Fate of internal solitary wave and enhanced mixing in Manado Bay, North Sulawesi, Indonesia

Abstract

The Sulawesi Sea is one of the most active regions where internal solitary waves (ISWs) are frequently observed through satellite images. These waves that result from the nonlinear evolution of internal tides are generated over the Sibutu Sills and propagate over a long distance, up to 700 km before shoaling in coastal areas. To date there is yet no specific in-situ observation or numerical studies dedicated to characterize their dynamics. We present results from the combination of two simulations and in-situ observations which allow to describe the full life cycle from generation to breaking of ISWs in the Sulawesi Sea. A first set of 2D non-hydrostatic numerical simulations reproduces the generation of ISW from the non-linear steepening of the internal tides. The tidal energy flux emanated from the generation site is of about 15 kW m−1 with progressive decrease in the along flux direction resulting the growth of ISWs. The ISW energy flux increases up to large fraction of 32% of the total energy flux at 700 km from the generation point while the M2 internal tide contributes to 64% of the total energy flux. A second simulation at higher resolution focusing on the shallow coastal area of the Manado Bay reproduces the breaking of the ISWs at the shelf slope and was compared with in-situ observations. About 3.5% of the energy contained in an ISWs packet is dissipated over a 3 km wide cross shelf region of the Manado Bay in the simulation. Both simulations and observations revealed the presence of energetic boluses identified as a sequence of high-frequency waves of elevations in the stratified bottom layer. A high level of turbulence was diagnosed in the observations and simulations from the Thorpe Method. The mean eddy kinetic energy dissipation rate and vertical diffusivity in the observation reaches as 10−6 W kg−1 and 10−3 m2 s−1, respectively; while the dissipation rate is one order of magnitude higher and the vertical diffusivity is one order lower in the model. This discrepancy likely reflects the sparse sampling in the observations as well as differences in the stratification and the absence of surface forcing in the model.