What it measures. A probabilistic forecast of future ground shaking from all earthquake sources combined — not any single quake, and not a prediction of when one will strike. The USGS 2025 NSHM expresses it as Peak Ground Acceleration (PGA, in %g) at a fixed probability — here, shaking with a 2% chance of being exceeded in 50 yrs (≈ a ~2,475-yr average return period). Darker = stronger expected shaking = higher building-design demand.

Magnitude vs. intensity. Richter / moment magnitude (M) is the size of a quake at its source — one number per event. Modified Mercalli Intensity (MMI, I–XII) is the shaking actually felt at a place — it varies across the map with distance, depth and local soil. This layer's PGA corresponds to MMI (what you'd feel & the damage), not to magnitude.

MMI	Shaking / typical damage	PGA %g
I	Not felt	<0.2
II–III	Weak — felt indoors	0.2–1.4
IV	Light — dishes/windows rattle	1.4–3.9
V	Moderate — felt by all, minor breakage	3.9–9.2
VI	Strong — slight damage	9.2–18
VII	Very strong — moderate damage	18–34
VIII	Severe — considerable damage	34–65
IX	Violent — heavy damage	65–124
X+	Extreme — very heavy to total	>124

Magnitude→intensity isn't fixed (it falls off with distance & depth), but as a felt anchor near the epicenter: M3–4 ≈ MMI III–IV; M5–6 ≈ MMI VI–VII; M6–7 ≈ MMI VII–IX (the 2020 SW-PR M6.4 reached ≈ VII–VIII); M7+ ≈ MMI VIII–X+ (the 1918 PR M7.3). PGA↔MMI bins per USGS ShakeMap (Wald et al. 1999).

What it measures. Vs30 is the average speed of a seismic shear-wave through the top 30 m of ground (m/s). It's the standard proxy for how much soft ground amplifies earthquake shaking: slower Vs30 = softer soil = stronger shaking at the surface for the same quake. Classes run 1 (hard rock, least amplification) to 6 (Vs30 ≤180 m/s, NEHRP class E, most amplification). It describes shaking amplification only — not liquefaction, landslide or fault-rupture risk.

How it's built. The base grid is the USGS Global Vs30 model (Heath et al. 2020), which estimates Vs30 from topographic slope (steep→rock→fast; flat→sediment→slow) at ~700 m resolution. In "geology-conditioned" mode that estimate is capped using mapped PR surface geology (USGS OFR-98-38): Quaternary alluvium, artificial fill and beach deposits are pinned to published soft-soil Vs30 ceilings, while hard rock (≥500 m/s) is left ungated. 49 measured field points are available for cross-check (calibration against them didn't improve fit, confirming residual gaps are resolution-driven, not threshold-driven).

Limits. It is a model, not a site measurement — the ~700 m grid can't resolve narrow coastal soft-soil strips, and it is no substitute for a borehole Vs profile. Ocean is masked.

What it is. Liquefaction is when strong earthquake shaking makes loose, water-saturated soil briefly behave like a liquid: the ground loses strength, so buildings can settle, tilt or sink and buried utilities can float or shear. It is a ground-failure risk — separate from how hard a site shakes (Vs30) and from the building's own construction.

How the score is built. A property's rating combines the three things liquefaction needs, all present at once (HAZUS screening method):
• Soil — loose sandy/silty ground and artificial fill are prone; solid rock and stiff clay are not (clay can soften but does not classically liquefy). Read from mapped surface geology, cross-checked against the soft-ground Vs30 class.
• Water — it needs a shallow water table, so dry or elevated ground scores low. Depth-to-water is estimated from elevation and distance to the coast (a screening proxy, not a measured well).
• Shaking — it has to be triggered by strong motion, so the site's own earthquake demand (PGA) is applied. The same soil rates higher in high-shaking SW Puerto Rico than in the calmer northeast.

Those combine into a liquefaction probability that maps to a six-step label — None → Very High — with the % beside the label being that probability.

San Juan / Bayamón. Inside the metro footprint a detailed borehole study (Bachhuber/Hengesh 2008; 685 SPT borings) replaces the island-wide screening wherever it is mapped, for higher accuracy.

Reading it. High / Very High is a flag to check the foundation type and whether ground improvement is warranted. It is a screening indicator, not a site-specific geotechnical (SPT/CPT) investigation.

Six plain-language classes, ordered from best to most demanding ground. Softer ground amplifies earthquake shaking and is more prone to liquefaction and settlement; each class is a consolidation of the underlying USGS rock-type units, which are preserved in the click popup.
• Hard Rock — volcanic & granite-type bedrock (island core). Best bearing, least amplification.
• Firm Rock — limestone/karst belts (north & southwest). Solid rock, low amplification, but sinkholes & voids possible.
• Medium Ground — layered sandstone/mudstone. Moderate firmness & amplification.
• Soft Soil — river & coastal deposits. Strongest amplification; flood- & liquefaction-prone.
• Man-made Fill — landfill / engineered fill. Softest class; settlement & liquefaction risk.
Screening-level mapped geology — not a substitute for a site-specific geotechnical investigation.