The time-averaged shear-wave velocity in the top 30 meters of soil and rock, VS30, is the most commonly used site-effect proxy for ground-motion prediction equations (GMPEs). This article presents a nonlinear site amplification model, referred to as the 1D TVS30 model, which uses VS30 as a site-effect proxy for the horizontal component of motion for GMPEs, based on seismic data from Japan. The site amplification ratios were from Hou and Zhao (2022) using site period TS, and the site shear-wave velocity profiles were also from Hou and Zhao (2022) based on the selected KiK-net and K-NET networks. The data distribution characteristics show that 1D site amplification ratio data could compensate well for the scarcity of empirical data at strong shaking levels, and provide a more confident constraint on model development. The 1D TVS30 model in this study has a larger between-site standard deviation and slightly smaller within-site standard deviation than the 1D TS model that we published previously. The predicted site amplification ratios from the two models are close for sites with shallow soil but are significantly different for sites with deep soil. The method proposed in our previous study is recommended to implement the 1D TVS30 model into GMPEs. We also performed an informal test to validate our model. This matches well with ground-motion records with moderately or strong nonlinear site response in the NGA-West2 and NGA-Subduction datasets, especially for the NGA-Subduction records at the spectral periods up to 1.0s. Further comparison with other published models suggests that the nonlinearity of our model is moderately stronger than, or close to that of the NGA-West2 and NGA-Subduction models and is weaker than that of an existing model for Taiwan.