计算非均匀地表通量的一种简化PDF及其应用

提出描述地表非均匀特性的简化概率密度函数（PDF），它可代替各种具体的PDF用于求解次网格尺度平均通量而不影响其精度，这种简化PDF可直接加入陆面过程模式方程组中进行陆－气通量交换的数值试验。它对非均匀地表陆面过程参数化具有一定的普适性。本文仅讨论了对称分布的简化计算（非对称分布另文讨论），列举了非均匀分布的观测事实，并以地表净红外辐射通量计算为例，详细验证了应用这种简化PDF的可行性和可靠性。     

风廓线雷达探测大气返回信号谱的仿真模拟

利用2008年1—9月北京延庆地区对流层CFL-08风廓线雷达和2009年7—12月广东东莞地区边界层CFL-03风廓线雷达晴空大气探测资料,对大气返回信号功率谱密度分布及雷达系统噪声幅度统计特征进行分析。基于大气返回信号的谱分布符合高斯分布的统计结果,采用高斯模型对大气返回信号谱进行模拟;雷达系统噪声是白噪声,其谱线幅度的统计特征呈高斯分布,在此基础上,通过生成高斯分布随机函数对雷达探测输出信号进行信号谱分布的仿真模拟,并采用雷达探测信号的谱参数对仿真模拟进行检验,效果较好。该文应用仿真模拟的数据,对风廓线雷达信息处理方法及其处理精度进行了初步试验和分析。     

时间序列混合模型及其在原油价格预测中的应用

在时间序列建模时,随机扰动项经常表现出非高斯性质,造成传统的时序模型的估计和预测偏误．针对这个问题,提出一种新的时间序列混合模型———多维高斯混合转移分布模型（ＭＧＭＴＤ模型）,用来处理残差项不是白噪声的情况,并提出采用遗传算法进行参数估计．将新提出的ＭＧＭＴＤ模型应用于对国际原油价格的预测分析中,进行模型的可行性检验．实证结果显示,ＭＧＭＴＤ模型可以得到较好的预测结果．     

全球多源海冰密集度融合资料研制试验

为了发展一套全球多源海冰密集度逐日融合资料,以欧洲气象卫星应用组织（EUMETSAT）海洋海冰应用中心（OSI SAF）海冰密集度数据、中国国家卫星气象中心（NSMC）的MWRI和VIRR全球海冰密集度数据、美国国家冰雪数据中心（NSIDC）的NISE海冰密集度数据、美国国家冰中心（NIC）的IMS北半球海冰数据为观测数据源,以ERA-Interim模式数据为背景场,采用以下方案开展融合试验。首先,对各数据源资料进行质量控制;其次,以OSI SAF海冰密集度数据为基准,采用概率密度函数（PDF）匹配方法订正其他卫星资料的系统误差;然后,根据订正后的误差生成超级观测场;最后,利用STMAS方法将超级观测场和作为背景场的ERA-Interim海冰密集度数据进行融合,生成全球逐日0.25°分辨率海冰密集度融合试验数据。通过与国际广泛使用的OISST、OSTIA海冰密集度数据对比,评估融合试验产品的质量。结果表明:融合方案中的PDF方法通过调整非基准资料的概率密度分布,实现非基准资料和基准资料概率密度分布一致,从而使3种海冰密集度卫星资料系统误差均显著减小;STMAS方法能够将超级观测场和背景场进行有效融合,生成融合试验产品;风云卫星数据的使用提高了融合数据生产的自主可控能力;同时,融合方案考虑了卫星数据源的时效性、获取的稳定性等因素。融合产品与OISST和OSTIA海冰密集度数据的空间分布在南、北极均高度吻合,相关系数均超过0.985,与OISST和OSTIA的偏差分别为?1.170%和0.276%,融合试验产品整体偏差介于两种资料之间,反映了试验产品系统误差较小的良好特性。可见,融合方案能够满足实时业务需要,融合试验产品具有较高的质量。      

基于执行依赖启发式动态规划的船舶减摇鳍在线最优控制

针对船舶线性横摇系统,设计了一种基于执行依赖启发式动态规划（ADHDP）方法的在线学习最优减摇鳍控制器.在设计过程中直接使用输入输出数据获取系统状态值.利用评价网络来逼近针对船舶减摇鳍控制系统设计的性能指标函数,并通过执行网络获得最优控制律,这两个网络都是多层前馈神经网络,即反向传播（BP）神经网络.在训练过程中,这两个神经网络不仅可以使用实时测量数据,也可以减少船舶横摇模型的内部误差和不确定性干扰的影响,从而提高系统的鲁棒性.最后,仿真结果表明所提出的ADHDP控制器对于降低船舶横摇有很好的控制效果.     

基于机器学习的气温要素空间插值

掌握近地表气温的空间分布对于科学理解生态环境变化和陆地-大气热力学相互作用机理等至关重要,然而气象观测站在地理空间上分布不均,为区域的气温插值带来了挑战。本文以全国711个气象站点的月平均气温为基础数据,分析了气温与经度、纬度和高程的相关性,利用随机森林、支持向量机、高斯过程回归三种机器学习方法,对中国陆地区域的气温进行了空间插值,并与反距离权重、普通克里金以及ANUSPLIN等传统的插值方法以及全球土地数据同化系统（Global Land Data Assimilation System,GLDAS）气温数据产品进行对比。结果表明：（1）在整个中国陆地区域,利用机器学习进行气温插值的精度明显高于传统的反距离权重和普通克里金插值,并且机器学习方法明显提高了夏季的气温插值精度（R2）;(2)在三种机器学习方法中,高斯过程回归表现最好,支持向量机次之,然后是随机森林,ANUSPLIN具有较高的插值精度,但是整体低于高斯过程回归模型;（3）在气象站点稀少且分布不均匀的青藏高原地区,三种机器学习方法插值精度明显高于反距离权重和普通克里金方法,且整体优于ANUSPLIN的结果,尤其是在夏季,机器...     

多维非高斯系统最小熵控制

针对多维非高斯系统提出了最小熵控制方法,控制的目标是使系统的非高斯输出概率密度函数跟踪一个已知的联合概率密度函数.首先,根据系统模型和辅助映射,构建了系统状态、跟踪误差与扰动输入之间的泛函算子模型,然后基于梯度算法设计了递归的次优控制律,最后通过仿真验证了最小熵控制算法的有效性.     

高阶非线性系统自适应模糊有限时间状态约束控制

针对一类具有状态约束的非严格反馈高阶非线性系统,研究一种自适应模糊有限时间跟踪控制问题.首先,利用模糊逻辑系统逼近不确定性非线性函数,在此基础上,采用障碍Lyapunov函数,解决状态约束问题,通过障碍加幂积分方法和反步递推技术,提出了一种有限时间控制设计方法.在有限时间Lyapunov稳定意义下,严格证明闭环系统半全局实际有限时间稳定且系统的状态不超出给定的约束边界,并实现了有限时间跟踪控制目标.最后,仿真研究进一步验证了所提出控制方法的有效性.     

气象服务系统数据统计功能二次开发的设计和实现

针对目前气象服务系统普遍存在二次开发能力不足、推广应用困难的技术难题,设计了一套具有强大自定义功能的服务数据分析系统。该系统采用内部函数和外部函数相结合的方法,内部函数采用动态链接库的形式开发,其接口可以连接各种业务数据库,并通过配置不同的计算方法和参数,实现多种需求输出;数据前台界面显示通过自定义的sql语句接口参数来实现;在内部函数计算的基础上,利用基于外部函数的文字模板实现数据描述文本的输出。基于该方案开发的气象服务系统从用户界面、数据分析、数据显示、绘图和文本输出等各种功能均可实现自定义,系统后期升级方便,跨区域推广能力强,其设计思路对相关业务系统的开发具有很好的指导意义。     

基于粒子群优化的投影寻踪聚类模型及其应用

投影寻踪聚类分析是根据设计的投影指标函数,并在相关约束条件下进行问题优化分析的过程.给出了用于求解投影指标函数的粒子群算法,并将构造的模型应用于森林承载力评价.仿真实验结果表明:与基于遗传算法优化的模型比较,基于粒子群优化的模型简单、容易实现并且没有许多参数需要调整;在应用上,基于粒子群优化的模型可获得更优的解,并可预计模型在森林承载力评价中具有重要的应用价值.     

粒子群优化算法在随机分布控制中的应用

将粒子群优化算法应用到随机分布系统中,其随机系统控制目标不局限于传统的均值和方差,而是估算某些变量的概率密度函数.该方法能减少基于泛函算子模型的随机分布控制算法在仿真过程中的计算量,避免计算中间变量的概率密度函数,且对模型的要求不高,从而使控制结果更加精确高效.     

Background Error Covariance Statistics of Hydrometeor Control Variables Based on Gaussian Transform

Use of data assimilation to initialize hydrometeors plays a vital role in numerical weather prediction(NWP).To directly analyze hydrometeors in data assimilation systems from cloud-sensitive observations,hydrometeor control variables are necessary.Common data assimilation systems theoretically require that the probability density functions(PDFs)of analysis,background,and observation errors should satisfy the Gaussian unbiased assumptions.In this study,a Gaussian transform method is proposed to transform hydrometeors to more Gaussian variables,which is modified from the Softmax function and renamed as Quasi-Softmax transform.The Quasi-Softmax transform method then is compared to the original hydrometeor mixing ratios and their logarithmic transform and Softmax transform.The spatial distribution,the non-Gaussian nature of the background errors,and the characteristics of the background errors of hydrometeors in each method are studied.Compared to the logarithmic and Softmax transform,the Quasi-Softmax method keeps the vertical distribution of the original hydrometeor mixing ratios to the greatest extent.The results of the D′Agostino test show that the hydrometeors transformed by the Quasi-Softmax method are more Gaussian when compared to the other methods.The Gaussian transform has been added to the control variable transform to estimate the background error covariances.Results show that the characteristics of the hydrometeor background errors are reasonable for the Quasi-Softmax method.The transformed hydrometeors using the Quasi-Softmax transform meet the Gaussian unbiased assumptions of the data assimilation system,and are promising control variables for data assimilation systems.     

A two-dimensional trajectory-simulation model for non-Gaussian,inhomogeneous turbulence within plant canopies

We report a two-dimensional (alongwind u, vertical w) trajectory-simulation model, consistent with Thomson's (1987) well-mixed criteria, that allows for the non-Gaussian turbulence typical of flow within a plant canopy. The effect of non-Gaussian turbulence was examined by formulating a non-Gaussian u, w joint probability density function (PDF) as the sum of two Gaussian joint-PDFs. The resultant PDF reproduced the desired means, variances, skewnesses, and kurtoses, and the correct covariance. In prediction of the location of maximum concentration downwind of a line source in homogeneous, slightly non-Gaussian turbulence, it proved advantageous to incorporate skewness and kurtosis. However, in the case of inhomogeneous, highly non-Gaussian turbulence, the addition of skewness and kurtosis in the model resulted in substantially worse agreement with measurements than the results of the model using Gaussian PDFs. This may be due to inaccuracy in our PDF formulation. Dispersion predictions from the model with Gaussian PDFs were generally not statistically different from measurements. These results indicate that a two-dimensional Gaussian trajectory-simulation approach is adequate to predict mean concentrations and fluxes resulting from sources within plant canopies.     

Probability Distribution Function of a Forced Passive Tracer in the Lower Stratosphere

The probability distribution function （PDF） of a passive tracer, forced by a ＂mean gradient＂, is studied. First, we take two theoretical approaches, the Lagrangian and the conditional closure formalisms, to study the PDFs of such an externally forced passive tracer. Then, we carry out numerical simulations for an idealized random flow on a sphere and for European Center for Medium-Range Weather Forecasts （ECMWF） stratospheric winds to test whether the mean-gradient model can be applied to studying stratospheric tracer mixing in midlatitude surf zones, in which a weak and poleward zonal-mean gradient is maintained by tracer leakage through polar and tropical mixing barriers, and whether the PDFs of tracer fluctuations in midlatitudes are consistent with the theoretical predictions. The numerical simulations show that when diffusive dissipation is balanced by the mean-gradient forcing, the PDF in the random flow and the Southern-Hemisphere PDFs in ECMWF winds show time-invariant exponential tails, consistent with theoretical predictions. In the Northern Hemisphere, the PDFs exhibit non-Gaussian tails. However, the PDF tails are not consistent with theoretical expectations. The long-term behavior of the PDF tails of the forced tracer is compared to that of a decaying tracer. It is found that the PDF tails of the decaying tracer are time-dependent, and evolve toward flatter than exponential.     

海气耦合非线性随机模式求解探讨

本文首先对一个假定随机强迫为高斯白噪声的四次方非线性海气耦合随机模式进行了ε一阶近似的摄动处理，将非线性随机模式化为显含随机项的线性系统和隐含随机项的非线性系统，通过对输出响应的分析表明:该模式结构和随机项的特性使得随机强迫输出的响应只能通过模式的线性区；非线性项仅对气候系统的方差及其对应的频谱有一定量的偏差作用，而对一般输出特性没有贡献。在这基础上，求出了模式输出谱，并给出了计算。     

Study of the Probability Density Functions From a Large-Eddy Simulation for a Stably Stratified Boundary Layer

Turbulence in a non-strongly stably stratified large-eddy simulation (LES) case is studied through probability density functions (PDFs) to obtain additional information than that provided by classical LES averages. The PDFs are computed for one hour within the steady-state regime at three different levels: near the surface, in the middle and at the top of the boundary layer, for the wind components and the temperature. The physical significance of these PDFs from LES is discussed and they are compared to those obtained from observations. The analysis of the eddy structures within the stably stratified boundary layer is made through the combined study of the fields, the spectra and the statistical moments obtained from the PDFs and joint PDFs. The homogeneity of the fields is inspected through a comparison of the ensemble to the temporal and the spatial PDFs, showing that the ergodic theorem is not fulfilled. To this end, the sensitivity of the PDF moments to the LES resolution is explored.     

随机非完整系统控制问题综述和展望

随着随机非线性控制的发展,随机非完整系统的控制引起了学者们的注意.本文首先探讨了随机非完整控制系统的镇定问题,涉及严反馈链式系统的反馈镇定和不满足严反馈的移动机器人镇定等;其次,介绍了该系统跟踪控制及现状;最后,在总结现有结果的基础上,分析了随机非完整系统发展的趋势,给出了6个可能的研究方向.     

基于数据驱动高阶学习律的轮式移动机器人轨迹跟踪控制

轮式机器人执行巡逻、播种和工业生产等任务是一个强非线性的间歇过程.针对重复运行的轮式机器人轨迹跟踪问题,本文提出了一种基于数据驱动的高阶迭代学习控制算法.首先,对轮式移动机器人的模型进行推导设计,并对推导得到的状态空间形式的离散时间模型利用基于状态转移的迭代动态线性化方法,将轮式机器人系统转化为线性输入输出数据模型;其次,设计高阶迭代优化目标函数得到控制律,并利用参数更新律估计线性输入输出数据模型中的未知参数.控制器的设计和分析只使用系统的输入输出数据,不包含任何显式的模型信息.通过采用高阶学习控制方法,在控制律中利用更多之前迭代的控制输入信息,提高了控制性能.最后,仿真结果验证了该方法在轮式机器人轨迹跟踪控制中的有效性.     

COMPARISON AMONG EMPIRICAL PROBABILITY DENSITY FUNCTIONS OF THE VERTICAL VELOCITY IN THE SURFACE LAYER BASED ON HIGHER ORDER CORRELATIONS

Vertical velocity fluctuations were measured in theatmospheric surface layer by means of an ultrasonic anemometer andhigher order correlations were calculated on two time series, recordedin unstable and neutral conditions, and selected for the wholemeasurement period on the basis of the inversion test (stationaritytest). Comparisons have been made between observed and predictedcorrelations by considering Gaussian joint-PDF and Gram-Charlierseries expansions truncated to the fourth and sixth order as doneearlier by Frenkiel and Klebanoff. A bi-Gaussian PDF, given by amixture of two Gaussian PDFs, has also been considered. This lasthas been constructed assuming that either the first three or the firstfour moments are given, and the relationships between correlationfunctions of different order are derived. The departure from Gaussianbehaviour in both stability conditions is derived. Though Gram-Charlier series expansions show a good correspondence toexperimental reality, their use as non-Gaussian probabilitydistributions cannot be suggested in theoretical approaches andshould be considered with care in practical applications, due topossible occurrences of small negative probabilities. The resultsshown in this paper support the applicability of the bi-Gaussian PDFcreated using up to the fourth moment.