Sensitivity of near-inertial internal waves to spatial interpolations of wind stress in ocean generation circulation models

The oceanic near-inertial internal waves (NIWs) have been extensively studied using ocean general circulation models (OGCMs). Currently most OGCMs use the bilinear or bicubic interpolation to interpolate wind stress onto models' surface grids. In this study, we examine the influences of bilinear and bicubic interpolations on the wavenumber and frequency spectra of wind stress and on the simulated NIWs in the ocean. It is demonstrated that both the bilinear and bicubic interpolations are equivalent to spatial low-pass filters with the former leading to more significant loss of wind stress variance at high wavenumbers. When coarse (e.g., 2 degrees) wind stress is used to force OGCMs, the bilinear and bicubic interpolations significantly damp the wavenumber spectrum of wind stress at mesoscales, leading to decreased near-inertial wind stress variance. Using the bilinear (bicubic) interpolation could weaken the near-inertial wind work by similar to 43% (22%) in the subtropical region (10 degrees N-30 degrees N) and by similar to 16% (4%) at the midlatitudes (30 degrees N-50 degrees N). We propose a new interpolation method, i.e., the bi-sinc-function interpolation, which is able to retain all the wind stress variance within the Nyquist wavenumber. Compared to the bilinear and bicubic interpolations, the bi-sinc-function interpolation improves the simulations of NIWs and should be incorporated into OGCMs especially when coarse wind stress is used. (C) 2015 Elsevier Ltd. All rights reserved.