本页面用于教学和交互演示，不是地震预警、工程设计或灾害评估产品。
红色虚线为板块边界。断层模式只借用最近板块边界估算破裂走向，不代表已知活动断层位置或真实破裂面。
An educational interactive model, not an earthquake warning, engineering or disaster-assessment product.
P/S speeds are fixed at 7/4 km/s; wavefronts remain idealized epicentral rings, without finite-fault rupture timing. Fault mode uses the NIED Morikawa–Fujiwara basic PGA equation, with the 2023 Acc. coefficients and amplitude-magnitude saturation at Mw 8.2. Actual magnitude, rupture dimensions and duration remain uncapped. Point and fault modes share a single empirical amplitude scale of 1.5× by default, with no additional fault-only boost. It is not measured site response or a change to the published coefficients. Shallow thrust sources are assumed to be subduction interface; other shallow mechanisms are assumed crustal. Unspecified deep sources blend to intraslab coefficients over 60–120 km. Geometric spreading retains the actual 3D distance; an illustrative deep-path extension assumes a 40 km crust and a mantle/crust attenuation ratio of 0.25. This bounded extension is not a published MF13 term or a calibrated global deep-focus model; distances beyond 200 km remain extrapolations. Rrup is the 3D shortest distance to a buried horizontal rectangle at the chosen depth, including its ends; dip and slip are unresolved. Point mode uses hypot(horizontal distance, depth), removing the extra empirical power on depth. Both modes share the deep-path extension; depth enters geometric distance once. Synthetic multi-frequency waveforms feed a short-window peak detector. Bundled NOAA ETOPO 2022 metre elevations (0.25°) provide land-only physical slopes and relief, excluding ocean depth. The USGS Allen–Wald active-region slope relationship provides a 180–900 m/s Vs30 proxy, not measured Vs30. The NIED Acc. shallow-soil factor (Vs30/350)−0.523210 applies to both source modes. A separate illustrative convex-ridge term is bounded at 1.00–1.12; altitude alone does not amplify PGA. The roughly 25 km DEM cannot resolve local soils, sediment basins or nonlinear response, and the active-region relationship is not globally calibrated. Missing land data uses neutral corrections; deep-event amplification remains schematic.
JMA uses I = 2 log10(a) + 0.94 with a proxy of 60% PGA. Official instrumental intensity instead requires filtered three-component acceleration and cumulative 0.3-second amplitude thresholding. MMI is estimated independently from PGA using a GMICE relation. Neither estimate is observed shaking.
Reasenberg–Jones / Omori–Utsu rate with Poisson counts and truncated Gutenberg–Richter magnitudes. a=−1.67, b=0.91, p=1.08, c=0.05 days are generic California parameters, not a regional forecast. The seed preserves the catalogue and rupture geometry. The upper magnitude is mainshock−1; the largest event is stochastic, rather than forced to mainshock−1.2.
A selected window replays in 60 seconds, separately from P/S-wave seconds. The mainshock station field is retained. Select an occurred aftershock to run its wave field separately. Event pulses are schematic and respect reduced motion.
Interface dimensions use Strasser et al. (2010); other mechanisms use Wells–Coppersmith (1994) all-slip regressions. High-magnitude extrapolation remains uncertain. Dip, slip and rupture-time variation are not resolved. PB2002 boundaries provide estimated strike, not mapped active faults.
WorldPop 2020 (CC BY 4.0) counts are summed from 1 km to 5 arc-minute cells. Mainshock peak PGA envelopes, finite-source distance and the same slope-site/convex-relief/deep terms classify each cell centre. Exposure at JMA ≥5 lower, ≥6 lower and 7 is cumulative. These are modeled counts using historical population, not current population, casualties or observed intensity.
NOAA ETOPO 2022 is sampled at 0.25°. Marine connectivity excludes inland depressions and closed lakes. The simplified source condition requires M≥7.5, depth≤70 km, assumed thrust or normal faulting and ocean cells within the initial rupture. Other tsunami mechanisms are not ruled out by a failed condition.
Fast Marching solves first-arrival time with speed √(gh), land barriers and periodic longitude. All initial-source cells start at time zero. Named locations use nearby ocean cells (offsets in tooltips). Narrow straits may be unresolved. The coarse grid is unsuitable for minute-level coastal decisions and does not compute amplitudes. A separate shallow-water solver estimates wave heights.
Moment M0=10^(1.5Mw+9.1) N·m and rigidity 30 GPa set mean slip over the rupture area. Eighteen seeded, moment-conserving patches assume enhanced shallow slip. Okada vertical deformation initializes the sea surface. Dip and strike are assumptions; the source plane is shifted down if needed to remain buried. This tsunami geometry is separate from the horizontal-plane PGA approximation, and is not a historical slip inversion.
A linear spherical C-grid shallow-water solver runs for 24 h on 0.5° cells with a CFL time step, periodic longitude, impermeable land and a sponge at ±80°. No Coriolis, friction, dispersion, tides, nonlinear flooding or land run-up. The amplitude is maximum absolute sea-level displacement, not crest-to-trough wave height. Peak amplitudes are sampled at every step; display snapshots are 30 min apart.
Green shoaling prefers a nearby resolved ≥100 m offshore cell, falling back to resolved shallow-water cells on broad shelves within the same 120 km search radius. Assumed coastal depth is 10 m. The 0.5–2× range is a scenario envelope, not a statistical confidence interval or validated error bound. Coast colours sample the same offshore envelope at 0.25° spacing along the basemap coastline. Sub-threshold sections and locations without a resolved offshore reference remain uncoloured; this does not increase simulation resolution. They follow JMA height categories at the range upper bound: ≤0.2 m minor, ≤1 m advisory level, ≤3 m warning level, >3 m major warning level. No official warning is issued. Harbour resonance and surveyed 37–40 m land run-up cannot be resolved or compared directly to offshore amplitudes.