Variation of shaking intensity for the 18 April 1906 earthquake in and around the city of San José, California, is established based on 607 damage locations, using a methodology suitable for high-resolution mapping of intensity. This detailed methodology uses a refined modified Mercalli intensity (MMI) scale and is based on the premise that an intensity estimate can be calculated from the damage to a single building, provided extent of damage and style of building construction are sufficiently well known. To reflect the rich detail in the damage descriptions, the refined scale recognizes partial steps in the range MMI 6–9.
For the 138 km2 area of study, earthquake damage is compiled from published descriptions, mostly made by building-industry professionals within 10 days after the 1906 earthquake and never before evaluated for an intensity study. Sanborn–Perris Map Company fire insurance maps provide precise location and nature of construction for 483 damaged structures within, and 124 outside, the 1906 San José city limits. The distribution of assigned intensities approximates a normal distribution with median intensity 7.5, mean intensity 7.36, and standard deviation 0.687. Maximum intensity (8.5–9.0) is identical to contoured MMI from previous intensity studies with lower density of data.
For analysis, damage data points are grouped at two scales. At the first, 227 modern census blocks contain as many as 31 buildings each, with mean of block median intensities 7.46 and standard deviation 0.660. In 71% of 101 blocks with two or more data points, intensity range is one point or less. This low variability indicates consistent building response in these blocks to ground shaking. Moderate to larger variations in intensity within some blocks may be attributed to differences in construction, prior damage to structures, or anthropogenic factors.
At the second scale, geographically distinct associations of neighboring blocks are grouped into six microzones from 1.5 to 7.0 km across, with mean microzone intensity 6.6 to 8.2 and standard deviation 0.160 to 0.637. These intensity differences, which match variations observed in modern ground-motion studies, are inferred to reflect real variation in damage-causing ground motion in 1906. Methods presented here can be used to constrain ground-motion models, identify seismic hot spots, and improve site-specific assumptions for probabilistic seismic hazard analysis (PSHA).