Search

The examination of currents by merging model results and radial velocity High-Frequency (HF) radar data has been undertaken in the Sunda Strait, which links Sumatera Island and Java Island, involving two coastal sites (Anyer and Labuan) and using the Ensemble Transform Kalman Filter (ETKF). Dependent validation involved the data used during model analysis while independent validation utilised observations from a different site. These validations are needed to obtain an optimal representativity error, which has the lowest averaged root mean square (RMS) value over time and is appropriate for all sites. Moreover, the optimal representativity error was evaluated using the relative error reduction and the associated skill score metric. The results show that the model analysis for both independent and dependent validation has better results than the model with no blending. Interestingly, independent validation has a smaller RMS value than the model with no blending, although it is still greater than the dependent validation. The best results were obtained from model analysis of all sites, with the representativity error value being 0.2704 m / s for Anyer and 0.4459 m / s for Labuan. However, they have a pattern in the RMS error over time series. It is necessary to consider the factor such as winds that would modulate the magnitude of radial velocity. • Model accuracy is improved by blending HF radial velocity using the ETKF. • The representativity error is optimized by using independent validation. • Every site has a different optimal representativity error. • Improved model analysis can be obtained from either the original HF radar data or the HF radar data obtained at a different site. • Every site has a different optimal representativity error. • The root mean square error values of the currents are modulated by the strength of the winds. [ABSTRACT FROM AUTHOR]