A U.S. national soil velocity model based on machine learning of geospatial information: National map products of Vs2.5, k, and n

This dataset contains a suite of raster files (geotiffs) which comprise a U.S. national soil velocity model and a README file. The soil velocity model is developed using geospatial machine learning of shear wave velocity measurements and is implemented at ~90m resolution across the conterminous U.S. The resulting model provides a practical, scalable “geotechnical layer” complementing the U.S. Geological Survey National Crustal Model and supporting applications such as physics‐based ground motion simulations and coseismic hazard assessments that require continuous characterization of near-surface velocity structure across the U.S. The README file outlines the file and data structures, and provides the equations needed to implement the soil velocity model. Model users are referred to the journal publication provided in this project description for details regarding model development, performance, limitations, and use:

Sanger, M.D. and Maurer, B.W. (2026). “A U.S. National Soil Velocity Model Based on Machine Learning of Geospatial Information.” Journal Publication In Review.