The Assimilation of Temperature and Salinity Profile-Observations For Forecasting the River-Estuary-shelf Waters

Current down-scaling numerical modeling system around Hong Kong achieved a considerable prediction skill for the estuarine-shelf circulation off the Pearl River Estuary without data assimilation (DA). In order to further improve the reliability of this modeling system, the cost-effective Ensemble optimal interpolation approach is implemented to test the potential benefits from assimilating the cruise temperature and salinity (T/S) profiles to reproduce the variable coast waters. Regarding assimilation parameters (e.g., assimilation window, observation spatiotemporal scales, and ensemble composition), four parallel experiments are conducted in summer 2015. Against the assimilated T/S profiles, the vertical structures of the analyzed T/S are improved by the DA, although the waters experience strong mixing on the shelf. Compared with the run without DA, the root mean square errors of the predicted T/S are generally reduced by 9.8%-23.5% and 4.2%-14.0% in the assimilation runs. The results also show the salinity stratification is improved in the shelf by the assimilation of T/S profiles, although the improvement is sensitive to the selected ensemble and the assimilation window. Further, we investigate the impact of the temporal scales of the river-estuary-shelf (RES) waters on the assimilation results by the sampling of the model-state ensemble. The water exchanges between the estuary and the shelf are also better captured through this assimilation. The assimilation impact analysis shows that DA has advantages in reproducing the distribution of water masses of the RES waters, although the quality of the reproduced water mass distribution is related to the adopted sample ensemble in DA. Plain Language Summary The river-estuary-shelf waters of the Pearl River Estuary have high spatiotemporal variability. In this research, we use a high-resolution model with data assimilation to obtain a reliable prediction of these waters. The temperature and salinity profiles are assimilated in the current system to test the potential benefits from the cruise profiles to reproduce the variable coast waters. Our experiments show that the prediction skill of temperature and salinity is improved by data assimilation. The water exchanges between the estuary and the shelf and the water mass distribution of the study regions are well captured through this assimilation.

Automated summary

The current high-resolution numerical modeling system around Hong Kong has achieved considerable prediction skill for the estuarine-shelf circulation off the Pearl River Estuary without data assimilation. By implementing Ensemble Optimal Interpolation approach to assimilate cruise temperature and salinity profiles, the prediction errors of temperature and salinity have been reduced and salinity stratification in the shelf has been improved. Water exchanges between the estuary and the shelf are better captured through data assimilation, showing advantages in reproducing the distribution of water masses in the study regions.