A way of ranking the risk to life from volcanoes is to establish how many people live in their vicinity. In addition to being an indicator of lives under threat, population exposure is a proxy for threat to livelihoods, infrastructure, economic assets and social capital. This report uses two indicators of population density around volcanoes to assess the current global exposure and as a risk indicator for individual volcanoes, and discusses this in combination with the Human Development Index (HDI) as a proxy for vulnerability.

Background

Ewert & Harpel (2004) introduced the Volcano Population Index (VPI), which estimates the number of people living within 5 and 10 km radii of volcanoes (VPI5 and VPI10). These population statistics, and VPI30 and VPI100 (population within 30 and 100 km of Holocene volcanoes) are reported in the VOTW4.0 (2013) database (www.volcano.si.edu; Siebert et al. (2010)). The Population Exposure Index, (PEI), was developed by Aspinall et al. (2011) for a study of volcanic risk in the World Bank's Global Facility for Disaster Reduction and Recovery (GFDRR) priority countries. Here, populations within 10 and 30 km radii were estimated and combined using weightings that reflect how historic fatalities vary with distance from volcanoes.

The VPI was developed on the basis that most eruptions are small to moderate in size (VEI ≤3), with footprints of less than 10 km. The VPI therefore represents the population exposures for most eruptions. Indeed, eruptions of VEI2 occur at a rate of approximately one every few weeks, and VEI3 several times a year (Siebert et al., 2010). Eruptions of larger magnitudes (VEI≥4) are less frequent, but often cause fatalities at distances well beyond 10 km (Auker et al., 2013). Hazard footprints from such eruptions commonly extend to tens of kilometres. The PEI thus complements the VPI, accounting for the high threat from large eruptions and potentially distal hazard types. An advantage of PEI is that only a single indicator parameter captures the exposure of populations around each volcano with the various VPI populations all contributing to the index and weighted according to historical evidence on the distribution of fatalities with distance. Here we develop and apply an amended version of the PEI, which correlates quite well with VPI10.