This article is part of The Purple Report September 2026 | Disaster Forecasting and Predictions.
Natural hazards do not usually know which ministry, company or engineering discipline owns the next system.
A landslide can remove a road.
The missing road can delay power repair.
The power loss can shut pumps and communications.
Communications loss can slow rescue.
Slow rescue can increase mortality from an event whose physical footprint was relatively small.
The most important disaster hotspot may not be where the strongest hazard occurs. It may be where one failed node removes many other capabilities at once.
Why Cascade Potential Changes the Forecast
Traditional hazard maps often begin with the physical event.
Where will the ground shake?
Where can the flood reach?
Which slopes are unstable?
Which volcanic valleys contain lahar pathways?
Those questions are essential.
The Purple Forecast adds:
- Which power system lies there?
- Which hospital depends on that road?
- Which port is the only practical import route?
- Which data centre, pipeline or substation has no close substitute?
- Which bridge connects two otherwise separated networks?
- Which water system stops when electricity fails?
- Which emergency teams cannot reach affected people after transport fails?
UNDRR’s Global Assessment Report 2025 explains why this matters economically: direct disaster losses substantially understate the wider cost once cascading and ecosystem effects are counted.
UNDRR — Global Assessment Report 2025
The Five Cascade Questions
- What fails first? The physical node directly struck by the hazard.
- What depends on it? Services, routes, organisations or communities that rely on that node.
- What alternative exists? A backup route, spare system, reserve supply or substitute technology.
- How long can the alternative carry the load? Redundancy can be temporary.
- What happens when repair itself depends on the failed system? This is where cascades become self-reinforcing.
Gyirong: Road Failure Became Rescue Failure
The Gyirong disaster provides the project’s clearest current example.
The physical event damaged a Himalayan road and border corridor.
That changed access.
Access affected rescue, communications repair, power restoration, logistics and border function.
The hazard therefore spread through linked dependencies:
mass movement → road loss → access loss → slower rescue/power/comms repair → longer disruption
The important forecast signal was not merely “unstable mountain”.
It was “unstable mountain above a high-criticality, low-redundancy corridor”.
Civilisation Atlas — Shigatse and the Gyirong Disaster
Power Is Often a Hidden Multiplier
Electricity rarely appears on the first hazard map.
But after a disaster, power can determine whether:
- water pumps operate;
- mobile towers function;
- hospitals maintain critical systems;
- fuel can be pumped;
- traffic control works;
- food refrigeration survives;
- digital payments continue;
- homes remain safe during heat or cold.
A power substation can therefore have a much larger disaster footprint than its physical size suggests.
The forecast should rank functional criticality, not just asset value.
Water and Power Can Form a Circular Dependency
Many water systems need electricity for pumping, treatment and control.
Power systems may need water for cooling, hydroelectric generation or fuel logistics.
When both systems depend strongly on one another, disruption can become circular.
The Purple Forecast therefore looks for:
- onsite emergency power;
- gravity-fed water;
- stored potable water;
- independent communications;
- alternative generation;
- fuel-delivery access;
- islanding capability;
- manual operation when digital controls fail.
These are not glamorous assets.
They often determine whether a cascade stops at one layer.
Ports: A Local Hazard Can Enter the World Economy
Ports concentrate cranes, channels, roads, rail, customs, fuel, warehouses, data and shipping schedules into one interface.
A port earthquake, storm surge or power failure can therefore propagate outside the city where it occurs.
The Pearl River Delta, Cascadia, Gulf Coast and Singapore-facing regional supply chains all illustrate this principle in different ways.
The relevant forecast field includes:
- local hazard;
- port throughput and substitution;
- alternate ports;
- road/rail evacuation;
- fuel and power;
- shipping-network rerouting;
- inventory buffers at receiving economies.
Hospitals: Available Beds Are Not Enough
A hospital is not resilient because the building survives.
It needs:
- electricity;
- water;
- oxygen;
- staff access;
- communications;
- pharmaceutical supply;
- working roads;
- waste removal;
- space for surge patients.
A disaster can leave the hospital building intact and still remove healthcare capacity.
This is the same distinction that appears throughout the Civilisation Atlas:
Installed capacity is not automatically deployable capacity.
Communications: Observation Can Fail During the Event
Modern emergency response depends heavily on information.
But the disaster can damage the systems used to observe and report the disaster.
Mobile towers lose power.
Fibre routes are cut.
Weather stations disappear.
Roads prevent technicians from reaching instruments.
That means the system can become least observable when it is most dangerous.
Satellite communications, independent power and multiple information routes therefore have forecast value because they preserve reliable reporting and coordination when primary networks fail.
The Cascade Matrix
| Initial loss | Common second-order effects | Potential third-order effects |
|---|---|---|
| Road/bridge | rescue and supply delay | health, power, food and repair duration worsen |
| Power | pumps, telecoms, refrigeration fail | water/health/food/digital services degrade |
| Port | imports/exports disrupted | fuel, food, industrial supply chains reroute |
| Communications | coordination and warning degrade | rescue slows; uncertainty rises |
| Hospital | care capacity falls | indirect mortality and evacuation load rise |
| Water | drinking/sanitation loss | health emergency and displacement grow |
| Dam/hydropower | power or water-flow state changes | downstream flood, energy and irrigation effects |
| Data centre/payment rail | digital coordination or transactions fail | commerce, logistics and public-service friction increase |
Criticality Without Catastrophising
A critical node should not automatically become HIGH-CONCERN.
Criticality tells us consequence if the node fails.
It does not tell us how likely the hazard is.
The Purple Forecast therefore keeps two questions separate:
- Hazard convergence: is failure becoming more physically plausible now?
- Cascade criticality: how much capability disappears if failure occurs?
A low-probability, enormous-cascade system belongs in long-term preparedness.
A high-probability, low-cascade event may deserve a different intervention.
Redundancy Is Not Always Duplication
A backup does not have to look identical to the primary system.
A blocked railway can be bridged temporarily by buses.
A failed fibre link can be replaced temporarily by satellite communication.
A damaged port can shift some cargo to another port.
A grid failure can be buffered by islandable local generation.
The function is what needs redundancy.
The form can differ.
How the Forecast Will Measure Cascade Potential
- Dependency count: how many important functions require the node?
- Substitutability: can another route or asset perform the same function?
- Time-to-substitute: how long does switching take?
- Buffer duration: how long can reserves carry demand?
- Repair dependence: does fixing the node require the node’s own function?
- Geographic concentration: can one hazard hit multiple redundancies simultaneously?
- Human sensitivity: which people lose essential capability first?
- External propagation: does failure move into other regions or countries?
The Forecasting Payoff
Hazard science tells us where the trigger may occur.
Cascade analysis tells us where prevention has the highest leverage.
Sometimes the best disaster intervention is not changing the mountain, storm or fault.
It is:
- a second road;
- a second communications path;
- emergency power;
- distributed water storage;
- alternate port agreements;
- hospital surge arrangements;
- redundant warning channels;
- moving one critical asset out of a shared hazard zone.
The catastrophe forecast becomes actionable when it shows which dependency can be changed before the hazard arrives.