dc.description.abstract | The discrepancies between model initial analysis and the true state may contribute to the factors for model prediction errors. To adjust the initial fields for reducing the discrepancies, many data, such as satellite observations, are available for assimilation, however still showing less impacts as compared to those from insertion of a bogus vortex. Using a four dimensional variational method (4DVAR), the bogus vortex can be effectively assimilated into the model to adjusts the initial field under constraints of model dynamics, which is the so-called bogus data assimilation (BDA). In this study, we address a new BDA method which adopts the better balanced vortex from MM5 4DVAR than the traditional simple Rankine vortex and then applies the three dimensional variational method (3DVAR) to assimilate this vortex data into WRF to investigate the impacts on track and intensity predictions of typhoons impinging Taiwan.
The target of this research is to evaluate the impacts of the new method on typhoon prediction. Two typhoons, Shanshan (2006) and typhoon Sepat (2007), were selected in this study, and they were simulated for 72 h. The results show that assimilation of the 3D wind of the virtual model vortex from 4DVAR in general have the largest improvement on typhoon simulation. Besides, the wind field tends to adjust to the mass field in 3DVAR and hence an unreasonable vortex structure may be produced when an unrealistic temperature from BDA has been assimilated as well by 3DVAR. The results also indicate that this assimilation tends to faster the typhoon movement, possibly due to the intensified 3DVAR vortex after the assimilation of the virtual vortex. For simulation of typhoon intensity, this approach may give significant improvement. Sensitive tests show this improvement, however, becomes much less when sea surface pressure was assimilated instead. Assimilation with the combined data (wind, temperature and moisture) in general does not lead to more consolidated improvement. | en_US |