渤黄东海三维风暴潮模式四维变分同化技术研究
发布时间：2018-07-21 11:49
【摘要】：本文利用区域海洋模式ROMS(Regional Ocean Modelling System)及其自带的四维变分同化模块，建立了渤黄东海三维风暴潮数值模式。四维变分同化能够保持水动力过程的协调稳定，通过积分切线性模式和伴随模式，并且运用梯度下降算法，使模拟结果在同化窗口内最大程度的靠近观测。 采用随水深变化的底摩擦系数，对潮汐进行了数值模拟，分析了四个主要分潮（M2，S2，K1，O1）的同潮时图。结果显示该模式具备较好的潮汐模拟能力，无论无潮点的位置还是等振幅、等迟角线都和前人的模拟结果非常接近。同时，大部分海洋站点的潮汐模拟结果和观测吻合较好。 选定两次温带风暴潮作为研究对象，通过将海洋站观测水位同化到模式中，很大程度上改进了风暴潮的模拟结果。设计了5组孪生敏感性试验对影响同化模拟结果的要素进行研究，结果表明，风应力对风暴潮的模拟质量起着决定性的作用，初始场只在短时间内影响模拟结果，风速越大，影响时间越短。因此，将风应力和初始场同时作为控制变量可以到的最优的模拟结果。同化模拟不仅能够提高参与同化海洋站的风暴潮模拟精度，还能改善其周围的模拟结果。在同化模拟的过程中，迭代次数是一个非常重要的参数，它决定了目标函数—模式结果和观测之间距离的下降程度。迭代次数越多，风暴潮的模拟结果就会与观测吻合越好，但同时，也要付出更加高昂的计算代价。 同化模拟一段时间之后，，能够得到一个与观测吻合较好的最优预报初始场和最优的风应力拖曳系数，分别用最优的拖曳系数和常数拖曳系数（0.0026）对风暴潮进行数值预报。结果表明，最优预报初始场对预报结果的改进存在不确定性。减水个例的改进效果不明显，前期预报结果产生了较大误差，可能是因为模式对减水过程的物理描述还不完善，在大风条件下，误差显著的表现出来；增水个例的改进效果可以从预报起始时刻一直持续到观测增水的峰值附近，表明最优初始场可以提高临近预报的准确度。 同化模拟反演的最优拖曳系数能否进一步改善预报结果，取决于两个方面的因素。一方面是同化窗口和预报窗口风速的误差变化情况，如果风速误差在同化前后差异很大，最优拖曳系数适用性一定会变差；另一方面是不同风速条件下拖曳系数的变化幅度，即在不同风速条件下，如果实际拖曳系数变化幅度很大，最优拖曳系数的适用性同样会受到挑战。
[Abstract]:In this paper, the regional ocean model ROMS (Regional Ocean Modeling system) and its own four-dimensional variational assimilation module are used to establish a three-dimensional storm surge numerical model in the Bohai Sea, Yellow Sea and East China Sea. Four-dimensional variational assimilation can keep the coordination and stability of hydrodynamic process. By means of integral tangent linear model and adjoint model and gradient descent algorithm, the simulated results can be observed in the assimilation window as close as possible. With the bottom friction coefficient varying with the depth of water, the tide is numerically simulated, and the isotidal time diagrams of four main tidal components (M2S2K1O1) are analyzed. The results show that the model has good tidal simulation ability, and the isobaric angle line is very close to the previous simulation results, regardless of the position of no tide point or the equal amplitude. At the same time, the tidal simulation results of most ocean stations are in good agreement with observations. Two temperate storm surges were selected as the research object. By assimilating the observed water levels of ocean stations to the model, the simulation results of storm surges were improved to a great extent. Five sets of twin sensitivity tests were designed to study the factors that affect the simulation results of assimilation. The results show that wind stress plays a decisive role in the simulation quality of storm surge, the initial field only affects the simulation results in a short time, and the wind speed increases. The shorter the influence time is. Therefore, the wind stress and the initial field are taken as the optimal simulation results for the control variables. Assimilation simulation can not only improve the accuracy of storm surge simulation, but also improve the simulation results around it. In the process of assimilation simulation, the number of iterations is a very important parameter, which determines the degree of decrease of the distance between the objective function-model result and the observation. The more iterations, the better the simulation result of storm surge will be, but at the same time, it will be more expensive to calculate. After a period of assimilation simulation, an optimal prediction of initial field and an optimal drag coefficient of wind stress can be obtained. The optimal drag coefficient and constant drag coefficient (0.0026) are used to predict storm surge numerically, respectively. The results show that the improvement of the initial field of the optimal prediction is uncertain. The improvement effect of water reduction case is not obvious, and the result of early forecast has a big error, which may be because the physical description of water reduction process is not perfect, and the error is obvious under the condition of strong wind. The improved effect of water increasing example can be continued from the beginning of prediction to near the peak value of observed water increase, which indicates that the optimal initial field can improve the accuracy of near prediction. Whether the optimal drag coefficient of assimilation simulation inversion can further improve the prediction results depends on two factors. On the one hand, the variation of wind speed error in assimilation window and forecast window, if the error of wind speed is very different before and after assimilation, the applicability of the optimal drag coefficient will certainly become worse; on the other hand, the variation range of drag coefficient under different wind speed conditions, In other words, the applicability of the optimal drag coefficient will also be challenged if the actual drag coefficient varies greatly under different wind speeds.
【学位授予单位】：国家海洋环境预报中心
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P731.23
本文编号：2135429
[Abstract]:In this paper, the regional ocean model ROMS (Regional Ocean Modeling system) and its own four-dimensional variational assimilation module are used to establish a three-dimensional storm surge numerical model in the Bohai Sea, Yellow Sea and East China Sea. Four-dimensional variational assimilation can keep the coordination and stability of hydrodynamic process. By means of integral tangent linear model and adjoint model and gradient descent algorithm, the simulated results can be observed in the assimilation window as close as possible. With the bottom friction coefficient varying with the depth of water, the tide is numerically simulated, and the isotidal time diagrams of four main tidal components (M2S2K1O1) are analyzed. The results show that the model has good tidal simulation ability, and the isobaric angle line is very close to the previous simulation results, regardless of the position of no tide point or the equal amplitude. At the same time, the tidal simulation results of most ocean stations are in good agreement with observations. Two temperate storm surges were selected as the research object. By assimilating the observed water levels of ocean stations to the model, the simulation results of storm surges were improved to a great extent. Five sets of twin sensitivity tests were designed to study the factors that affect the simulation results of assimilation. The results show that wind stress plays a decisive role in the simulation quality of storm surge, the initial field only affects the simulation results in a short time, and the wind speed increases. The shorter the influence time is. Therefore, the wind stress and the initial field are taken as the optimal simulation results for the control variables. Assimilation simulation can not only improve the accuracy of storm surge simulation, but also improve the simulation results around it. In the process of assimilation simulation, the number of iterations is a very important parameter, which determines the degree of decrease of the distance between the objective function-model result and the observation. The more iterations, the better the simulation result of storm surge will be, but at the same time, it will be more expensive to calculate. After a period of assimilation simulation, an optimal prediction of initial field and an optimal drag coefficient of wind stress can be obtained. The optimal drag coefficient and constant drag coefficient (0.0026) are used to predict storm surge numerically, respectively. The results show that the improvement of the initial field of the optimal prediction is uncertain. The improvement effect of water reduction case is not obvious, and the result of early forecast has a big error, which may be because the physical description of water reduction process is not perfect, and the error is obvious under the condition of strong wind. The improved effect of water increasing example can be continued from the beginning of prediction to near the peak value of observed water increase, which indicates that the optimal initial field can improve the accuracy of near prediction. Whether the optimal drag coefficient of assimilation simulation inversion can further improve the prediction results depends on two factors. On the one hand, the variation of wind speed error in assimilation window and forecast window, if the error of wind speed is very different before and after assimilation, the applicability of the optimal drag coefficient will certainly become worse; on the other hand, the variation range of drag coefficient under different wind speed conditions, In other words, the applicability of the optimal drag coefficient will also be challenged if the actual drag coefficient varies greatly under different wind speeds.
【学位授予单位】：国家海洋环境预报中心
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P731.23
【参考文献】
相关期刊论文 前2条
1 赵庆良;许世远;王军;胡蓓蓓;叶明武;刘耀龙;;沿海城市风暴潮灾害风险评估研究进展[J];地理科学进展;2007年05期
2 傅赐福;于福江;王培涛;刘秋兴;董剑希;;滨海新区温带风暴潮灾害风险评估研究[J];海洋学报(中文版);2013年01期
本文编号：2135429
本文链接：https://www.wllwen.com/kejilunwen/haiyang/2135429.html
