统一性与台风在时空上的差异性形成了强烈的反差.最近,基于国家气象中心全球资料分析同化-预报循环系统,设计和发展了一套新的台风初始化业务方案,它主要由初始涡旋形成、涡旋重定位和涡旋调整3部分过程组成.相比于业务中使用的人造bogus涡旋台风初始化方案,新方案在很大程度上减少了人为因素对台风涡旋结构的影响,而更多地是依靠数值模式自身的动力和物理过程来协调约束产生三维空间的涡旋结构.应用新方案,文中对生成于西北太平洋的2006年0605号台风格美(Kaemi)进行了数值试验,初步分析表明,新方案在实现台风涡旋环流结构的初始化方面效果较好,同时,对台风格美多个时次的预报结果也显示,相比于业务使用的bogus方案而言,新方案对台风路径平均预报误差有了大幅度的降低.     

OPERATIONAL ENSEMBLE FORECASTING AND ANALYSIS OF TROPICAL CYCLONES OVER THE WESTERN NORTH PACIFIC (INCLUDING THE SOUTH CHINA SEA)

Based on the tropical cyclone data from the Central Meteorological Observatory of China, Japan Meteorological Agency, Joint Typhoon Warning Center and European Centre for Medium-Range Weather Forecasts （ECMWF） during the period of 2004 to 2009, three consensus methods are used in tropical cyclone （TC） track forecasts. Operational consensus results show that the objective forecasts of ECMWF help to improve consensus skill by 2%, 3%-5% and 3%-5%, decrease track bias by 2.5 kin, 6-9 km and 10-12 km for the 24 h, 48 h and 72 h forecasts respectively over the years of 2007 to 2009. Analysis also indicates that consensus forecasts hold positive skills relative to each member. The multivariate regression composite is a method that shows relatively low skill, while the methods of arithmetic averaging and composite （in which the weighting coefficient is the reciprocal square of mean error of members） have almost comparable skills among members. Consensus forecast for a lead time of 96 h has negative skill relative to the ECMWF objective forecast.     

Improving Multi-Model Ensemble Forecasts of Tropical Cyclone Intensity Using Bayesian Model Averaging

This paper proposes a method for multi-model ensemble forecasting based on Bayesian model averaging (BMA), aiming to improve the accuracy of tropical cyclone (TC) intensity forecasts, especially forecasts of minimum surface pressure at the cyclone center (P_min). The multi-model ensemble comprises three operational forecast models: the Global Forecast System (GFS) of NCEP, the Hurricane Weather Research and Forecasting (HWRF) models of NCEP, and the Integrated Forecasting System (IFS) of ECMWF. The mean of a predictive distribution is taken as the BMA forecast. In this investigation, bias correction of the minimum surface pressure was applied at each forecast lead time, and the distribution (or probability density function, PDF) of P_min was used and transformed. Based on summer season forecasts for three years, we found that the intensity errors in TC forecast from the three models varied significantly. The HWRF had a much smaller intensity error for short lead-time forecasts. To demonstrate the proposed methodology, cross validation was implemented to ensure more efficient use of the sample data and more reliable testing. Comparative analysis shows that BMA for this three-model ensemble, after bias correction and distribution transformation, provided more accurate forecasts than did the best of the ensemble members (HWRF), with a 5%–7% decrease in root-mean-square error on average. BMA also outperformed the multi-model ensemble, and it produced “predictive variance” that represented the forecast uncertainty of the member models. In a word, the BMA method used in the multi-model ensemble forecasting was successful in TC intensity forecasts, and it has the potential to be applied to routine operational forecasting.     

Impact of Coastal Radar Observability on the Forecast of the Track and Rainfall of Typhoon Morakot(2009)Using WRF-based Ensemble Kalman Filter Data Assimilation

This study explored the impact of coastal radar observability on the forecast of the track and rainfall of Typhoon Morakot(2009)using a WRF-based ensemble Kalman filter(EnKF)data assimilation(DA)system.The results showed that the performance of radar EnKF DA was quite sensitive to the number of radars being assimilated and the DA timing relative to the landfall of the tropical cyclone(TC).It was found that assimilating radial velocity(Vr)data from all the four operational radars during the 6 h immediately before TC landfall was quite important for the track and rainfall forecasts after the TC made landfall.The TC track forecast error could be decreased by about 43% and the 24-h rainfall forecast skill could be almost tripled.Assimilating Vr data from a single radar outperformed the experiment without DA, though with less improvement compared to the multiple-radar DA experiment.Different forecast performances were obtained by assimilating different radars, which was closely related to the first-time wind analysis increment, the location of moisture transport, the quasi-stationary rainband, and the local convergence line.However, only assimilating Vr data when the TC was farther away from making landfall might worsen TC track and rainfall forecasts.Besides, this work also demonstrated that Vr data from multiple radars, instead of a single radar, should be used for verification to obtain a more reliable assessment of the EnKF performance.     

登陆台风强度变化过程预报的影响试验研究

探索了基于WRF模式的集合卡尔曼滤波同化方法（WRF-EnKF,简称EnKF）在近海有可能达到更强台风连续循环同化中国大陆高时空分辨率多普勒天气雷达径向风观测资料的效果,同时检验台风Vicente（2012）的三维结构演变及其动力学特征。通过短期集合预报得到跟随当前流场变化着的背景误差协方差的台风涡旋和动力学结构。研究发现,EnKF同化预报系统能有效地同化高时空分辨率雷达径向速度观测资料,显著改善初始场中台风Vicente的中小尺度内核结构,同时提高对台风Vicente的路径和强度及其相伴随的短期强降水预报。在台风最强时刻同化雷达径向风观测能快速（1~2 h）得到真实的暖核台风结构,同时进一步提高台风路径和强度的预报。另外,EnKF同化雷达径向风观测资料还能有效提高短期降水预报,1 h和3 h累积降水的分布、降水中心以及降水随时间演变都能得到显著改善,这与改善台风路径、结构和强度有密切关系。因此,对中国东南沿海有可能达到较强的台风进行同化雷达径向风观测资料可改善登陆台风的预报水平,这为利用我国地基多普勒天气雷达观测资料改善模式的初始场从而提高台风预报提供一定的指示作用。      

热带气旋奇异向量在GRAPES全球集合预报中的初步应用

基于副热带奇异向量的初值扰动方法已应用于GRAPES （Global and Regional Assimilation PrEdiction System）全球集合预报系统,但存在热带气旋预报路径离散度不足的问题。通过分析发现,热带气旋附近区域初值扰动结构不合理导致预报集合不能较好地估计热带气旋预报的不确定性,是路径集合离散度不足的可能原因之一。通过建立热带气旋奇异向量求解方案,将热带气旋奇异向量和副热带奇异向量共同线性组合生成初值扰动,以弥补热带气旋区域初值扰动结构不合理这一缺陷,进而改进热带气旋集合预报效果。利用GRAPES全球奇异向量计算方案,以台风中心10个经纬度区域为目标区构建热带气旋奇异向量求解方案,针对台风“榕树”个例进行集合预报试验,并开展批量试验,利用中国中央气象台最优台风路径和中国国家气象信息中心的降水观测资料进行检验,对比分析热带气旋奇异向量结构特征和初值扰动特征,评估热带气旋奇异向量对热带气旋路径集合预报和中国区域24 h累计降水概率预报技巧的影响。结果表明,热带气旋奇异向量具有局地化特征,使用热带气旋奇异向量之后,热带气旋路径离散度增加,路径集合平均预报误差和离散度的关系得到改善,路径集合平均预报误差有所减小,集合成员更好地描述了热带气旋路径的预报不确定性;中国台风降水的小雨、中雨、大雨、暴雨各量级24 h累计降水概率预报技巧均有一定提高。总之,当在初值扰动的生成中考虑热带气旋奇异向量后,可改进热带气旋初值扰动结果,并有助于改善热带气旋路径集合预报效果。      

ASSIMILATION OF REAL OBSERVATIONAL DATA WITH THE GSI-HYBRID DATA ASSIMILATION SYSTEM TO IMPROVE TYPHOON FORECAST

A hybrid GSI (Grid-point Statistical Interpolation)-ETKF (Ensemble Transform Kalman Filter) data assimilation system has been recently developed for the WRF (Weather Research and Forecasting) model and tested with simulated observations for tropical cyclone (TC) forecast. This system is based on the existing GSI but with ensemble background information incorporated. As a follow-up, this work extends the new system to assimilate real observations to further understand the hybrid scheme. As a first effort to explore the system with real observations, relatively coarse grid resolution (27 km) is used. A case study of typhoon Muifa (2011) is performed to assimilate real observations including conventional in-situ and satellite data. The hybrid system with flow-dependent ensemble covariance shows significant improvements with respect to track forecast compared to the standard GSI system which in theory is three dimensional variational analysis (3DVAR). By comparing the analyses, analysis increments and forecasts, the hybrid system is found to be potentially able to recognize the existence of TC vortex, adjust its position systematically, better describe the asymmetric structure of typhoon Muifa and maintain the dynamic and thermodynamic balance in typhoon initial field. In addition, a cold-start hybrid approach by using the global ensembles to provide flow-dependent error is tested and similar results are revealed with those from cycled GSI-ETKF approach.     

B08RDP区域集合预报温度场质量评估与综合偏差订正

针对B08RDP(The Beijing 2008 Olympics Research and Development Project)5套区域集合预报资料,系统分析了各套集合预报温度场的预报质量。在此基础上运用集合预报的综合偏差订正方法对温度场进行偏差订正,并对其效果进行了分析讨论。结果显示:5套B08RDP区域集合预报中,美国国家环境预报中心(NCEP)区域集合预报温度场的整体预报质量最高,平均预报误差最小,离散度也最为合理,预报可信度和可辨识度均较优;而中国气象科学研究院(CAMS)的温度预报误差过大,预报质量最差。整体上看,除NCEP之外的4套集合预报的温度场均存在集合离散度偏小的问题;综合偏差订正能有效减小各集合预报温度场的集合平均均方根误差,改善集合离散度的质量,显示出综合偏差订正方案对集合预报温度场偏差订正的良好能力。     

1999—2003年西北太平洋热带气旋综合预报的误差分析

应用1999—2003年中国中央气象台 (CMO)、日本气象厅 (JMA) 以及美国联合台风警报中心 (JTWC) 发布的西北太平洋热带气旋综合预报资料, 从总误差、逐年误差趋势、不同海区误差、不同路径趋势误差、不同强度趋势误差等5个方面对各预报中心的路径及强度预报结果进行分析, 结果表明:5年总的平均误差以JTWC的路径预报误差最小, 而JMA的强度预报较准确; 在不同海域, 各预报中心的路径预报能力各有优势, 但在热带气旋的强度预报方面, JMA的方法在各海区都较稳定; 对不同路径趋势热带气旋的预报方面, 除了南海转向热带气旋的路径预报比JMA和CMO稍差一些之外, JTWC的路径预报在大多数情况下都是好于或相当于JMA和CMO; 在不同强度变化趋势热带气旋的预报方面, JTWC在大多数情况下都优于其他中心。上述结果帮助业务和科技人员全面了解各预报中心的预报能力优劣, 也为今后改进我国的热带气旋预报提供有益的参考。     

CONDITIONAL NONLINEAR OPTIMAL PERTURBATIONS: APPLICATION IN TROPICAL CYCLONE FORECASTS

Considering the feature of tropical cyclones (TCs) that strong positive vorticity exists in the lower layers of troposphere, this study proposed to use vorticity at 850 hPa as cost function to find the conditional nonlinear optimal perturbation (CNOP), which was largely different from those previous studies using total energy of perturbed forecast variables. The CNOP was obtained by an ensemble-based approach. All of the sensitive areas determined by CNOP with vorticity at 850 hPa as cost function for the three cases were located over the TC core region and its vicinity. The impact of the CNOP-based adaptive observations on TC forecasts was evaluated with three cases via observational system simulation experiments (OSSEs). Results showed obvious improvements in TC intensity or track forecasts due to the CNOP-based adaptive observations, which were related to the main error source of the verification area, i.e., intensity error or location error.     

Evaluation of Tropical Cyclone Intensity Forecasts from Five Global EnsemblePrediction Systems During 2015-2019

This study presented an evaluation of tropical cyclone (TC) intensity forecasts from five global ensemble prediction systems (EPSs) during 2015-2019 in the western North Pacific region. Notable error features include the underestimation of the TC intensity by ensemble mean forecast and the under-dispersion of the probability forecasts.The root mean square errors (brier scores) of the ensemble mean (probability forecasts) generally decrease consecutively at long lead times during the five years, but fluctuate between certain values at short lead times.Positive forecast skill appeared in the most recent two years (2018-2019) at 120 h or later as compared with the climatology forecasts. However, there is no obvious improvement for the intensity change forecasts during the 5-yearperiod, with abrupt intensity change remaining a big challenge. The probability forecasts show no skill for strongTCs at all the lead times. Among the five EPSs, ECMWF-EPS ranks the best for the intensity forecast, while NCEP-GEFS ranks the best for the intensity change forecast, according to the evaluation for ensemble mean and dispersion. As for the other probability forecast evaluation, ECMWF-EPS ranks the best at lead times shorter than 72 h, while NCEP-GEFS ranks the best later on.     

T213与T639模式热带气旋预报误差对比

应用国家气象中心全球谱模式T213L31(简称T213) 及其升级版本T639L60(简称T639) 对2009—2010年西北太平洋热带气旋数值预报的结果进行对比。结果表明:T213与T639模式24～120 h预报平均距离误差基本相近,但由于T639模式分辨率较高,T639模式的热带气旋强度预报明显好于T213模式。从分类误差来看,T639模式对于西北行登陆及转向热带气旋的路径预报好于T213模式,但对西行及北上热带气旋预报误差偏大。对于异常路径热带气旋预报,T639模式能较好预报环流形势的突然调整,对路径突变的热带气旋预报比T213模式有明显优势;从登陆类热带气旋预报的移向误差来看,T213模式存在东北偏北向系统性偏差,T639模式存在东北偏东向系统性偏差。