The number in question is called the local dissipation ratio, written as q*. It describes the fraction of internal tidal energy that is dissipated locally, near where the waves are generated, rather than radiating away across the ocean basin. Internal tides are enormous internal waves generated when the surface tides, driven by the gravitational pull of the moon and sun, flow over ridges and seamounts on the seafloor. Globally, roughly one terawatt of tidal energy is converted into these waves, supplying about half of the mechanical power needed to maintain mixing in the abyss. That mixing, in turn, shapes the meridional overturning circulation, redistributes heat, and carries nutrients through the deep ocean. Because climate models cannot resolve internal tides explicitly, they must rely on parameterizations in which q plays a decisive role, and for decades most models have simply prescribed it as a uniform value of 0.3 everywhere.

The new work shows just how misleading that constant can be. Using output from the MITgcm LLC4320 simulation, a state-of-the-art ocean model running at a horizontal resolution of about two kilometers and forced by sixteen tidal constituents together with atmospheric reanalysis data, the team computed complete internal tidal energy budgets for two contrasting regions. The first is the Luzon Strait, the famous double-ridge passage between Taiwan and the Philippines that is one of the strongest internal tide generators on the planet. The second is the Nansha Islands, a sprawling archipelago of reefs and seamounts in the southern South China Sea that lies directly in the path of waves radiating from the strait.

In the Luzon Strait, the results largely confirm what oceanographers expected, but with important new detail. Barotropic-to-baroclinic conversion there reaches approximately 45 gigawatts, with semidiurnal constituents accounting for roughly 60 percent of that total, a consequence of resonance between the tide and the strait’s double-ridge topography. The local dissipation ratio for the total internal tide fluctuates between 0.3 and 0.7, and the team traced this variability primarily to the local breaking of modes 3 through 5, mid-to-high vertical modes whose energy stays close to the ridges. Mode 1, by contrast, radiates energetically away, carrying peak semidiurnal energy fluxes of up to 25 kilowatts per meter westward into the South China Sea and eastward into the Pacific. The q* values of around 0.5 are comparable to estimates of about 0.58 for the M2 internal tide at the Aleutian Ridge, suggesting that submarine ridge source regions worldwide may share a similar magnitude.

The Nansha Islands tell a radically different story. Local generation of internal tides there is feeble, less than 1.5 gigawatts, yet the region receives a substantial convergence of internal tidal energy flux arriving from afar. As a result, the local dissipation ratio generally exceeds 1 and occasionally surpasses 2.5, meaning the region dissipates far more energy than it generates. This is only possible because the numerator of q* includes energy delivered from the far field, not just local conversion. Modal analysis shows that the intensified dissipation originates predominantly from topographic scattering and breaking of mode-1 internal tides arriving from the Luzon Strait, with reflection from the southwestern shelf slope of the South China Sea furnishing an additional supply. When incident low-mode waves strike the steep reefs and seamounts, their energy is scattered into higher, shorter-wavelength modes that break readily, converting wave energy irreversibly into turbulence.

This distinction between source and sink regions is the conceptual core of the study. In a source region like the Luzon Strait, q* behaves like the conventional dissipation ratio q and stays between 0 and 1. In a sink region like the Nansha Islands, the incoming beams make q* exceed 1, with mode-1 diurnal values climbing above 4. The authors emphasize that scattering itself does not dissipate energy; it redistributes energy among modes and directions, and it is the subsequent breaking of the scattered high modes that removes energy from the wave field. Ignoring this spatial heterogeneity, previous work has warned, can cause an order-of-magnitude miscalculation of the mechanical energy supplying internal tidal mixing, degrading the ability of models to simulate water mass transformation and the strength of the global overturning circulation.

To translate these findings into tools that models can actually use, the team developed two complementary parameterizations of q*. The first is a physically grounded power-law formulation built from three variables: the internal tidal shear averaged over the bottommost 500 meters, the barotropic tidal kinetic energy, and the mode-1 internal tidal energy. Fitted to a 429-day time series, this formulation reveals opposite dynamics in the two regions. In the Luzon Strait, stronger barotropic forcing enhances local dissipation and raises q*, while an accumulation of mode-1 energy favors radiation of coherent beams to the far field and lowers it. Around the Nansha Islands the coefficients flip sign: weak local forcing combined with abundant far-field mode-1 energy drives the ratio upward, confirming that remote inputs, not local generation, sustain the extreme dissipation there.

The second parameterization takes a data-driven approach, using an eXtreme Gradient Boosting, or XGBoost, machine learning model trained on six physical features and fed with data from both regions. On the combined test set the model achieves an R-squared of 0.70, with 0.71 for the semidiurnal component alone, and it successfully separates the data into a low-q* cluster for the Luzon Strait and a high-q* cluster for the Nansha Islands. It also tracks the spring-neap periodicity of the ratio reasonably well. Performance degrades, however, when the model is asked to predict individual regions or tidal constituents separately, and the authors attribute this to the fact that extreme q* values depend on far-field beam interference and nonlinear wave-wave interactions that local, area-averaged input features simply cannot capture. Long-range internal tides from multiple sources form complex interference patterns, further modulated by mesoscale refraction, that smooth out in any purely local framework.

The study is candid about its limitations. At two-kilometer resolution the simulation resolves major topographic features and low-mode internal tides, but small-scale roughness, high-mode waves, and the breaking and turbulence that ultimately dissipate the energy remain subgrid processes. The dissipation term is inferred as the residual of the baroclinic energy budget rather than diagnosed directly, so it bundles together parameterized viscosity, bottom drag, and numerical damping, and its relative uncertainty grows where local conversion is weak, particularly over rough topography and for higher modes. Higher-resolution simulations with explicit dissipation diagnostics will be needed to separate these contributions.

Even so, the message for the modeling community is clear and potentially far-reaching. The comfortable assumption that a fixed fraction of 0.3 of internal tidal energy dissipates locally is wrong in both direction and magnitude across much of the ocean, and it fundamentally misses regions where dissipation is powered by waves generated hundreds or thousands of kilometers away. The authors argue that future tidal mixing parameterizations must move beyond the traditional framework based solely on local generation and explicitly incorporate non-local processes, including the far-field propagation, interference, and scattering of the internal wave field. As climate scientists work to refine projections of ocean heat uptake and the overturning circulation, the hidden lives of these underwater tides, and the distant reefs where they finally die, may prove to be a crucial piece of the puzzle.

Subject of Research: Local dissipation ratio of internal tides at the Luzon Strait and Nansha Islands in the South China Sea

Article Title: Local dissipation ratio of internal tides at key topographic features in the South China Sea

Article References: Yuan, Z., Sun, H., Yang, Q., Han, F., & Li, J. (2026). Local dissipation ratio of internal tides at key topographic features in the South China Sea. Ocean Science, 22(5), 2957-2972. https://doi.org/10.5194/os-22-2957-2026

Image Credits: AI Generated

DOI: 10.5194/os-22-2957-2026

Keywords: internal tides, South China Sea, Luzon Strait, Nansha Islands, tidal dissipation, diapycnal mixing, ocean modeling, parameterization, MITgcm LLC4320, topographic scattering, mode-1 internal tide, XGBoost