The　imaging　of　subsurface　soil　velocity　structures　from　ambient　noise　inversion　is　a　difficult　problem.　Few　recording　points　and　a　simplified　1-D　layered　profile　lead　to　important　nonuniqueness.　From　our　point　of　view,　improving　the　reliability　of　processing　methods　of　the　observed　data　to　obtain　noise　horizontal-to-vertical　spectral　ratio　(NHV)　curves　and　setting　a　complete　model　parameter　space　are　important　tasks　to　reduce　the　non-uniqueness　of　inversion.　In　this　study,　using　a　local　site　near　the　border　of　the　Tonghai　Basin,　China,　as　a　case　study,　we　first　demonstrate　how　to　identify　and　mitigate　the　influence　of　industrial　sources　using　surface　observations　to　obtain　more　reliable　NHV　curves.　Then,　a　new　strategy　to　determine　model　parameter　space　is　proposed,　that　is,　stratifying　soil　layers　based　on　the　number　of　NHV　peaks　and　determining　the　shear　wave　velocities,　thicknesses,　and　their　ranges　based　on　the　empirical　relationship　between　sedimentary　thickness　and　resonant　frequency　(h-fr).　Subsequently,　combining　the　model　parameter　space　acquisition　strategy　with　the　NHV　inversion,　a　novel　NHV　inversion　approach　is　developed　and　applied　to　obtain　the　2-D　VS　profile　of　the　investigated　Tonghai　site.　The　inverted　2-D　VS　profile　aligns　favorably　with　the　frequency-depth　conversion　results　of　the　measured　NHV　curves　(NHV-profiling)　and　the　measured　borehole　profiles,　affirming　the　reliability　of　the　proposed　NHV　inversion　method.　Finally,　by　comparing　the　empirical　transfer　functions　from　the　strong-motion　recordings,　we　validated　the　applicability　of　the　inverted　models　for　characterizing　site　effects.　The　model　parameter　space　acquisition　strategy　proposed　in　this　paper　and　the　analysis　procedure　of　the　observed　data　are　also　applicable　to　other　study　areas,　which　can　provide　a　referable　approach　to　quickly　and　effectively　acquire　the　soil　layer　velocity　structure　of　the　site.