Impact of synthetic abyssal hill roughness on resolved motions in numerical global ocean tide models

Abstract

Global models of seafloor topography have incomplete and inconsistent resolution at horizontal wavelengths less than about 10-20 km, notably due to their inability to resolve abyssal hills in areas unsurveyed by ships (that is, about 90% of the global seafloor). We investigated the impact of this unresolved bottom roughness on global numerical simulations of the HYbrid Coordinate Ocean Model (HYCOM) that are forced exclusively by the M2 and K1 internal tides. Simulations were run with horizontal resolutions of 0.08 °and 0.04 °, 10 isopycnal layers in the vertical direction, and two versions of bathymetry: one derived from the SRTM30_PLUS global bathymetry model, and one merging SRTM30_PLUS with a synthetic fractal surface simulating the expected roughness of abyssal hills in the 2-10 km horizontal wavelength band. Power spectra of the two bathymetry versions diverge at wavenumbers of order 4 * 10-4 radians/m and higher (wavelengths of order 15 km and lower), with more pronounced differences evident on the 0.04 °grid, as the 0.08 °grid has a more limited ability to capture bathymetric details at the abyssal hill scale. Our simulations show an increase in the amount of kinetic and potential energy in higher vertical modes, especially in the 0.04 °simulation, when the synthetic roughness is added. Adding abyssal hills to the 0.04 °simulation increases the M 2 kinetic energy for modes 3 and 4 by 12-18% and the potential energy by 5-15%. Adding abyssal hills to the 0.08 °simulation yields a reduced, though still measurable, impact on simulated baroclinic tidal energies. Baroclinic tidal energy conversion rates increase by up to 16% in regions of high roughness, and by up to 3.4% in the global integral. The 3.4% increase in global conversion rates in the numerical simulations is less than the 10% increase computed from a linear analysis on a 0.008 °grid because of the resolution limitations of the numerical simulations. The results obtained in the present study, though limited by the horizontal and vertical resolutions of the simulations, are consistent with those of previous studies indicating that abyssal hills on the seafloor transfer energy into higher vertical mode internal tides. The method employed here to add synthetic roughness could easily be replicated in other models, with higher resolution and/or more complex forcing.

The Global Mode‐2 M2 Internal Tide

“…Although mode‐1 internal tides are mainly generated by steep topographic features (Egbert & Ray, , ; Zhao et al, ), higher‐mode internal tides may be generated by the wide‐spreading small‐scale bottom features. The small‐scale gentle topographic features such as abyssal seamounts are ubiquitous in the ocean (Lefauve et al, ; Timko et al, ; Zhang et al, ). My satellite results show that these small‐scale topographic features are prone to generate higher‐mode internal tides (mode‐2 in this paper).…”

A Primer on Global Internal Tide and Internal Gravity Wave Continuum Modeling in HYCOM and MITgcm

“…In Timko et al (2017), the statistical abyssal hill roughness was employed in HYCOM simulations of the internal tides, set up under the same simplified conditions used in Arbic et al (2004) and Simmons et al (2004b), to examine the extra explicitly resolved internal tide activity arising from the small-scale roughness missing in global bathymetric datasets. Timko et al (2017) found that the roughness does indeed increase internal tide generation and energy levels, especially in the higher vertical modes that are generated by smaller-scale topography (St. Laurent and Garrett, 2002;Simmons et al, 2004b). Fig.…”

“…13.42 shows the small-scale roughness that we introduce through the Goff and Arbic (2010) and Goff (2010) work, as it appears once filtered down to 1/12.5 horizontal resolution. The method employed by Timko et al (2017) to add statistical roughness to a global internal tide model could be adapted to researchers studying other phenomena, in other models, and with different grid resolutions. In fact, in regional models with higher resolutions than the 1/12.5 and 1/25° resolutions used in Timko et al (2017)'s global models, the effects of the statistical roughness will be greater because as resolution increases the radius of the smoother that must be applied to put the raw statistical bathymetry onto a grid will decrease.…”

“…Substantial analysis was also done at NRL, and most of the HYCOM simulations were performed there. All of the HYCOM simulations reported on here, except for those used in Timko et al (2017), were supported by grants of computer time from the Department of Defense (DoD) High Performance Computing Modernization Program at the Navy DoD Supercomputing Resource Center. MCB, EJM, HEN, JGR, JFS, IS, AJW, and LZ acknowledge support from the ONR projects "Dynamics of the Indonesian Throughflow and its remote im impact", "Eddy resolving global ocean prediction including tides", "Global and remote littoral forcing in global ocean models", "Ageostrophic vorticity dynamics of the ocean", "NCOM-4DVAR, A multiscale approach for assessing predictability of ASW environment", "Extending predictability in coastal environments", "HYCOM global ocean forecast skill assessment", and the ONR and National Ocean Partnership Program (NOPP) sponsored project "Improving global surface and internal tides through two-way coupling with high resolution coastal models".…”

“…We thank Ayan Chaudhuri for helping with inclusion of tidal forcing in MITgcm. PGT and BKA acknowledge the Texas Advanced Computing Center (TACC) at The University of Texas at Austin for providing HPC resources that have contributed to the research results reported in Timko et al (2017). URL: http://www.tacc.utexas.edu.…”