How Population Exposure is Calculated in Disaster Monitoring

Why Population Exposure Matters

A magnitude 6.0 earthquake in the Sahara Desert and a magnitude 6.0 earthquake beneath Mexico City are physically similar events, but their human consequences are vastly different. Population exposure is the metric that captures this difference. By estimating how many people are within the affected area of a disaster event, monitoring platforms can distinguish between events that are scientifically interesting and events that require urgent humanitarian response.

Population exposure is one of the four components of the Calamity Score, contributing 30% of the final value. It is also used independently for triage: events with high population exposure receive priority in alert systems and are more likely to generate public-facing event pages.

The Population Exposure Index (PEI)

The Population Exposure Index is a weighted sum of population counts, where each population center contributes to the total based on its distance from the event. The fundamental formula is:

PEI = Sum of (city_population x decay(distance) x source_confidence)

This approach accounts for three factors simultaneously: how many people live near the event, how far away they are, and how precisely we know the event's location.

Gaussian Decay Model

The decay function determines how quickly population exposure diminishes with distance from the event. Rather than using a simple radius cutoff (everyone within 50 km is exposed, everyone outside is not), the system uses a gaussian decay curve.

The gaussian function produces a smooth, physically realistic falloff. People very close to an earthquake epicenter experience strong shaking, while those at moderate distances experience progressively weaker effects. The function shape mirrors how most disaster impacts actually attenuate in the real world.

The formula is: decay(d) = exp(-d squared / 2 sigma squared)

Here, d is the distance from the event to the population center, and sigma is the characteristic decay radius. The sigma parameter varies by disaster type because different hazards affect different spatial scales:

• Earthquakes: Sigma is relatively large because seismic waves propagate over hundreds of kilometers, though with diminishing intensity.
• Wildfires: Sigma is moderate, reflecting the typical smoke plume range and direct fire threat area.
• Volcanic eruptions: Sigma can be very large for ash dispersal but small for lava flow impact zones.
• Floods: Sigma is typically smaller, constrained to river valleys and low-lying areas.
• Air quality events: Sigma is large, as airborne pollutants can affect wide regions.

The choice of sigma for each disaster type is calibrated against historical damage and casualty data. The goal is to produce PEI values that correlate well with observed human impact.

Nearest City Analysis

The population model uses a global city-level dataset that includes population estimates for thousands of populated places worldwide. For each event with known coordinates, the system performs a spatial query to identify all cities within a maximum search radius.

Each city within range contributes to the PEI proportionally to its population and inversely to its distance. A city of 500,000 people at 20 km contributes far more than a town of 5,000 at the same distance, and both contribute more than equally sized settlements at 200 km.

The nearest city is also used for human-readable event descriptions. When the platform reports an earthquake occurred "near Kathmandu", it is citing the nearest significant population center from this spatial analysis.

Source Confidence Factor

Not all event locations are equally precise. A well-instrumented seismic network can localize an earthquake to within a few kilometers. A satellite fire detection from a moderate-resolution sensor might only narrow the location to a 10-20 km pixel. A flood report from a news aggregator might reference a city name with no precise coordinates.

The source confidence factor, ranging from 0.7 to 1.0, adjusts the PEI to account for location uncertainty. Events with high-confidence coordinates (from well-instrumented networks) receive the full population weight. Events with lower-confidence coordinates are attenuated slightly, preventing the system from over-counting population exposure for imprecisely located events.

This factor is determined by the monitoring source's known spatial resolution and instrumentation quality, not by any manual assessment.

Worked Example

Consider an earthquake detected at coordinates (28.4, 84.1) in central Nepal:

1. The system identifies cities within the search radius: Kathmandu (population 1,400,000 at 85 km), Pokhara (population 400,000 at 50 km), Bharatpur (population 300,000 at 30 km), and several smaller towns.
2. The gaussian decay function is applied with the earthquake-specific sigma. Bharatpur at 30 km receives a high decay multiplier, Pokhara at 50 km a moderate one, Kathmandu at 85 km a smaller but still significant one.
3. Each city's population is multiplied by its decay factor and by the source confidence (0.95 for a major seismic network).
4. The weighted populations are summed to produce the final PEI value.

This PEI value feeds into the Calamity Score's population impact component, ensuring that this event scores substantially higher than an equivalent-magnitude earthquake in an uninhabited region.

Limitations of Population Exposure Estimation

The gaussian decay model is a simplification. Real disaster impacts depend on local topography, building construction quality, time of day, warning lead time, and many other factors that a global monitoring platform cannot assess for every event in real time.

Population data has inherent limitations as well. City population figures may lag behind census cycles, and they do not capture transient populations such as tourists, commuters, or displaced persons. The model estimates potential exposure, not actual casualties or damage.

For precise impact assessments, national disaster management agencies use detailed building inventories, vulnerability models, and ground-truth reports. The PEI provides a rapid first estimate that can be computed within seconds of event detection, serving as a triage tool rather than a definitive damage assessment.