products/atlas
01 · Catastrophe risk

Atlas. Catastrophe risk, on the map.

A map-first way to read catastrophe risk. Atlas lays hazard over place, so risk reads where it actually lands. The current public study shows one tropical-cyclone layer over one region, the Greater Bay Area, on Edgion's own 3 km storm simulations. Further perils and regions are on the roadmap as their physics lands.

[ open the risk map ] [ scope a region ]
↓ the map
3 kmconvection-permitting mesh
1public study module (TC)
3storms simulated
1current study region
The idea

Risk is not a row in a table. It is a place on a map.

Most catastrophe output arrives as a single loss number for a portfolio. That hides the thing a decision actually turns on: where the hazard lands, how far it reaches, and which assets sit inside the footprint.

Atlas is map-first. Each peril is a layer you can switch on over a real region, read against a calibrated scale, and trace down to the asset. Today only one layer is modelled, built on physics rather than a fitted curve.

The risk map

One region, every storm, layer by layer.

The tropical-cyclone module over the Greater Bay Area. Pick a storm, switch the hazard layer, and read the footprint against its scale. Every map below is an Edgion 3 km simulation, not stock imagery.

ATLAS · TROPICAL CYCLONE
Greater Bay Area tropical-cyclone hazard map, Edgion 3 km simulation.
Mangkhut 2018 · 48 h rainfall
99th-pct response 1.83× vs thermodynamic reference 1.0×
48 h rainfall · ensemble meanmm
hazard colour scale
025100200400700
0.3member fraction crossing ≥50 mm/hr1.0
Ensemble-mean fields, CPAS 3 km convection-permitting mesh. Rainfall shows 48 h accumulation; flash-flood shows where the warming ensemble crosses the ≥50 mm/hr threshold beyond the historical footprint.
From map to asset

Every footprint resolves to a per-region hazard readout.

Drop a location and Atlas reads the current study layer at that place. Where a peril is modelled, the readout carries the real metric and its scale; where it is not yet, the row stays explicitly open rather than guessing a number.

// HAZARD READOUTstorm set Mangkhut · Hato · Hagupit
TC rainfall48 h, ensemble mean
0.63–1.83× ref
Flash-floodnew ≥50 mm/hr zone
26.8–92.6k km²
TC windhigh-wind area shift
+10 to +22%
Storm surgecoastal inundation
layer in build
Seismicground shaking
roadmap
Current-study rows carry the real range across the three simulated storms; roadmap rows are planned layers not yet modelled for this region. Figures are illustrative of the method, not a guarantee for any specific asset.
How it runs

From observed storm to a layer on the map.

01
Ingest
Observed storm track, terrain, and bias-corrected climate state for the region.
02
Simulate
Replay at 3 km on an adaptive mesh; historical and warming controls plus ensemble.
03
Resolve
Extract rainfall, flood, and wind fields as calibrated hazard layers.
04
Map
Render each peril over the region against a documented scale.
05
Read
Score any location inside the footprint and export the readout.
The demo

Built to go multi-hazard. Honest about what is modelled today.

Current study · tropical cycloneRainfall and flash-flood layers over the Greater Bay Area, on Edgion's own 3 km simulations of Mangkhut, Hato, and Hagupit.
One engine, many perilsThe same map, layer, and readout framework carries surge, seismic, and landslide as each peril's physics and data are brought in.
Region by regionThe GBA is the current study region today. New regions come online as their storm set is simulated and calibrated.
[ scope a region ] [ see the science ]

// note: Atlas is a map-first catastrophe risk demo with one current public study module (tropical cyclone) over the Greater Bay Area. Layers shown as roadmap are not yet modelled and carry no result figures. Current maps are Edgion 3 km storyline simulations; they are physics-based stress tests of how past storms could reorganise under warming, not forecasts of specific future events, and are illustrative rather than a guarantee for any individual asset.