Study on the validity of wave prediction models by direct numerical simulations

常规的数值波浪预报已成为可能，而且实际上具有足够的精度。但是，这一成功在很大程度上归功于基于现场观测和卫星遥感获得的大量数据的模型的参数化，这并不一定意味着成功地阐明了与海浪谱演变有关的所有基本机制。本项目旨在通过与大尺度直接数值模拟的对比研究，探讨当前海浪预报模式中所包含的各种假设和假设的有效性。本文首先对无风输入和破碎耗散的波浪场进行了一系列直接数值模拟，以检测各分量波之间非线性相互作用引起的波谱变化，结果表明，直接模拟得到的波谱变化率与波浪预报模型的预测结果吻合较好。这显然支持了波浪预报模式的有效性。然而，在相当短期的确定性演化中观测到的波谱变化率在每一时刻的变化，似乎对目前所接受的波浪预报模式的非线性项的推导方式提出了一些疑问，有必要进行进一步的研究。关于破断耗散，我们新建立了一个简单而有效的模型，可用于确定性直接数值模拟。将该模型应用于具有JONSWAP或P-M谱的随机波场时，发现该模型产生的耗散谱与WAM模型所采用的标准模型相当吻合。目前正在将该模型应用于各种研究，如破碎对波高分布的影响。作为这项研究的副产品，我们还研究了反常波现象，并获得了各种有趣的结果。

It has become possible to perform routinely numerical wave forecast with practically sufficient accuracy. But much of this success owes the parameterizations of the models based on huge amount of data acquired by in situ observations and satellite remote-sensing, and it does not necessarily mean the success of elucidating all the basic mechanisms that are relevant to the evolution of the ocean wave spectrum. This project aims at investigating the validity of various assumptions and hypothesis included in the present-day wave forecasting models through comparative study with large-scale direct numerical simulations.We first performed a series of direct numerical simulations of the wave fields without input from wind or dissipation due to breaking to detect the change of spectrum due to nonlinear interactions among component waves, and found that the rate of change of the spectrum which the direct simulations gives agrees quite well with the prediction by wave forecast models. This apparently supports the validity of the wave forecast models. However, the fact that the rate of the change of the spectrum was observed at every instant of time in a rather short-term deterministic evolution seems to throw some doubt on the way of derivation of the nonlinear term of the wave forecast models which are currently accepted, and further investigation should be necessary.With regard to the dissipation due to breaking, we have newly developed a simple but efficient model which can be built in the deterministic direct numerical simulations. When applied to random wave fields with JONSWAP or P-M spectrum, the model was found to generate a dissipation spectrum which agrees fairly well with the standard model employed by the WAM model. Application of this model to various studies, such as the effect of breaking on the wave-height distributions is now underway. As a byproduct of this study, we also investigated the phenomenon of freak waves and obtained various interesting results.