Evaluating the impact of atmospheric forcing resolution and air-sea coupling on near-coastal regional ocean prediction

Atmospheric forcing applied as ocean model boundary conditions can have a critical impact on the quality of ocean forecasts. This paper assesses the sensitivity of an eddy-resolving (1.5 km resolution) regional ocean model of the North-West European shelf (NWS) to atmospheric forcing resolution and air-sea coupling. The analysis is focused on a month-long simulation experiment for July 2014 and evaluation of simulated sea surface temperature (SST) in a shallow near-coastal region to the south-west of the UK (Celtic Sea and western English Channel). Observations above and below the sea surface at the L4 ocean buoy from the Western Channel Observatory are used to evaluate ocean and atmosphere model data. The impacts of differences in the atmospheric forcing are illustrated by comparing results from an ocean model run in forcing mode using operational global-scale numerical weather prediction (NWP) data with a run forced by a convective scale regional atmosphere model. The value of dynamically representing feedbacks between the atmosphere and ocean state is assessed through use of these model components within a fully coupled ocean-wave-atmosphere system. Simulated SST show considerable sensitivity to atmospheric forcing and to the impact of model coupling in near-coastal areas. A warm ocean bias relative to in-situ observations in the simulation forced by global-scale NWP (0.7 K mean difference, warmer relative to all observations in the model domain) is shown to be reduced (to 0.4 K) through use of the 1.5 km resolution atmosphere forcing. When simulated in coupled mode, this bias is further reduced by 0.2 K. Results demonstrate much greater variability of both surface energy balance terms and near-surface winds in the higher resolution atmosphere model data, as might be expected. Assessment of the surface energy balance and wind forcing over the ocean is challenging due to a scarcity of observations. It can however be demonstrated that the wind speed over the ocean simulated by the high resolution atmosphere agreed with the limited number of observations less well than the global-scale NWP data. Further partially-coupled experiments are discussed to better understand why the degraded wind forcing does not detrimentally impact on SST results.