高海况海洋遥感信息提取技术研究进展

中等海况下,星载合成孔径雷达(Synthetic Aperture Radar,SAR)已经广泛应用于海洋动力环境要素的监测(风场、波浪、流场)。近年来,SAR高海况遥感,尤其是探测台风海面风场、巨浪、流场已经成为国内外研究热点,并突破了一些关键技术。利用SAR多极化成像模式对海观测和新发展的地球物理模式函数,可以提取高海况下的海面风速、风向、有效波高、流速和流向等海洋表面关键物理参数。这些环境要素可以用于海洋灾害监测预警;为海洋和大气数值模式提供准确的初始场和同化源,改进模式预报精度;为研究全球气候变化提供有力的观测依据。     

海洋科技人才需求预测方法研究

海洋科技人才需求预测是海洋领域人才规划的重要内容,是制定海洋人才战略的基础性工作。在现有人才需求模型的基础上,结合海洋科技和海洋产业发展的特点,分析了海洋科技人才需求预测的步骤和指标体系,建立了海洋科技人才需求的预测模型。     

南海及周边海域风浪流耦合同化精细化数值预报与信息服务系统简介

面向社会需求,建立覆盖南海及周边海域的高分辨率风-浪-流耦合同化数值预报与信息服务系统。系统包含耦合同化数值预报模式、海洋动力环境数据库与可视化平台两部分。其中,耦合同化数值预报模式由中尺度大气数值预报模式、海浪数值预报模式和区域海洋环流数值模式,在C-Coupler耦合器中进行耦合,引入集合调整Kalman滤波同化模块,在耦合预报前进行大气、海浪和海流的同化后报模拟,为耦合预报模式提供更为精确的初始场。预报结果经海洋动力环境数据库和可视化平台处理后,通过二维和三维可视化展示,向用户提供直观的南海及周边海域海洋环境预报产品。     

Marsh-Schulkin与Rogers模型比对分析

基于实测传播损失数据对比分析Marsh-Schulkin和Rogers经验传播损失模型。通过讨论东海声学调查具体的海洋环境,较详细地标定了两模型对应的声学参数,并融入模型,结合实测数据进行对比验证,得出关于两模型在东海适应性的如下结论:两模型预报结果与实测数据吻合较好,且两者预报曲线大致相同;其中MarshSchulkin模型预报曲线斜率略大于Rogers模型,预报差异随距离增大有所增加;各频率预报曲线存在交点,交点左侧Rogers模型更接近实测数据,右侧Marsh-Schulkin更为准确。     

Short-Range Prediction Experiments with Operational Data Assimilation System for the Kuroshio South of Japan

The short-range (one month) variability of the Kuroshio path was predicted in 84 experiments (90-day predictions) using a model in an operational data assimilation system based on data from 1993 to 1999. The predictions started from an initial condition or members of a set of initial conditions, obtained in a reanalysis experiment. The predictions represent the transition from straight to meander of the Kuroshio path, and the results have been analyzed according to previously proposed mechanisms of the transition with eddy propagation and interaction acting as a trigger of the meander and self-sustained oscillation. The reanalysis shows that the meander evolves due to eddy activity. Simulation (no assimilation) shows no meander state, even with the same atmospheric forcing as the prediction. It is suggested therefore that the initial condition contains information on the meander and the system can represent the evolution. Mean (standard deviation) values of the axis error for all 84 cases are 13, 17, and 20 (10, 10, and 12) km, in 138.5°E, in the 30-, 60-, and 90-day predictions respectively. The observed mean deviation from seasonal variation is 30 km. The predictive limit of the system is thus about 80 days. The time scale of the limit depends on which stage in the transition is adopted as the initial condition. The gradual decrease of the amplitude in a stage from meander to straight paths is also predicted. The predictive limit is about 20 days, which is shorter than the prediction of the opposite transition. This revised version was published online in July 2006 with corrections to the Cover Date.     

卫星高度计资料在三维海温和盐度数值预报中的应用

随着卫星遥感观测技术的发展,越来越多的卫星观测资料被应用于数值模式的同化研究中.基于国家海洋环境预报中心西北太平洋三维湿盐流预报系统,利用法国CLS中心的沿轨高度计资料的海表面高度异常的融合数据,结合基于三维变分的OVALS(ocean variational analysis system)同化系统,在垂向将海面高度...     

基于物理关联深度学习的海浪浪高预测方法

精确的海浪有效波高（简称浪高）预测对于海上生产生活具有重要意义。针对现有海浪浪高预测模型对不同海洋要素间关联信息考虑不足,以及长时序浪高数据本身存在非平稳性的问题,本文设计了一种考虑物理约束与差值约束的海浪浪高时间序列预测方法。该方法基于风速与浪高之间的物理关联,设计物理约束,并通过提取差分信息设计差值约束,结合现有基于深度学习的时间序列预测模型,实现浪高预测。采用黄海和东海的6个不同站点浮标数据进行了大量实验。实验结果表明,本文提出的方法可以利用海洋要素间的物理关联,有效提高浪高预测精度,并避免因不同要素间融合造成的信息间干扰;同时,利用差值约束,限制时间序列预测结果的变动范围。本文方法可以与不同类型的时间序列预测模型相结合,显著提升原有模型的性能,并在长时间序列的预测中体现出很好的鲁棒性,为海洋要素预测中物理与数据驱动模型的有效结合提供了思路和验证。     

一种基于小波分解和模糊推理神经网络的潮位预测方法

潮位预测严重影响沿海区域，尤其是近海浅水沿岸地区居民的生产生活和涉海活动。谐波分析是长周期潮位预测的传统方法，但无法预测非周期性气象过程发生时的水位变化。与数据处理方法相结合，人工智能的方法通过拟合输入与输出数据的历史数值关系，能够有效预测高度非线性和非平稳的流模式，因而在时间序列数据预测领域得到了广泛的应用。本文结合自适应模糊推理系统（Adaptive Neuro-Fuzzy Inference System, ANFIS）和小波分解方法，利用水位异常和风切变分量作为输入数据，实现了一种综合的多时效潮位预测方法。文中测试了多种输入变量组合和小波-ANFIS（WANFIS）模型，并与人工神经网络（Artificial Neural Network, ANN）、小波-ANN（WANN）和ANFIS模型进行了预测结果对比。通过不同指数的误差分析来看，相比ANN模型，ANFIS模型能够更准确的预测潮位变化，小波分解对ANFIS预测精度有一定的提高，且模型中水位异常和风切变分量数据的加入比单一的潮位数据输入能取得更好的预测结果。     

一个嵌套于MM5下的蒸发波导预报模式

Evaporation duct is an abnormal refractive phenomenon in the marine atmosphere boundary layer. It has been generally accepted that the evaporation duct prominently affects the performance of the electronic equipment over the sea because of its wide distribution and frequent occurrence. It has become a research focus of the navies all over the world. At present, the diagnostic models of the evaporation duct are all based on the Monin-Obukhov similarity theory, with only differences in the flux and character scale calculations in the surface layer. These models are applicable to the stationary and uniform open sea areas without considering the alongshore effect.This paper introduces the nonlinear factor av and the gust wind item wg into the Babin model, and thus extends the evaporation duct diagnostic model to the offshore area under extremely low wind speed. In addition, an evaporation duct prediction model is designed and coupled with the fifth generation mesoscale model(MM5).The tower observational data and radar data at the Pingtan island of Fujian Province on May 25–26, 2002 were used to validate the forecast results. The outputs of the prediction model agree with the observations from 0 to 48 h. The relative error of the predicted evaporation duct height is 19.3% and the prediction results are consistent with the radar detection.     

FIO-ESM气候模式北太平洋海表温度和降水的季节性预报技巧评估

The seasonal prediction of sea surface temperature(SST) and precipitation in the North Pacific based on the hindcast results of The First Institute of Oceanography Earth System Model(FIO-ESM) is assessed in this study.The Ensemble Adjusted Kalman Filter assimilation scheme is used to generate initial conditions, which are shown to be reliable by comparison with the observations. Based on this comparison, we analyze the FIO-ESM 6-month hindcast results starting from each month of 1993–2013. The model exhibits high SST prediction skills over most of the North Pacific for two seasons in advance. Furthermore, it remains skillful at long lead times for midlatitudes. The reliable prediction of SST can transfer fairly well to precipitation prediction via air-sea interactions.The average skill of the North Pacific variability(NPV) index from 1 to 6 months lead is as high as 0.72(0.55) when El Ni?o-Southern Oscillation and NPV are in phase(out of phase) at initial conditions. The prediction skill of the NPV index of FIO-ESM is improved by 11.6%(23.6%) over the Climate Forecast System, Version 2. For seasonal dependence, the skill of FIO-ESM is higher than the skill of persistence prediction in the later period of prediction.     

江苏近海地层原位剪切波速相关特性及预测方法研究

剪切波速测试是原位勘测常用且有效的技术之一,其测试成果可用于分析场地土层动力学特性。海洋地层测试条件恶劣,在某些情况下对剪切波速的预测分析尤为重要。为了研究海洋地层精确的剪切波速预测方法,结合江苏近海及潮间带的剪切波速原位测试成果,总结和对比分析了剪切波速预测方法,评判了剪切波速的变化特性和与土体物理力学指标的统计关系。基于广义回归神经网络(GRNN)方法,通过剪切波速与土体各参数的统计关系,建立了剪切波速与土体各物理力学指标的非线性映射关系,进行了剪切波速的预测分析,得到了较好的预测结果。     

The Adriatic Sea M2 and K1 tides by 3D model and data assimilation

In the present work we explore the impact of assimilating local tide-gauge and altimetric data on the quality of predicting the major Adriatic tides (M₂ and K₁). To that end we compute optimal tidal open boundary conditions for a 3D high-resolution finite-element model by using an incremental assimilation formalism. The essence of the method is the use of two dynamical models where the solution in the complex 3D high-resolution model is sought via assimilation of prediction errors into the simpler 2D model with explicit inverse. In the central numerical experiment, harmonic constants from 12 tide gauges are assimilated and the results are analysed at 31 locations, hence 19 independent ones. The data assimilation contributes to the reduction of maximum amplitude error from 5.6 to 0.5 cm for M₂ and from 3.9 to 0.1 cm for K₁. The assimilation procedure is repeated by assimilating suitably processed Topex/Poseidon altimeter data, again validating the outcome at 31 tide gauge locations. The result was very similar to the gauge-data assimilation outcome. The model output is also validated with the current data, not used in the assimilation. At two locations and at three depths the model was able to reproduce the major and the minor semi-axes of tidal ellipses, as well as their orientations very well.     

Prediction of maximum wave-induced liquefaction in porous seabed using multi-artificial neural network model

In the last few decades, considerable efforts have been devoted to the phenomenon of wave-induced liquefactions, because it is one of the most important factors for analysing the seabed and designing marine structures. Although numerous studies of wave-induced liquefaction have been carried out, comparatively little is known about the impact of liquefaction on marine structures. Furthermore, most previous researches have focused on complicated mathematical theories and some laboratory work. In the present study, a data dependent approach for the prediction of the wave-induced liquefaction depth in a porous seabed is proposed, based on a multi-artificial neural network (MANN) method. Numerical results indicate that the MANN model can provide an accurate prediction of the wave-induced maximum liquefaction depth with 10% of the original database. This study demonstrates the capacity of the proposed MANN model and provides coastal engineers with another effective tool to analyse the stability of the marine sediment.     

海洋资料同化对气候季节-年际预测技巧及初始场的影响试验

集合卡尔曼滤波(Ensemble Kalman filter, EnKF)是一种国内外广泛使用的海洋资料同化方案, 用集合成员的状态集合表征模式的背景误差协方差, 结合观测误差协方差, 计算卡尔曼增益矩阵, 有效地将观测信息添加到模式初始场中。由于季节、年际预测很大程度上受到初始场的影响, 因此资料同化可以提高模式的预测性能。本文在NUIST-CFS1.0预测系统逐日SST nudging的初始化方案上, 利用EnKF在每个月末将全场(full field)海表温度(sea surface temperature, SST)、温盐廓线(in-situ temperature and salinity profiles, T-S profiles)以及卫星观测海平面高度异常(sea level anomalies, SLA)观测资料同化到模式初始场中, 对比分析了无海洋资料同化以及加入同化后初始场的区别、加入海洋资料同化后模式提前1~24个月预测性能的差异以及对于厄尔尼诺-南方涛动(El Niño-southern oscillation, ENSO)预测技巧的影响。结果表明, 加入海洋资料同化能有效地改进初始场, 并且呈现随深度增加初始场改进越显著的特征。加入同化后, 对全球SST、次表层海水温度的平均预测技巧均有一定的提高, 也表现出随深度增加预测技巧改进越明显的特征。但加入海洋资料同化后, 模式对ENSO的预测技巧有所下降, 可能是由于模式误差的存在, 使得同化后的预测初始场从接近观测的状态又逐渐恢复到与模式动力相匹配的状态, 加剧了赤道太平洋冷舌偏西、中东部偏暖的气候平均态漂移。     

Multi-step ahead short-term predictions of storm surge level using CNN and LSTM network

Storm surges pose significant danger and havoc to the coastal residents' safety, property, and lives, particularly at offshore locations with shallow water levels. Predictions of storm surges with hours of warning time are important for evacuation measures in low-lying regions and coastal management plans. In addition to experienced predictions and numerical models, artificial intelligence (AI) techniques are also being used widely for short-term storm surge prediction owing to their merits in good level of prediction accuracy and rapid computations. Convolutional neural network (CNN) and long short-term memory (LSTM) are two of the most important models among AI techniques. However, they have been scarcely utilised for surge level (SL) forecasting, and combinations of the two models are even rarer. This study applied CNN and LSTM both individually and in combination towards multi-step ahead short-term storm surge level prediction using observed SL and wind information. The architectures of the CNN, LSTM, and two sequential techniques of combining the models (LSTM–CNN and CNN–LSTM) were constructed via a trial-and-error approach and knowledge obtained from previous studies. As a case study, 11 a of hourly observed SL and wind data of the Xiuying Station, Hainan Province, China, were organised as inputs for training to verify the feasibility and superiority of the proposed models. The results show that CNN and LSTM had evident advantages over support vector regression (SVR) and multilayer perceptron (MLP), and the combined models outperformed the individual models (CNN and LSTM), mostly by 4％–6％. However, on comparing the model computed predictions during two severe typhoons that resulted in extreme storm surges, the accuracy was found to improve by over 10％ at all forecasting steps.     

COSMIC掩星资料同化对台风“天兔”预报影响的试验

台风数值预报的关键之一在于给定准确的模式初始场。本文使用气象、电离层和气候星座观测系统(COSMIC)的卫星资料,通过资料同化技术改善数值天气预报模式的初始场,进而评估数据同化技术对台风(或热带气旋)预报的改善效果。使用我国自主研发的数值天气预报模式,对1319号台风“天兔”进行模拟试验,对温度场、湿度场、涡度散度场及假相当位温场进行诊断分析。试验结果表明,与同化常规探空资料相比,COSMIC资料的同化可减小模式预报台风路径误差,台风附近降水落区更为集中。研究评估了掩星资料同化对模式的改进作用,并能为台风预报技术和防灾减灾提供参考。     

Filters for linear sea-wave prediction

Deterministic sea-wave prediction (DSWP) models are appearing in the literature designed for quiescent interval prediction in marine applications dominated by large swell seas. The approach has focused upon spectral methods which are straightforward and intuitively attractive. However, such methods have the disadvantage that while the sea is aperiodic in nature, the standard discrete spectral processing techniques force an absolutely periodic structure onto the resulting sea surface prediction models. As it is the shape of the sea surface that is important in such applications, particularly near the end of the domain which is important, the standard windowing techniques used in signal processing work to reduce leakage artifacts cannot be employed. This has necessitated the use of end matching methods that can be both inconvenient and may reduce the fraction of the time for which legitimate predictions are available. As a result, an investigation has been undertaken of the use of finite impulse response prediction filters to provide the necessary dispersive phase shifting required in DSWP systems. The present work examines the theoretical basis for such filters and explores their properties together with their application to both long and short crested swell seas. It is shown that wide band forms of such filters are only convergent in the sense of distributions having both infinite duration impulse responses and asymptotically divergent first derivatives. However, appropriate band limitation can produce useful finite impulse responses allowing implementation via standard discrete convolution methods. It is demonstrated that despite the prediction filters having a non-causal impulse response such filters can be used in practice due to a combination of the asymmetric nature of the impulse response and the fundamental nature of the prediction process. The findings are confirmed against actual sea-wave data.     

基于LSTM的海洋表面短期风速预测研究

为实现对海面风速精确的短期预测,提出了一种基于长短期记忆(LSTM,long short-term memory)神经网络的短期风速预测模型,选取OceanSITES数据库中单个浮标站点采集的风速历史数据作为模型输入,经过训练设置最佳参数等步骤,实现了以LSTM方法,对该站点所在海区海面风速在各季节性代表月份海面风速的24h短期预测。同时通过不同预测时长的实验以及与BP(back propagation)神经网络神经网络和径向基函数神经网络(radial basis function neural network,RBF)的预测效果对比实验,证明了LSTM预测方法相比上述两种神经网络预测方法,在海表面风速预测应用中的优越性。最后通过多个海域对应的站点风速数据预测实验,证明了LSTM神经网络模型的普遍适用性,由相关系数和预测误差的分析可知该方法具备应对急剧变化数据的预测稳定性,可以作为海洋表面风速短期预测的一种可靠方法。     

基于最小二乘和卡尔曼滤波方法进行原子时预报的研究

介绍和分析了最小二乘和卡尔曼滤波方法在时间预报中的应用。通过IGS站提供的钟差数据，分别运用这两种方法对其中六个原子钟进行了时间预报的实验。通过对预报结果进行分析，结论证明为了取得较好的预报效果，不同的预报方法和钟参数的模型对于观测数据的要求有一些差别。从实验数据所成图形来看，当采用一天的观测数据进行模型预报时，最小二乘法的预报精度比卡尔曼滤波法稍高一些。     

全球大洋潮汐模式在北印度洋潮汐预报准确性的评估

为评估DTU10、TPXO8、GOT00.2和NAO.99b 4个全球大洋潮汐模式对北印度洋潮汐的预报能力,采用英国海洋资料中心提供的海区中部和沿岸站潮汐调和常数资料,检验了这些模式4个主要分潮(M_2、S_2、K_1、O_1)的准确度。它们的各分潮调和常数资料准确度都比较高,振幅绝均差的最大值仅5.61 cm,迟角绝均差的最大值仅9.13°。这些模式的调和常数给出潮波传播特征差别不大。基于这些模式提供的调和常数,分别建立了北印度洋4、8和16分潮潮汐预报模型,将预报结果与中国海事服务网提供的沿岸24个站潮汐表资料进行对比。各模式的8分潮(M_2、S_2、N_2、K_2、K_1、O_1、P_1、Q_1)潮汐预报模型均优于4分潮(M_2、S_2、K_1、O_1)潮汐预报模型,NAO.99b模式可以提供16分潮(M_2、S_2、N_2、K_2、K_1、O_1、P_1、Q_1、MU_2、NU_2、T_2、L_2、2N_2、J_1、M1、OO_1)潮汐预报模型,但是对预报结果改善不明显;在各模式中,GOT00.2模式的8分潮潮汐预报模型对北印度洋沿岸的预报效果最好,平均绝均差为14.97 cm。