Improvement in wave forecast using data assimilative coastal wave model

Abstract

Attempt has been made to improve wave forecast using a data-assimilative coastal model. Synthetic ocean wave spectra have been computed using scatterometer forced coastal wave model SWAN. The spectra have been assimilated in the same model forced by ETA model wind from the National Center for Medium Range Weather Forecast (NCMRWF). Model runs have been made for the Indian coastal region. Assimilation techniques used are direct insertion and blending. The present study reveals that the model performance has improved both for the significant wave height and swell height during the assimilation period and the forecast period. Direct insertion performs better than blending. Improvement is more in the case of slow moving low frequency swell as compared to the wind-generated wave. Impact of the assimilation remains significant for about 2 days in the case of total significant wave height, while for the swell; the memory of assimilation is about 3 days.