Study And Application Of Swan Model And Data Assimilation Technology In Wave Forecast

SWAN (Simulating WAves Nearshore) model (Booij et al 1999) was firstly introduced in NMEFC (National Marine Environment Forecast Center) at the end of twentieth century and several operational wave forecast systems such as Northwest pacific and global wave forecast have been developed based on SWAN model during the first decade of 21st century. One of the main purposes of this paper is to improve the operational wave model and wave forecast in NMEFC by numerical experiments. The second purpose, as the pre-study of altimetry significant wave height data of HY-2 satellite, is to evaluate the effect of assimilation of satellite altimetry significant wave height on three days'wave forecast. It will also contribute to the future operational applications of satellite altimetry data from HY-2 satellite.The process of wave energy dissipation is the least understood part of the physics of wave modeling. Despite so many theories and observations related to wave dissipation, people do not know exactly when and how waves lose their energy and where the dissipated energy to go. It is impossible to derive a rational wave dissipation function directly from fluid dynamical equation. In this paper a quasilinear wind wave dissipation function is proposed based on the study of Alves and Banner (2003) and Westhuysen (2007), in which the so called"two-phase"viewpoint is taken. The new dissipation function comprises saturation-based dissipation above 2 times of peak frequency and an improved"Whitecapping"dissipation otherwise. The saturation-based dissipation does not take the Phillips parameter as a constant, but varies with wave age. At the lower-frequency part of wave spectrum, the ideal experiment in this paper and other study (Banner and Young 1994; Rogers 2003) σandσ4 (σis wave angular frequency) and it also may be related to wave age. Those approaches are supported by numerical experiments of wind wave growth and spectral growth in this paper. There are two main improvements due to the new dissipation. Firstly it improves the simulation of sea mature; secondly it improves the simulation of higher-frequency spectrum which is very important to compute the characteristic quantity related to the high order moment of wave spectrum.The initial wave field is also an important factor for wave forecasting as we know, but it is very difficult to get observational data in the ocean. From 1980s, people can acquire enough near real-time data of ocean waves from satellite altimetry so that they can assimilate those data into wave models to improve their wave forecasts. The practical technology of satellite altimetry is being developed in our country now and we will soon obtain near real-time wave data from our own satellite altimetry. It will help us to improve forecasts of ocean waves. In this paper, a wave assimilation model based on optimal interpolation scheme (OI) is established after sufficient investigations of approaches about errors of background field and altimetry data from Topex/Poseidon and Jason-1. Reconstruction of wave spectra refers to the assimilation model in ECMWF, in which the background wave spectra are divided into Wind Sea and swell. For Wind Sea, the error of wind is believed to be the main cause of background error, so the satellite altimetry data are used to correct both the significant wave height and wind speed. The correction of frequency of swell spectra is acquired from the correction of significant wave height by supposing that assimilation does not change the steepness of swell.The assimilation wave model and improvements of model physics are all based on SWAN model (Booij et al 1999). Although SWAN model is designed for coastal application originally, many authors believe that it can be used in any scale relevant for wind-generated surface waves recent years (The SWAN team 2006). In this paper I try to apply the improved SWAN model for global wave forecasting. The everyday wave forecasts in 2009 are simulated using different parameterization configurations and the results are tested and compared with buoy and satellite altimetry wave data respectively. The results indicate that the new model physics can improve the forecasts of significant wave height and wave spectra. The satellite altimetry wave assimilation decreases the error of three days'wave forecast and increase the correlation coefficient significantly. It can be drawn a conclusion that the improved SWAN model can be used in the area of oceanic scale as well as coastal area.There are three innovations in this paper. Firstly, an improved dissipation function is proposed and tested in SWAN model, which improves the performance of three days'wave forecasting. Secondly, an assimilation module is developed and added to SWAN model, which makes it capable of simulating satellite altimetry wave data. Finally, the improved SWAN model is extended for global application successfully, which indicate that the SWAN model can be used for wave forecasting from coastal area to global.