Relationships between wind predictions and model resolution over coastal regions

The impact of resolution on wind predictions within regions of complex coastal geometry is evaluated using a quadruple nest of COAMPS® (27km to 1km) to find an optimal configuration of spatial and temporal resolution. Two regions, Turkish Straits System and Chesapeake Bay, are selected because of their diverse coastal environments, the availability of wind observations and to determine if the relationships between resolution and wind prediction accuracy would be valid for geographically different regions. The coarse resolution model successfully simulates the general trend of the surface wind variation, but cannot capture peak events accurately. Increased spatial resolution results in more accurate wind predictions. The coastline representation and land features impact friction over land and blocking of the winds and affect accuracy of wind predictions. 27-km resolution products lack important details over coastal waters and are not adequate to force high resolution ocean models. No evident improvement in accuracy is observed when increasing the resolution from 3-km to 1-km. An increase in frequency of the wind records from 3-hourly to hourly is required to capture frontal events with strong wind speeds and sharp gradients. Our analysis for both regions suggests the use of hourly atmospheric products at 3-km resolution for oceanic forcing purposes. •Wind predictions at 4 resolutions are compared using a nested grid atmospheric model.•Coarse resolution grid lack complex land features causing less accurate predictions.•Better representation of coastline is required to improve accuracy in coastal waters.•3-km resolution is found to be more optimum than 1-km for coastal wind predictions.•The use of hourly predictions is strongly needed to accurately capture storm fronts.