The offshore wind industry in the United States is gaining strong momentum to achieve sustainable energy goals, and the need for observations to provide resource characterization and model validation is greater than ever. Pacific Northwest National Laboratory (PNNL) operates two lidar buoys for the U.S. Department of Energy (DOE) in order to collect hub height wind data and associated meteorological and oceanographic information near the surface in areas of interest for offshore wind development. This work evaluates the performance of commonly used reanalysis products and spatial approximation techniques using lidar buoy observations off the coast of New Jersey and Virginia, USA. Reanalysis products are essential tools for setting performance expectations and quantifying the wind resource variability at a given site. Long-term accurate observations at typical wind turbine hub heights have been lacking at offshore locations. Using wind speed observations from both lidar buoy deployments, biases and degrees of correspondence for the Modern-Era Retrospective Analysis for Research and Applications-2 (MERRA-2), the North American Regional Reanalysis (NARR), ERA5, and the analysis system of the Rapid Refresh (RAP) are examined both at hub height and near the surface. Results provide insights on the performance and uncertainty of using reanalysis products for long-term wind resource characterization. A slow bias is seen across the reanalyses at both deployment sites. Bias magnitudes near the surface are on the order of 0.5 m s−1 greater than their hub height counterparts. RAP and ERA5 produce the highest correlations with the observations, around 0.9, followed by MERRA-2 and NARR.