The average shear-wave velocity in the first 30 m of subsoil, Vs30, is a key indicator of site response affecting the ground-motion amplification for many earthquake-engineering applications. It is, therefore, of great importance to estimate Vs30 accurately, especially for significant civil projects. This paper presents a framework that utilizes multiscale random field models incorporated with spatial dependence of soil properties to estimate Vs30 at any location neighboring measured data with the secondary information at hilly area (U.S. Geological Survey global Vs30 map that is generated based on the topographic slope). The capability to consistently refine and generate estimates across different scales, from regional scale all the way to local site-specific scale, makes the developed framework able to provide high resolution estimates within any specified regions. Monte Carlo simulations are performed and a multiresolution map of Vs30 at Suzhou site is generated. The new map then applied to site classification. Feasible results obtained.