Improving knowledge of near‐surface structure—i.e., structure of the shallowest crust, <1 km—is crucial to numerous areas of Earth science, including seismic-hazard assessment, earthquake science, and diverse surface processes.  However, constraining near-surface structure remains challenging, as conventional seismic tomography techniques have poor sensitivity and resolution at shallow depths.  Drilling or local seismic surveys are performed to image the near surface but provide extremely limited coverage of sites worldwide.  In this talk, I introduce a new computationally-efficient algorithm that determines near-surface seismic wave speeds.  The algorithm is based on theory that relates the seismic wave speed to the orientation of arriving seismic waves.  Applying this method, I generate maps of shear and compressional wave speeds for Japan, using seismic data recorded by the High-Sensitivity Seismograph Network.  The maps show that sedimentary basins and mountainous regions are characterized by low and high wave speeds, respectively.  Wave speeds also vary with thermal features, such as volcanos, and geological units of different ages.  Furthermore, I obtain wave speeds for different depths by analyzing different frequencies.  Lastly, I apply the frequency-dependent analysis to the United States to explore structure from tens of meters to several kilometers, bridging the gap between the Earth’s surface and deep tomographic images.  This versatile technique can serve as a new path to reliable and non-invasive shallow seismic imaging and hazard assessment at any part of the world with a single seismic instrument.  It can also provide a means of monitoring changes within the very upper crust caused by volcanic or hydrological processes.