Wave resource characterization is an essential step for wave energy converter development in the ocean. However, accurate and detailed resource characterization at a regional scale poses a great challenge because of the requirements for high model grid resolution, extensive model validation, and a high-performance-computing resource. This study presents a multi-scale, multi-resolution approach using the WaveWatchIII and Simulating WAve Nearshore (SWAN) wave models to provide accurate long-term wave hindcasts with a spatial resolution of approximate 300 m in the nearshore region on the U.S. West Coast. Extensive model validation for the six wave resource parameters recommended by the International Electrotechnical Commission, bivariate histograms, and frequency-directional spectra distributions were conducted using a set of model performance metrics and measurements from 28 wave buoys along the West Coast. Model skills in simulating large waves under extreme storm events were also evaluated. Model results showed that the high-resolution SWAN model is able to accurately simulate the wave climate on the West Coast, especially in the nearshore region. This study also demonstrates that the multi-scale and multi-resolution modeling framework is an efficient approach for generating accurate long-term, high-resolution wave hindcasts for wave resource characterization at the regional scale. •A multi-scale, multi-resolution wave modeling system for the U.S. West Coast.•Unstructured-grid SWAN model with 300 m resolution for the nearshore region.•Extensive model validation using up to 32 years of wave data at 28 buoys.•Good model skills in simulating spatial and temporal variabilities of wave climate.•High-resolution modeling system improved model accuracy in large wave prediction.