考虑风向分布的芝罘岛极值风速季节变化研究

Study of Seasonal Change of the Extreme Wind Speed at Zhifudao Observation Station Considering Wind Direction Distribution

基于角度-线性模型,提出利用风玫瑰图数据直接建立风速与风向联合分布的方法,对烟台芝罘岛地区四季度风玫瑰图建立联合分布模型,采用决定系数评价模型与原始玫瑰图一致性,结果表明统计模型可靠性高,与原始风向风速分布相关性极强;利用联合分布模型,结合根据条件概率推导得到的二维风向风速伪随机数生成算法,对芝罘岛地区100年的风况进行了蒙特卡洛模拟,从中筛选各季节、方位年极值风速数据,结合收集到的该地区1981—1992年极值风速观测数据进行比较,结果显示模拟数据可较好的反映各方位极值风速的变化;进一步利用模拟的年极值数据,对芝罘岛地区各季节,不同方位的10、20及50年风速重现值进行了预测。

Wind is one of the most important meteorological elements with two main parameters of speed and direction.Wind speed is the characteristic we concern most,and Weibull distribution is widely taken as the most suitable statistic model.Wind direction is also very important,for its distribution,a mixture of von Mises distribution is widely used.Wind speed and direction are closely related,thus it is more accurate to research on wind based on the joint distribution of wind speed and direction.Many researchers conduct extensive research on the approach of building joint probability model of these two variates.However,since wind speed is linear variable while wind direction varies on circle,their joint probability distribution,for example,different with the joint distribution of wave height and period,which are both linear variables,belongs to a special type of linear-circular joint distribution,and there are scarce literatures about constructing the smooth joint distribution of linear and circular variables.Meanwhile,construction of joint bivariate probability model needs adequate continuous observed in-situ data.Lack of long term observation also restricts its modeling.In comparison,wind rose is a more common and extensive datum to analyze wind characteristics.Based on the principle of maximum entropy,Johnson and Wehrly firstly proposed a linear-circular distribution with systematicly measured wind data,which attracted widespread attention for its clear form and better applicability.In this paper,based on the continuous angular-linear distribution by Johnson and Wehrly,a new approach is proposed to establish the joint probabilistic distribution of wind speed and direction just by using the data of wind rose.The method was taken to construct the seasonal statistical models of wind regime at Zhifudao Observation Station located in Yantai.In order to assess the fitness of the joint probabilistic models to the original wind roses,the coefficients of determination were calculated for both the joint cumulative probability and interval frequency.The results show that the coefficients of determination calculated are quite closed to one,which means that constructed statistical models have higher reliability and strong correlation with their original wind roses.Based on conditional probability method of multi-variables statistics,we also present an algorithm to generate pseudo-random numbers of wind speed and concomitant direction.The built joint probabilistic models were applied to generate the wind data with a span of 100 years by Monte Carlo simulation.The directional seasonal extreme wind speed was picked out.Then the boxplots were used to analyze the simulated 100 year seasonal extreme wind speed data.Comparing with the observed extreme wind speed data of every direction from 1981 to 1992,the results show that the simulated data can give a reasonable estimation of the ranges of the extreme wind speed for all four seasons and every direction.Furthermore,the directional seasonal extreme wind speeds under different return periods(e.g.10,20,and 50 years)are estimated by the simulated extreme wind data.For the wind regime at Zhifudao,we got following conclusions through analyses:(1)the wind regime of Zhifudao is mainly influenced by three monsoons,varing significantly with seasons,(2)for all year around,although the monsoons coming from SSE and W play a dominant role in summer and fall-winter respectively,the most serious extreme wind speeds mainly occur in NNW-N orientation,even though it is not an obvious wind direction,which should pay more attention for the engineering practice at this site.