条件非线性最优扰动方法在适应性观测研究中的初步应用

针对适应性观测中敏感性区域的确定问题，考虑初始误差对预报结果的影响, 比较了条件非线性最优扰动(CNOP)与第一线性奇异向量(FSV)在两个降水个例中的空间结构的差异，考察了它们总能量范数随时间发展演变的异同。结合敏感性试验的分析，揭示了预报结果对CNOP类型的初始误差的敏感性要大于对FSV类型的初始误差的敏感性，因而减少初值中CNOP类型误差的振幅比减少FSV类型的收益要大。这一结果表明可以把CNOP方法应用于适应性观测来识别大气的敏感区。关于将CNOP方法有效地应用于适应性观测所面临的挑战及需要采取的对策等也进行了讨论。     

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。      

基于奇异矢量目标观测的梅雨锋低涡观测系统模拟试验

就江淮梅雨锋低涡预报基于奇异矢量目标观测作了观测系统模拟试验,目的在于对基于奇异矢量目标观测实际实施作预先研究,寻找目标观测中所要遵循原则和实施细节,以及用奇异矢量确定目标观测区的恰当方法。经分析实际奇异矢量相关误差如何影响预报特征,得出在实施目标观测时应遵循的原则：只对奇异矢量相关误差进行订正,不对非奇异矢量相关误差订正;对奇异矢量强相关误差区域优先订正能更为高效率地改进预报;对整个垂直气柱进行订正,而不只对满足阈值区域进行订正;应优先采用效率较高的第一类斜压订正方案。文中两种方法确定的目标观测区与实际奇异矢量相关误差区域在位置、大小、形状上比较相似,两种方法的目标观测区误差影响预报方式与真实奇异矢量相关误差影响预报方式很相近。     

集合预报多尺度奇异向量初值扰动方法研究

目前国际上采用的奇异向量集合预报初值扰动法对于初值不确定性的描述存在一定的不足,为了更有效地反映初始误差的时空多尺度特性,基于GRAPES全球奇异向量计算技术,计算了不同空间分辨率及不同最优时间间隔的多个尺度的奇异向量,并采用基于高斯分布的线性组合法来构造多尺度奇异向量的扰动初值,以代表在相空间中增长最快的多尺度初值误差模态。通过2019年1月19日的初值扰动集合预报试验,对比分析了单一尺度奇异向量初值扰动法与多尺度初值扰动法的扰动特征及集合预报效果。结果表明,多尺度奇异向量初值扰动法为区域集合预报提供的初始扰动场是合理的,扰动的大小随时间增长,且在空间分布上较好地反映了当前大气的斜压不稳定特征。此外,多尺度奇异向量扰动可以描述一定的大尺度以及中小尺度运动误差特征,较单一尺度奇异向量扰动能反映出更多初始场的不确定性信息。检验分析表明,GRAPES多尺度奇异向量集合预报在集合一致性、连续等级概率评分、离群值等方面有一定的优势,相比于单一尺度奇异向量法有较好的预报技巧。因此,基于GRAPES的多尺度奇异向量初值扰动法对于集合预报的预报效果有一定的提高,能为构建一套完善的GRAPES区域奇异向量集合预报系统提供一定的科学依据和应用基础。      

基于总能量模的GRAPES全球模式奇异向量扰动研究

以发展基于奇异向量技术为初值扰动的GRAPES全球集合预报系统为目的,在GRAPES模式及其干动力框架下的切线性、伴随模式基础上开展了以总能量模为权重算子的奇异向量计算技术研究,建立奇异向量的计算求解模块,并通过奇异向量检验方法和切线性近似方法验证了奇异向量求解的正确性.通过对中高纬度的GRAPES奇异向量水平结构的线性演变分析,证实了在最优时间间隔内GRAPES奇异向量能够快速增长,并能描述中高纬度大气的斜压不稳定特征.分析在初始时刻和最优化时间间隔时刻的GRAPES奇异向量总能量及其分量(动能和势能)的垂直分布特征,发现在中高纬度区域,GRAPES奇异向量能够描述对流层不同层次的斜压不稳定增长特征.     

一种求解条件非线性最优扰动的快速算法及其在台风目标观测中的初步检验

条件非线性最优扰动(CNOP)是Mu等2003年提出的一个新的理论方法,它是线性奇异向量在非线性情形的推广,克服了线性奇异向量不能代表非线性系统最快发展扰动的缺陷,成为非线性系统可预报性和敏感性等研究新的有效工具.然而,由于以往CNOP的求解需要采用伴随技术,计算量相当巨大,限制了该方法的推广应用.为了克服这一困难,本文基于经验正交分解(EOF),提出了一种求解CNOP的快速算法,利用GRAPES区域业务预报模式实现了CNOP快速计算,并在台风"麦莎"的目标观测研究中得到初步检验,通过观测系统模拟实验(OSSE)检验了该方法确定敏感性区域(瞄准区)的有效性和可行性.试验结果表明,用快速算法求解的CNOP,其净能量随时间快速地发展,而且发展呈非线性.在台风"麦莎"个例的目标观测试验中,用快速算法得到的预报时间为24 h的CNOP可以有效地识别瞄准区,并通过瞄准区内初值的改善,可明显减少目标区域(检验区)内24 h累计降水预报误差.尤其,累计降水预报的这种改进效果能够延伸到更长时间(如72 h),尽管检验时间是设在第24小时.进一步分析发现,24 h累计降水预报误差的减少是通过利用瞄准区内改善的初值改进初始时刻台风暖心结构、高空相对涡度以及水汽条件等而得以实现的.     

适应性观测及其策略问题

从适应性观测(目标观测)概念提出后,确定进行适应性观测的时间、敏感区域的方法,即适应性观测策略得到不断发展,本文介绍了目前最主要的几种适应性观测策略,其中包括奇异矢量、繁殖矢量、伴随敏感性、集合转换Kalman滤波等适应性观测策略,以及用于台风的适应性观测策略.总结了适应性观测及其策略的相关理论问题,以及各种适应性观测策略之间的相关关系和不同,讨论了适应性观测对预报改进的影响因素,如观测误差、同化方案、模式误差等.为了实施适应性观测的业务应用、比较不同的适应性策略的适用性,国际上针对不同的高影响天气过程,在不同地区开展了一系列适应性观测外场试验.本文总结了近几年来开展的适应性观测外场试验.这些试验结果表明,平均而言适应性观测可有效地改进高影响天气过程的数值天气预报,但实施高影响性天气的适应性观测业务仍然是一个挑战性任务.     

北京地区暴雨个例的观测敏感区研究

利用基于集合预报的相关方法对2009年7月23日发生在北京及周边地区的暴雨过程的观测敏感区进行了分析。通过WRF(Weather Research Forecast)三维变分方法对初始场进行随机扰动,形成30个初始集合样本,做了预报时效为12 h的集合预报。利用该方法分析检验区(北京及周边地区)累积降水[14:00(北京时间,下同)至20:00]相对于初始时刻(08:00)各基本要素的敏感性,确定感性要素及其对应的区域。研究发现初步确定的敏感性要素为水汽和温度,对应的敏感区分别位于北京的西南侧和北京的东北侧,且通过实况分析可知初步确定的敏感性要素和对应的敏感区具有明确的物理意义。还进一步通过观测系统模拟试验(OSSE)的资料同化验证所确定的敏感区,结果表明在水汽对应的敏感区内同化水汽对降水的预报结果有明显的改进;在温度对应的敏感区内同化温度,降水的预报准确率有了明显的提高,说明了初步确定的敏感性要素和敏感区的正确性。在水汽对应的敏感区内同化水汽的同时在温度对应的敏感区内同化温度,使降水预报的技巧有大幅度的提高,说明了温度和水汽的共同作用对提高降水预报准确率贡献最大。因此,通过基于集合预报的相关方法能够快速的确定敏感区。研究结果将为确定北京暴雨的观测敏感区提供参考。     

适应性观测与集合变换卡尔曼滤波方法介绍

给出适应性观测理论和集合变换卡尔曼滤波方法及其研究现状的综述。重点介绍了集合变换卡尔曼滤波方法及其相关的一些问题。在数值预报领域,一种新的途径是利用数值预报系统信息在预报时效内确定出某些区域,如果在这些区域进行补充观测,可以最有效地改进预报技能。这种方法被称为适应性或目标观测,所确定的观测区域称为敏感区,敏感区内增加观测后分析质量将得到改善,对后续的预报技能可产生最大的预期影响。目前适应性观测研究已经成为世界气象组织(WMO)组织的THORPEX计划的一个子计划。集合变换卡尔曼滤波(The Ensemble Transform Kalman Filer,简称ETKF)是一种次优的卡尔曼滤波方案,最早是作为一种适应性观测算法提出的,现在还被用于集合预报初始扰动的生成。ETKF方法不仅可以同化观测资料,而且可以估计出观测对预报误差的影响。它与其它集合卡尔曼滤波方案不同之处在于:ETKF利用集合变换和无量纲化的思想求解与观测有关的误差协方差矩阵,可以快速估计出不同附加观测造成的预报误差协方差的减少量,预报误差减少最多的一组观测所对应的区域就是所寻找的敏感区。     

台风强度模拟的海温目标观测研究

根据非线性强迫奇异向量(NFSV)型海温(SST)强迫误差识别的敏感性特征,通过观测系统模拟试验(OSSE)确定了12个热带气旋(TC)的强度模拟的海温目标观测最优布局.NFSV型SST强迫误差敏感区一般沿着台风移动路径,主要位于台风快速增强阶段.结果 表明,在NFSV型SST强迫目标观测敏感区内以90 km间隔加密海...     

A Comparison Study of the Contributions of Additional Observations in the Sensitive Regions Identified by CNOP and FSV to Reducing Forecast Error Variance for the Typhoon Morakot

The sensitive regions of conditional nonlinear optimal perturbations (CNOPs) and the first singular vector (FSV) for a northwest Pacific typhoon case are reported in this paper. A large number of probes have been designed in the above regions and the ensemble transform Kalman filter (ETKF) techniques are utilized to examine which approach can locate more appropriate regions for typhoon adaptive observations. The results show that, in general, the majority of the probes in the sensitive regions of CNOPs can reduce more forecast error variance than the probes in the sensitive regions of FSV. This implies that adaptive observations in the sensitive regions of CNOPs are more effective than in the sensitive regions of FSV. Furthermore, the reduction of the forecast error variance obtained by the best probe identified by CNOPs is twice the reduction of the forecast error variance obtained by FSV. This implies that dropping sondes, which is the best probe identified by CNOPs, can improve the forecast more than the best probe identified by FSV. These results indicate that the sensitive regions identified by CNOPs are more appropriate for adaptive observations than those identified by FSV.     

Which Features of the SST Forcing Error Most Likely Disturb the Simulated Intensity of Tropical Cyclones?

Among all of the sources of tropical cyclone(TC) intensity forecast errors, the uncertainty of sea surface temperature(SST) has been shown to play a significant role. In the present study, we determine the SST forcing error that causes the largest simulation error of TC intensity during the entire simulation period by using the WRF model with time-dependent SST forcing. The SST forcing error is represented through the application of a nonlinear forcing singular vector(NFSV)structure. For the selected 12 TC cases, the NFSV-type SST forcing errors have a nearly coherent structure with positive(or negative) SST anomalies located along the track of TCs but are especially concentrated in a particular region. This particular region tends to occur during the specific period of the TCs life cycle when the TCs present relatively strong intensity, but are still intensifying just prior to the mature phase, especially within a TC state exhibiting a strong secondary circulation and very high inertial stability. The SST forcing errors located along the TC track during this time period are verified to have the strongest disturbing effect on TC intensity simulation. Physically, the strong inertial stability of TCs during this time period induces a strong response of the secondary circulation from diabatic heating errors induced by the SST forcing error. Consequently, this significantly influences the subsidence within the warm core in the eye region, which,in turn, leads to significant errors in TC intensity. This physical mechanism explains the formation of NSFV-type SST forcing errors. According to the sensitivity of the NFSV-type SST forcing errors, if one increases the density of SST observations along the TC track and assimilates them to the SST forcing field, the skill of TC intensity simulation generated by the WRF model could be greatly improved. However, this adjustment is most advantageous in improving simulation skill during the time period when TCs become strong but are still intensifying just prior to reaching full maturity. In light of this, the region along the TC track but in the time period of TC movement when the NFSV-type SST forcing errors occur may represent the sensitive area for targeting observation for SST forcing field associated with TC intensity simulation.     

Estimating the error in vertically interpolated forecast‐observation differences: U‐ and v‐wind components

Abstract

Data assimilation in numerical weather forecasting corrects current forecast values by subtracting a portion of interpolated forecast‐minus‐observation differences at the points of a three‐dimensional grid. Deviations used in updating a forecast data field are forecast errors obtained or derived from observations available at update time. When observations are missing at mandatory levels, construction of full vertical soundings by interpolation introduces extraneous errors. The present paper is concerned with determination of the error in vertical extrapolations of surface winds, and of aircraft and satellite cloud‐tracked winds. In addition it examines the effect on accuracy of using location‐specific statistics compared to averaged statistics as the basis for the interpolation weighting scheme and compares errors of one‐ and two‐variable interpolations.

Interpolation accuracy tests demonstrate the influence of the interpolation scheme on the quality of interpolated information used in forecast updating. The results show that the level of accuracy exceeds the benchmark provided by monthly mean forecast error values only with bivariate interpolation of wind components from off‐level data sources.     

Improvement in Background Error Covariances Using Ensemble Forecasts for Assimilation of High-Resolution Satellite Data

Satellite data obtained over synoptic data-sparse regions such as an ocean contribute toward improving the quality of the initial state of limited-area models. Background error covariances are crucial to the proper distribution of satellite-observed information in variational data assimilation. In the NMC (National Meteorological Center) method, background error covariances are underestimated over data-sparse regions such as an ocean because of small differences between different forecast times. Thus, it is necessary to reconstruct and tune the background error covariances so as to maximize the usefulness of the satellite data for the initial state of limited-area models, especially over an ocean where there is a lack of conventional data. In this study, we attempted to estimate background error covariances so as to provide adequate error statistics for data-sparse regions by using ensemble forecasts of optimal perturbations using bred vectors. The background error covariances estimated by the ensemble method reduced the overestimation of error amplitude obtained by the NMC method. By employing an appropriate horizontal length scale to exclude spurious correlations, the ensemble method produced better results than the NMC method in the assimilation of retrieved satellite data. Because the ensemble method distributes observed information over a limited local area, it would be more useful in the analysis of high-resolution satellite data. Accordingly, the performance of forecast models can be improved over the area where the satellite data are assimilated.     

Observation System Experiments for Typhoon Nida (2004) Using the CNOP Method and DOTSTAR Data

This study investigated the influence of dropwindsonde observations on typhoon forecasts. The study also evaluated the feasibility of the conditional nonlinear optimal perturbation (CNOP) method as a basis for sensitivity analysis of such forecasts. This sensitivity analysis could furnish guidance in the selection of targeted observations. The study was performed by conducting observation system experiments (OSEs). This research used the fifth-generation Mesoscale Model (MM5), the Weather Research and Forecasting (WRF) model, and dropsonde observations of Typhoon Nida at 1200 UTC 17 May 2004. The dropsondes were collected under the operational Dropsonde Observations for Typhoon Surveillance near the Taiwan Region (DOTSTAR) program. In this research, five kinds of experiments were designed and conducted:(1) no observations were assimilated; (2) all observations were assimilated;(3) observations in the sensitive area revealed by the CNOP method were assimilated;(4) the same as in (3), but for the region revealed by the first singular vector (FSV) method;and (5) observations within a randomly selected area were assimilated. The OSEs showed that (1) the DOTSTAR data had a positive impact on the forecast of Nida’s track;(2) dropsondes in the sensitive areas identified by the MM5 CNOP and FSV remained effective for improving the track forecast for Nida on the WRF platform;and (3) the greatest improvement in the track forecast resulted from the CNOP-based (third) simulation, which indicated that the CNOP method would be useful in decision making about dropsonde deployments.     

FSU-GSM forecast error sensitivity to initial conditions: Application to Indian summer monsoon

Summary The full-physics adjoint of the FSU Global Spectral Model of version T42L12 is applied to carry out sensitivity analysis of the localized model forecast error to the initial conditions for a case test occurring on June 8, 1988 during the Indian summer monsoon. The results show that adjoint sensitivity based on ECMWF analysis can be used to identify regions with large analysis uncertainties. The conclusion is that more observations are required over the northern Bay of Bengal to improve the quality of analyses so as to ameliorate the model forecast skill.With 17 Figures     

京津冀站点风温湿要素的机器学习订正方法

基于线性回归方法、梯度提升回归方法(GBRT方法)、XGBoost方法和堆叠集成学习方法(Stacking方法)4种机器学习方法,采用误差分析建模思路,针对北京城市气象研究院研发的睿图-睿思系统对2020年12月—2021年11月所有起报时次未来3~12 h的2 m温度、2 m相对湿度、10 m风速以及10 m风向4种气象要素预报,开展京津冀复杂地形下的站点预报误差订正技术研究及试验应用。结果表明:基于预报误差分析构建的4种订正模型中,由于Stacking方法集成了前3种方法的优势,在4个季节的4种气象要素订正中均表现最佳,其他3种单一机器学习方法试验中,XGBoost方法表现最佳,其后依次为GBRT方法、线性回归方法,但均对预报准确率有明显的正向提升效果。总体上,基于机器学习方法构建的预报误差订正模型可有效降低系统原始预报误差,有助于进一步提升复杂地形下站点客观释用产品的预报准确性。