In this paper, by using the wave data from a few oceanographic observation stations in the coastal zone of the Yellow Sea, the East China Sea and the South China Sea, the long-term joint distribution of the one-tenth ...In this paper, by using the wave data from a few oceanographic observation stations in the coastal zone of the Yellow Sea, the East China Sea and the South China Sea, the long-term joint distribution of the one-tenth large (or significant) wave height with average period is studied. The statistical data demonstrate that the long- term distribution of the one- tenth wave height or average period fits the log-normal distribution, thus the joint distribution also fits the two-dimensional log-normal distribution. Then the conditional probability distribution of the average period is derived, and the range as well as the mode of the average wave period corresponding to a certain return period of wave height can be calculated easily.展开更多
By analysing the scatter diagrams of characteristic the wave height H and the period T on the basis of instrumental data from various ocean wave stations, we found that the conditional expectation and standard deviati...By analysing the scatter diagrams of characteristic the wave height H and the period T on the basis of instrumental data from various ocean wave stations, we found that the conditional expectation and standard deviation of wave period for a given wave height can be better predicted by using the equations of normal linear regression rather than by those based on the log- normal law. The latter was implied in Ochi' s bivariate log-normal model(Ochi. 1978) for the long-term joint distribution of H and T. With the expectation and standard deviation predicted by the normal linear regression equations and applying proper types of distribution, we have obtained the conditional distribution of T for given H. Then combining this conditional P(T / H) with long-term marginal distribution of the wave height P(H) we establish a new parameterized model for the long-term joint distribution P(H,T). As an example of the application of the new model we give a method for estimating wave period associated with an extreme wave height.展开更多
The modified versions of the linear theoretical model of Longuet-Higgins （1983） are derived in this work and also compared with the laboratory experiments carried out in MAR1NTEK. The main feature of modifications i...The modified versions of the linear theoretical model of Longuet-Higgins （1983） are derived in this work and also compared with the laboratory experiments carried out in MAR1NTEK. The main feature of modifications is to replace the mean frequency in the formulation with the peak frequency of the wave spectrum. These two alternative forms of joint distributions are checked in three typical random sea states characterized by the initial wave steepness. In order to further explore the properties ＆these models, the associated marginal distributions of wave heights and wave periods are also researched with the observed statistics and some encouraging results are obtained.展开更多
在林业研究中,胸径-树高二元联合分布多由相同边缘分布构造,而林分的胸径与树高的实际分布状况可能有所差异。为降低这种差异带来的影响,依据佳木斯市孟家岗林场的115块长白落叶松人工林数据,选择适用条件低、适应范围广的Copula函数方...在林业研究中,胸径-树高二元联合分布多由相同边缘分布构造,而林分的胸径与树高的实际分布状况可能有所差异。为降低这种差异带来的影响,依据佳木斯市孟家岗林场的115块长白落叶松人工林数据,选择适用条件低、适应范围广的Copula函数方法拟合落叶松胸径-树高二元联合分布模型。首先选择威布尔(Weibull)、广义威布尔(G-Weibull)、逻辑斯蒂(Logistic)、轻量逻辑斯蒂(Logit-Logistic)、伽马(Gamma)、对数正态(Log-Normal)6个分布函数作为备选基础模型,根据K-S(kolmogorov smirnov test)检验与半参数估计结果筛选并构建Copula胸径-树高二元联合分布模型,再通过负对数似然(negative log-likelihood,NLL)、Sn拟合优度统计量和似然比检验(likelihood ratio test,LRT)与二元对数logistic分布函数和二元Weibull分布函数进行比较,最后使用雷诺误差指数(error index of Reynolds,EI)对模型预测能力进行评估。结果表明,基于Copula函数的二元分拟合结果与模型(EI=0.3184)预估能力皆优于二元Weibull分布(EI=0.6381)和二元对数Logistic分布(EI=0.9490),说明此方法构建胸径-树高二元联合Copula分布模型能够很好地描述落叶松人工林胸径树高联合分布,以Copula方法构建树高-胸径联合分布是可行的。展开更多
文摘In this paper, by using the wave data from a few oceanographic observation stations in the coastal zone of the Yellow Sea, the East China Sea and the South China Sea, the long-term joint distribution of the one-tenth large (or significant) wave height with average period is studied. The statistical data demonstrate that the long- term distribution of the one- tenth wave height or average period fits the log-normal distribution, thus the joint distribution also fits the two-dimensional log-normal distribution. Then the conditional probability distribution of the average period is derived, and the range as well as the mode of the average wave period corresponding to a certain return period of wave height can be calculated easily.
文摘By analysing the scatter diagrams of characteristic the wave height H and the period T on the basis of instrumental data from various ocean wave stations, we found that the conditional expectation and standard deviation of wave period for a given wave height can be better predicted by using the equations of normal linear regression rather than by those based on the log- normal law. The latter was implied in Ochi' s bivariate log-normal model(Ochi. 1978) for the long-term joint distribution of H and T. With the expectation and standard deviation predicted by the normal linear regression equations and applying proper types of distribution, we have obtained the conditional distribution of T for given H. Then combining this conditional P(T / H) with long-term marginal distribution of the wave height P(H) we establish a new parameterized model for the long-term joint distribution P(H,T). As an example of the application of the new model we give a method for estimating wave period associated with an extreme wave height.
基金financially supported by the European Union(Grant No.234175)the Portuguese Foundation for Science and Technology(Grant No.SFRH/BD/98983/2013)
文摘The modified versions of the linear theoretical model of Longuet-Higgins （1983） are derived in this work and also compared with the laboratory experiments carried out in MAR1NTEK. The main feature of modifications is to replace the mean frequency in the formulation with the peak frequency of the wave spectrum. These two alternative forms of joint distributions are checked in three typical random sea states characterized by the initial wave steepness. In order to further explore the properties ＆these models, the associated marginal distributions of wave heights and wave periods are also researched with the observed statistics and some encouraging results are obtained.
文摘在林业研究中,胸径-树高二元联合分布多由相同边缘分布构造,而林分的胸径与树高的实际分布状况可能有所差异。为降低这种差异带来的影响,依据佳木斯市孟家岗林场的115块长白落叶松人工林数据,选择适用条件低、适应范围广的Copula函数方法拟合落叶松胸径-树高二元联合分布模型。首先选择威布尔(Weibull)、广义威布尔(G-Weibull)、逻辑斯蒂(Logistic)、轻量逻辑斯蒂(Logit-Logistic)、伽马(Gamma)、对数正态(Log-Normal)6个分布函数作为备选基础模型,根据K-S(kolmogorov smirnov test)检验与半参数估计结果筛选并构建Copula胸径-树高二元联合分布模型,再通过负对数似然(negative log-likelihood,NLL)、Sn拟合优度统计量和似然比检验(likelihood ratio test,LRT)与二元对数logistic分布函数和二元Weibull分布函数进行比较,最后使用雷诺误差指数(error index of Reynolds,EI)对模型预测能力进行评估。结果表明,基于Copula函数的二元分拟合结果与模型(EI=0.3184)预估能力皆优于二元Weibull分布(EI=0.6381)和二元对数Logistic分布(EI=0.9490),说明此方法构建胸径-树高二元联合Copula分布模型能够很好地描述落叶松人工林胸径树高联合分布,以Copula方法构建树高-胸径联合分布是可行的。
