The team constructed a large-scale two-dimensional finite element model in ABAQUS, using borehole data to capture the real geometry of the seabed and its underlying strata. At three representative boreholes, labeled ZK1, ZK3, and ZK6, they extracted matching one-dimensional soil columns for direct comparison. Input motions were drawn from far-field bedrock recordings of the Japanese KiK-net strong-motion network, and the soil was represented with a non-Masing DCZ hysteretic constitutive model, which captures the nonlinear stiffness degradation and energy dissipation that real soils exhibit under strong cyclic loading.

The simulations revealed clear spatial patterns. At ZK1, relatively large horizontal surface accelerations appeared, which the authors associate with nearshore topographic variation and complex local geological conditions rather than with any single factor. The vertical component of motion proved especially sensitive to site geometry.

When horizontal and vertical excitations were applied together, the ratio of vertical peak ground acceleration between the two-dimensional and one-dimensional models reached 2.15 to 2.92 at ZK3. In other words, the one-dimensional approach underestimated vertical response by factors of roughly two to three at that location, indicating that horizontal-vertical interaction is amplified where bedrock undulates and sediments vary laterally. Frequency-domain and response-spectrum analyses added a spectral dimension to the picture.

The combined site conditions amplified low-frequency motion below 1 hertz and produced localized amplification around 1 to 2 hertz. The normalized acceleration response spectra also showed shifted predominant periods and multiple peaks, meaning the site does not simply magnify the input motion uniformly but redistributes energy across frequencies in ways that depend on the full two-dimensional geometry. The authors draw a practical boundary for model selection.

Where the bedrock surface is relatively flat and sediment layers are nearly horizontal, a one-dimensional model may still give useful preliminary information about the frequency characteristics of horizontal seismic response. But where the bedrock undulates and sediments change laterally, as at ZK3, two-dimensional modeling should be considered for a more reliable seismic evaluation. The study is a numerical investigation grounded in real borehole data and recorded input motions, not an observational record of an actual earthquake at the site, and its conclusions are specific to the Hangzhou Bay geometry and the modeled input motions.

Subject of Research: Nonlinear seismic site response of the Hangzhou Bay seabed under combined topographic, sedimentary, and bedrock conditions

Article Title: Effects of comprehensive site conditions on the nonlinear seismic response of the Hangzhou Bay seabed site

Article References: Jiang, Z., Miao, H., Wang, Y., Liu, J., Wang, B., & Chen, G. (2026). Effects of comprehensive site conditions on the nonlinear seismic response of the Hangzhou Bay seabed site. Frontiers in Marine Science, 13, Article 1892853. https://doi.org/10.3389/fmars.2026.1892853

Image Credits: AI Generated

DOI: 10.3389/fmars.2026.1892853

Keywords: Hangzhou Bay, seismic site response, nonlinear soil behavior, two-dimensional modeling, ABAQUS, KiK-net, bedrock undulation, marine sediments, ground motion amplification, earthquake engineering, seabed topography, response spectrum