The contribution of sensitivity analysis and data assimilation to tidal-stream resource assessment: The example of the Alderney Race

Highly accurate modeling of tidal current is crucial in the assessment of tidal-stream resource. We use a method for refining the predictions of a two-dimensional tidal model applied to estimate the resource of the Alderney Race (English Channel). The first phase consists in identifying the input parameters which have the greatest influence on the model’s performance. The sensitivity analysis relies on the Sobol’ indices. It is found that the adjustment of the phase of the M 2 (lunar semidiurnal) constituent is the most efficient way to reduce model’s errors. Bottom friction and amplitudes of M 2 and S 2 (solar semidiurnal) are also influential on model performance. The second phase relies on data assimilation. The 3DVar algorithm is used to adjust the calibration parameters so that they reduce the errors between model predictions and depth-averaged current speed measured by ADCPs. The method, although common in other coastal engineering applications, is tested in the context of estimating tidal resource. It gives promising results as it reduces root mean square errors in current speed by 19%. This corresponds to a reduction of root mean square error in power density of 31% at the ADCP locations. By comparing maps of power density before and after the model calibration, we show that the adjustment of model parameters significantly modifies the resource assessment in the Alderney Race. • Parameters most influencing tidal model performance are identified by Sobol indices. • Data assimilation adjusts model parameters and significantly reduces errors. • Data assimilation is used in the rare context of tidal resource assessment. • Improved model calibration strongly impacts tidal resource assessment.