Some Problems on the Global Wavelet Spectrum

In order to test the validity of the global wavelet spectrum - a new period analysis method based on wavelet analysis, we carried out some simple experiments. In our experiments we used idealized time series and real Nifio 3 sea surface temperature （SST） for testing purposes. First we combined different signals which have the same power but different periods into some new time series. Then we calculated the global wavelet spectra and Fourier power spectra for the testing time series. The testing results revealed that on some occasions the global wavelet spectrum tends to amplify the relative power of longer periods. By making comparisons with the results obtained by the traditional Fourier power spectrum, we demonstrated that on an occasion when the global wavelet spectrum does not work the Fourier power spectrum can be used to achieve the right results. Hence it is recommended that when making period analysis with the global wavelet spectrum one needs to do further tests to confirm their results.     

����С�������LS-SVM�Ļ���ʱ������Ԥ��ģ�ͼ���Ӧ��

?????С???????????????????LS-SVM??????????μ???????????????????μ????????н???С???????????C-C??????????????????????????????????????????????????????????????????LS-SVM??????н?????????BP??????????????????????????????????????С???????LS-SVM????????????????????н??????Ч????     

基于神经网络的混沌时间序列预测

为了提高了RBF神经网络对于混沌时间序列的预测性能,从相空间重构理论出发,建立RBF神经网络,并且改变RBF神经网络训练参数及训练样本数,以Logistic及Lorenz混沌时间序列进行预测仿真,取得了不错的预测效果。     

桥梁水平位移混沌特征识别与神经网络预测研究

跨海大桥系统受外界影响扰动,其变形伴有混沌现象发生.对桥梁变形监测数据实现了混沌识别,运用C-C法计算时间序列的延迟时间,用G-P方法求得最佳嵌入维数,通过求取的时间延迟和最佳嵌入维数对桥梁变形监测数据进行相空间重构,为混沌时间序列预测模型的建立奠定基础;基于RBF神经网络建立混沌时间序列预测模型,对实测数据进行桥梁变...     

库岸滑坡地下水位时间序列混沌特征识别与PSO-LSSVM模型预测

地下水位预测对滑坡稳定性分析具有重要意义,三峡库区库岸滑坡地下水位时间序列在季节性强降雨和周期性库水位涨落等诸多因素影响下呈现混沌特征。在对地下水位序列进行相空间重构的基础上,采用饱和关联维数法和最大Lyapunov指数法对其混沌特征进行验证。再用预测性能优秀的最小二乘支持向量机(LSSVM)模型对其进行预测,并用粒子群算法优化选取LSSVM模型的参数,以克服LSSVM模型参数选取困难的缺点。以三峡库区三舟溪滑坡前缘STK-1水文孔日平均地下水位序列为例进行了混沌分析,分别运用粒子群优化的LSSVM模型(PSO-LSSVM)和BP神经网络模型对STK-1水文孔地下水位进行了预测。结果表明库岸滑坡地下水位序列存在混沌特征,PSO-LSSVM模型预测结果的均方根误差为0.193m,拟合优度为0.815,说明预测效果较理想,且PSO-LSSVM模型预测精度高于BP网络模型,具有较强的实用性。      

EEMD-PSOGSVM耦合模型在深基坑位移预测中的应用

针对单一预测模型的不足,提出EEMD分解与粒子群灰色支持向量机（particle swarm optimization grey support vector machine,PSOGSVM）相结合的基坑位移预测模型。以基坑时间序列的混沌性为基础,利用EEMD分解时间序列,采用相空间重构技术构造样本,应用PSOGSVM模型进行基坑预测,并与GM(1,1)、SVM、遗传小波神经网络进行对比。结果表明,该算法预测精度好,具有良好的稳定性,可有效地应用于基坑位移预测。     

����-֧���������ڴ�Ӱ�ȫ���Ԥ���е�Ӧ��

????????????????????г?????????????????μ?????????-??????????????????????????????????????????????????????????????????????????????????????????????????????????С????????????????н???????г????????????????????Ч????á?     

������������С��ȥ�������Ԥ��

?纣??????????硢????????????????????????????α???????????????????????????????????????db10С???????????????????????????????е???????????????С???????????????????????????????????????Lyapunov??????????????????ж???????????????????????????????????????з??????????????????????????     

Placement Optimization Method of FPSO Gas Detectors Based on Leakage Risk

As the first step of the fire/gas-detection systems of floating production storage and offloading(FPSO)units is to iden-tify leakage accidents,gas detectors play an important role in controlling the leakage risk.To improve the leakage scenario detection rate and reduce the cumulative risk value,this paper presents an optimization method of the gas detector placement.The probability density distribution and cumulative probability density distribution of the leakage source variables and environmental variables were calculated based on the Offshore Reliability Data and the statistical data of the relevant leakage variables.A potential leakage sce-nario set was constructed using Latin hypercube sampling.The typical FPSO leakage scenarios were analyzed through computational fluid dynamics(CFD),and the impacts of different parameters on the leakage were addressed.A series of detectors was deployed according to the simulation results.The minimization of the product of effective detection time and gas leakage volume was the risk optimization objective,and the location and number of detectors were taken as decision variables.A greedy extraction heuristic algo-rithm was used to solve the optimization problem.The results show that the optimized placement had a better monitoring effect on the leakage scenario.     

Satellite monitoring of sea surface temperatures off China

Satellite measurements of global sea surface temperatures (SST) have been made since 1982 using the multi-channel radiometers (AVHRR) on NOAA polar orbiting satellites. A four year data set was accumulated at daily intervals and a spatial resolution of about 100 kilometers on an interactive computer system. The time lapse evaluation of the data revealed variations of the SST which were related to coastal and equatorial upwelling events as well as to the pronounced equatorial warming associated with the 1982–1983 El Niño. In the present study, satellite time series are used to describe the annual variability of the SST at selected locations along the coast of China, the Yellow Sea, the Sea of Japan and the Equatorial Indian and Pacific Oceans. Further study of the SST off China using higher resolution satellite data are also described.

     

Quantification of the impact of environmental factors on chlorophyll in the open ocean

Many previous studies of the impact of oceanic environmental factors on chlorophyll(CHL)in a specific region focused on sea surface temperature(SST),mixed-layer depth(MLD),or wind stress(WS) alone.In this study,relationship between CHL and all those environmental factors(SST,MLD,and WS) in the open ocean was quantified for five regions within the subtropical gyres and the variation trend of 13-year(2003-2015) was analyzed using satellite observations and Argo measurements.The correlation analysis results show that MLD was correlated positively with CHL,SST was correlated negatively with CHL,and the correlation between CHL and WS was either positive or negative.Based on the significance of the correlations,models representing the relationships were established using the multiple linear regression and analyzed,showing that the environmental factors were the major determinants of CHL change.The regression coefficients show that both SST and MLD have remarkable effect on CHL.Our derived models could be used to diagnose the past changes,understand present variability,and predict the future state of CHL changes based on environmental factors,and help us understand the dynamics of CHL variation in the open ocean.     

Interannual variability of SST,SLA and wind waves in the Hawaii area and their responses to ENSO

Time series of sea surface temperature (SST),wind speed and significant wave height (SWH) from meteorologicalbuoys of the National Data Buoy Center (NDBC) are useful for studying the interannual variability and trend of these quantities at the buoy areas.The measurements from 4 buoys (B51001,B51002,B51003 and B51004) in the Hawaii area are used to study theresponses of the quantities to EI Nino and Southern Oscillation (ENSO).Long-term averages of these data reflect precise seasonaland climatological characteristics of SST,wind speed and SWH around the Hawaii area.Buoy observations from B51001 suggest asignificant warming trend which is,however,not very clear from the other three buoys.Compared with the variability of SST andSWH,the wind speeds from the buoy observations show an increasing trend.The impacts of El Nifio on SST and wind waves arealso shown.Sea level data observed by altimeter during October 1992 to September 2006 are analyzed to investigate the variabilityof sea level in the Hawaii area.The results also show an increasing trend in sea level anomaly (SLA).The low-passed SLA in theHawaii area is consistent with the inverse phase of the low-passed Sol (Southern Oscillation Index).Compared with the low-passedSOl and PDO (Pacific Decadal Oscillation),the low-passed PNA (Pacific-North America Index) has a better correlation with thelow-passed SLA in the Hawaii area.     

Metocean design parameter estimation for fixed platform based on copula functions

Considering the dependent relationship among wave height, wind speed, and current velocity, we construct novel trivariate joint probability distributions via Archimedean copula functions. Total 30-year data of wave height, wind speed, and current velocity in the Bohai Sea are hindcast and sampled for case study. Four kinds of distributions, namely, Gumbel distribution, lognormal distribution, Weibull distribution, and Pearson Type III distribution, are candidate models for marginal distributions of wave height, wind speed, and current velocity. The Pearson Type III distribution is selected as the optimal model. Bivariate and trivariate probability distributions of these environmental conditions are established based on four bivariate and trivariate Archimedean copulas, namely, Clayton, Frank, Gumbel-Hougaard, and Ali-Mikhail-Haq copulas. These joint probability models can maximize marginal information and the dependence among the three variables. The design return values of these three variables can be obtained by three methods: univariate probability, conditional probability, and joint probability. The joint return periods of different load combinations are estimated by the proposed models. Platform responses (including base shear, overturning moment, and deck displacement) are further calculated. For the same return period, the design values of wave height, wind speed, and current velocity obtained by the conditional and joint probability models are much smaller than those by univariate probability. Considering the dependence among variables, the multivariate probability distributions provide close design parameters to actual sea state for ocean platform design.     

Climate anomalies in the southern high latitudes associated with the Subtropical Dipole Mode

1　Introduction EnoughhasbeensaidabouttheteleconnectionofENSO(ElNi o SouthernOscilla tion)inthetropicalPacificOceanandtheclimateinthesouthernhighlatitudes(Fletcher etal.1982;SmithandStearns1993;SimmodsandJacka1995;Liuetal.2002;Kwok andComiso2002).Inres…     

Parameter identification of JONSWAP spectrum acquired by airborne LIDAR

In this study, we developed the first linear Joint North Sea Wave Project (JONSWAP) spectrum (JS), which involves a transformation from the JS solution to the natural logarithmic scale. This transformation is convenient for defining the least squares function in terms of the scale and shape parameters. We identified these two wind-dependent parameters to better understand the wind effect on surface waves. Due to its efficiency and high-resolution, we employed the airborne Light Detection and Ranging (LIDAR) system for our measurements. Due to the lack of actual data, we simulated ocean waves in the MATLAB environment, which can be easily translated into industrial programming language. We utilized the Longuet-Higgin (LH) random-phase method to generate the time series of wave records and used the fast Fourier transform (FFT) technique to compute the power spectra density. After validating these procedures, we identified the JS parameters by minimizing the mean-square error of the target spectrum to that of the estimated spectrum obtained by FFT. We determined that the estimation error is relative to the amount of available wave record data. Finally, we found the inverse computation of wind factors (wind speed and wind fetch length) to be robust and sufficiently precise for wave forecasting.     

The Equilibrium Range of Wind Wave Spectra： an Explanation Based on White Noise

Laboratory experiments and field observations show that the equilibrium range of wind wave spectra presents a – 4 power law when it is scaled properly. This feature has been attributed to energy balance in spectral space by many researchers. In this paper we point out that white noise on an oscillation system can also lead to a similar inverse power law in the corresponding displacement spectrum, implying that the – 4 power law for the equilibrium range of wind wave spectra may probably only reflect the randomicity of the wind waves rather than any other dynamical processes in physical space. This explanation may shed light on the mechanism of other physical processes with spectra also showing an inverse power law, such as isotropic turbulence, internal waves, etc.     

Warmer-Get-Wetter or Wet-Get-Wetter? A Criterion to Classify Oceanic Precipitation

In this study, the temporal and spatial variations of observed global oceanic precipitation during 1979–2010 are investigated. It is found that the global trend in precipitation during this period varies at a rate of 1.5%/K of surface warming while the rate is 6.6%/K during 2006–2010. The precipitation is highly correlated with Sea Surface Temperature(SST) in both the temporal and the spatial patterns since the strong 1997–98 El Nino event. Considering the distributions of precipitation and SST, seven oceanic regions are classified and presented using the observed Global Precipitation Climatology Project(GPCP) data and Extended Reconstructed Sea Surface Temperatures, version 3(ERSST.v3) data. Further examining the mechanisms of the classified oceanic precipitation regions is conducted using the Tropical Rainfall Measuring Mission(TRMM) satellite, GFDL-ESM-2G model precipitation and SST data and Hadley Center sea ice and SST version 1(Had ISST1) data. More than 85% of global oceanic precipitations are controlled by either one or both of the warmer-get-wetter mechanism and wet-get-wetter mechanism. It is estimated that a 0.5 SST signal-to-noise ratio, representing the trend of SST time series to the standard deviation, is a criterion to distinguish the mechanism of a region. When the SST ratio is larger than 0.5, the precipitation of this region is controlled by the warmer-get-wetter mechanism. SST, rather than the humidity, is the pivotal factor. On the other hand, when the SST ratio is less than 0.5, the precipitation is controlled by the wet-get-wetter mechanism. The SST variability is a significant factor contributing to the precipitation variation.     

一种新的基于模糊分析观测信息的局地化方法

地球表层系统是一个极其复杂的巨系统,为了更精确地表达地球表层系统各种过程的动态演进,解决数据同化系统观测误差的估计与处理已经成为地球科学领域备受关注的问题之一。在地球科学系统数值模拟中,一般采用集合数据同化来探讨地学变量预报时的各种误差。集合类卡尔曼滤波通常会由于集合数过小而带来欠采样、协方差低估、滤波发散和远距离虚假相关等问题。针对背景误差协方差被低估问题,局地分析方法（Local Analysis, LA）在一定程度上能起到抑制作用,但无法彻底解决背景误差协方差的虚假相关问题。因此,本文在集合卡尔曼滤波的算法框架下提出了一种与模糊逻辑控制算法相耦合的局地化分析方法（Fuzzy Analysis, FA）。在强非线性Lorenz-96模型中,对不同模型误差下的LA和FA方法进行了性能优劣方面的探讨,并比较分析了2种方法在集合数、观测数和观测位置、放大因子以及强迫参数变化时的同化性能。实验采用均方根误差作为算法评判依据,同时用功率谱密度（Power Spectral Density, PSD）更直接地对2种算法性能优劣作出了评价。结果表明：在完美模型下,FA相对于LA降低了17.5%的均方根误差（Root Mean Square Error, RMSE）;随着模型误差增大,RMSE减小的百分比和减小幅度都在降低;在严重模型误差下,FA降低了8.6%的RMSE。总体而言,新算法FA的有效性和鲁棒性都得到了验证,并且在EnKF同化基础下有效改进了传统的局地化分析方案,优化了观测误差处理,为今后的数据同化研究提供了一个较为全面的观测误差研究平台。