Abstract: Based on the Regional Oceanic Modeling System (ROMS) and multi-scale three-dimensional variational data assimilation scheme, the sea surface currents (SSCs) observed by double HF Radar stations Bohe and Doulong are assimilated into the ocean model. The effect of assimilation using different observation coverages and the sustainability of assimilation effect using different assimilation intervals are investigated through sensitivity experiments. The results show that the assimilation of HF Radar SSC significantly improves the SSC simulation. The experiments show no significant difference under different coverages of SSC observations when the coverage is over 40%, which indicates the background error covariance is able to make up for the missing data of the HF Radar SSC in space. The results also show that the sustainability of assimilation effect is influenced by the interval of assimilation: assimilating data too often will exacerbate the dynamic imbalance of the initial field, which makes the sustainability of the improvement worse, while a too large interval of assimilation will lose the assimilation effect before the next assimilation starts. The sustainability of the improvement in assisimilation achieves the best optimization when the assimilation interval is 6 or 12 hours. The sustainability of the improvement of model forecast after the assimilation lasts for 4 to 6 hours.

Key words: high frequency ground wave radar, three-dimensional variational assimilation, sea surface current assimilation, regional oceanic modeling system