【作者】 王际朝；

【导师】 张杰；

【作者基本信息】 中国科学院研究生院（海洋研究所） ， 物理海洋学， 2014， 博士

【摘要】 海浪数据同化是提高海浪模拟和预报精度的重要途径。影响海浪同化模式精度的四个主要因素为：海浪模式，强迫风场，观测数据以及同化方法。论文基于海浪模式WAVEWATCH III (WW3)，从上述四个因素入手，开展了南海、缅因湾和夏威夷三个海域的海浪模拟和同化实验，对海浪模式的三种输入耗散方案进行了评估，研究了同步观测的风场和海浪数据对海浪同化效果的影响机制，研究了基于谱分割策略的EnvisatASAR二维海浪谱最优插值同化。论文的工作可作为中法海洋星CFOSAT卫星应用的预研，对提高我国海浪模拟和预报水平具有实际意义。论文的第一部分工作是基于HY-2高度计和测波雷达的有效波高数据评估了WW3中提供的三种输入耗散方案，为高质量的海浪模拟提供基础。WW3提供的三种输入耗散方案WAM3，TC96和WAM4都能较好地模拟出海浪的演变趋势，但是在涌浪占主导的情况下，模拟效果均不甚理想。在这些输入耗散方案中，基于‘有效风速’策略的TC96方案的模拟效果最佳，其中的风速校正参数是非常敏感的，为获得高精度的模拟结果，需对此参数进行合理选择。论文的第二部分工作是以QSCAT/NCEP和CCMP风场为背景风场，通过多组海浪有效波高同化实验，分析了同步观测的风场和海浪数据对同化效果的影响。首先基于NDBC浮标数据，评估了QSCAT/NCEP和CCMP风场，二者精度都较高，2008年3月QSCAT/NCEP和CCMP风速的均方根误差分别为2.93和1.61m/s.然后以这两种风场和融合了浮标风速、风向的风场作为海浪模式的强迫风场，进行了多组海浪同化实验，最后得出结论：既融合观测风场又同化同步获取的海浪数据的同化效果最佳；海浪数据与风场数据在同化效果中所起的作用约为4:1，海浪数据同化能减小海浪模式对强迫风场的高度敏感性。论文的第三部分工作是基于谱分割策略，实现了Envisat ASAR二维海浪谱数据的最优插值同化，进行了多组模拟和同化实验，比较了有效波高同化与海浪谱同化的同化效果。首先基于NDBC浮标的一维频率谱数据，对2009年至2011年共三年的Envisat ASAR波模式的二维海浪谱数据进行了对比分析。然后对ASAR有效波高、浮标有效波高和ASAR二维海浪谱数据采用不同的同化策略进行了同化实验，最后得出结论：基于谱分割策略的ASAR二维海浪谱同化效果最佳，与仅同化波高相比，均方根误差降低了30%，同化率提高了35%；用于同化的数据源的质量是直接决定同化效果好坏的主要因素之一，浮标的波高同化优于ASAR的波高同化，除了数据质量，待同化数据的观测密度也会对同化效果产生严重影响；虽然ASAR海浪谱的数据质量不如浮标的高，但是ASAR海浪谱同化比浮标波高同化的均方根误差改善了5.6%，同化率提高了17.7%，这是由于二维海浪谱中除了包含表示海浪总能量的有效波高之外，还含有波周期，波传播方向等海浪信息，基于谱分割策略的最优插值同化方法可以用这些信息订正模式背景谱。更多还原

【Abstract】 Wave data assimilation is an important way to improve the precision of wavesimulation and forecast. Four main factors affected the accuracy of the waveassimilation model are the wave model, the forcing wind field, the observation dataand the assimilation method. In view of the above four factors, based on the wavemodel WAVEWATCH III (WW3), wave simulation and assimilation experiments areprocessed in the three sea areas, the South China Sea, the gulf of Maine and theHawaii. This article evaluates the three input/dissipation schemes of WW3model,and explores the impact mechanism of synchronous wind and wave dataobservations to the assimilation effects. Based on the spectrum partition strategy,Envisat ASAR two-dimensional wave spectra are assimilated using the optimalinterpolation method. The work in this paper could serve as the pre-study ofCFOSAT satellite application and has practical significance to improving the level ofour wave simulation and forecast.First, based on the HY-2altimeter and the significant wave height of the waveobservation radar, this work evaluates WW3’s three input/dissipation schemes inorder to establish high quality wave simulation and forecast model. The threeschemes, WAM3, TC96and WAM4, could simulate the evolution trend of the wavenicely, but in the case of ocean swell dominated, simulation results are not very ideal.The simulation result of TC96scheme is the best which is based on the strategy of"effective wind speed" considered the atmospheric instability. Because the windspeed correction parameter is very sensitive, it should be selected reasonably in orderto obtain high accuracy of the simulation result.Secondly, taking the QSCAT/NCEP and CCMP wind field as the backgroundwind field, this work fuses wind speed and direction of10buoys, analyzes dataassimilation effect of the synchronous wind and wave through several groups ofsignificant wave height assimilation experiments. The two wind fields have highprecision, the root mean square error of the QSCAT/NCEP and CCMP wind speed in March2008are2.93m/s and1.61m/s, respectively. Taking the two original windfields and the wind fields fused wind speed and wind direction of buoys as the forcewind field of wave model, wave assimilation experiments are carried out. Theconclusions are that: the assimilation effect is the best which both fusesobservational wind field and assimilates synchronous wave data; the ratio of waveand wind data effect to the assimilation is about4:1; wave data assimilation canreduce sensitive of the wave model to the force wind field.Thirdly, based on the spectra partition scheme, Envisat ASAR two-dimensionalwave spectral is assimilated with the optimal interpolation method. This workcompares assimilation effect of the significant wave height assimilation and wavespectral assimilation. The conclusions are that: the effect of Envisat ASARtwo-dimensional wave spectral assimilation based on the spectra partition scheme isthe best, compared with the significant wave height assimilation, and the RMSreduces about30%and AI improves about35%; The quality of the data source forassimilation is one of the main factors that determine assimilation effect. Waveheight assimilation of the buoy is better than wave height assimilation of ASAR, thusbeside the data quality, observation density will be a serious impact on assimilationeffect; compared with buoy wave height assimilation, the RMS of ASAR wavespectral assimilation reduces5.6%and AI improves17.7%. It is due to that thetwo-dimensional wave spectral contains not only the significant wave height, whichrepresents total energy of wave, but also other wave information such as wave period,wave propagation direction and so on. Based on the spectrum partition strategy, theoptimal interpolation assimilation method can use those information to correct themodel background spectral.更多还原