3.1. Global Population Distribution from 1975 to 2015 in Relation to Volcanism
This section presents and discusses the global distribution of population from 1975 to 2015 in relation to volcanism, considering both overall population in each period and annual growth rates by distance to both HV and SV. While the first part focuses on significant distance ranges up to 100 km of volcanoes, a second part analyses the specific spatio-temporal patterns of population density within 50 km of volcanoes in more detail.
shows total accumulated population within selected distance ranges (radial distances) from Holocene volcanoes (HV), for each period considered. Results show that the total population living in proximity of Holocene volcanoes (HV) is very significant and that this amount has been increasing steadily in last 40 years. In 2015, more than 1 billion people, 14.3% of the global population, are estimated to live within a range of potential direct impact of volcanic eruptions (i.e., within 100 km). Close to 60 million people are estimated to live within potential evacuation range worldwide (i.e., within 10 km).
Concerning population growth rates in vicinity of Holocene volcanoes (HV) (Table 2
), these are generally decreasing with time, accompanying global population trends. These rates were especially high from 1975 to 2000, and were increasing with proximity to HV. However, despite slowing down with time, rates in vicinity of HV were largely above the global average in all periods.
and Table 4
refer to variation of population within selected distances of volcanoes with significant eruptions (SV). Results show that this population has also been increasing steadily and amounts to 616 million in 2015, or 8.4% of global population. Also very significant are the 21 million people living within 10 km of these volcanoes.
Regarding population change with distance bands, growth rates are lower than in proximity to HV. Also differently from HV, population growth has been generally decreasing with time but also with proximity to SV volcanoes, except for the closest distance band (10 km). Only in the period 1975–1990 did population in these areas increase above the overall growth rate, having dipped below the global rate in the most recent period. In the period 2000–2015, population growth in these areas dipped below global growth, especially closer than 30 km from volcanoes. This may be related to: (i) the specific demographic trends of the countries and areas where SV volcanoes are located and/or (ii) with an increased awareness of the threat posed by these volcanoes and limitations to settlement in their vicinity.
and Figure 3
show overall global cumulative population as a function of radial distance to Holocene volcanoes (HV) and volcanoes in Significant Volcanic Eruption Database (SV), respectively, for each period under analysis. The plots show that very large numbers of people live in proximity to volcanoes, and that this presence has increased from 1975 to 2015. Concerning HV, while 100 million people would be spread over a radial distance of 20 km in 1975, the same amount were concentrated in a radius of only 13 km in 2015.
This has noteworthy consequences for the variation of population densities in space and time, as shown in Figure 4
and Figure 5
Differently from some previous analyses (e.g., [8
]), we have considered all volcanoes listed in the databases regardless of their type or location, as e.g., submarine volcanoes may still affect populations on land upon erupting, as many are located in proximity to inhabited areas. This is especially relevant for remote archipelagos (such as Tonga, French Polynesia, Solomon Islands Guam, Hawaii, Mariana Islands), where potential exposure and risk is typically overlooked in global analyses and where DRM poses specific challenges. In fact, we think this may be a ‘hidden’ source of potential exposure and risk that should deserve more attention in future research. Of 118 submarine volcanoes in the HV list, 76 have documented eruptions.
and Figure 5
show the average population densities (over land) by radial distance to volcanoes. These charts show that relatively high population densities occur in the vicinity of all volcanoes, especially of those with significant eruptions (above the overall global densities over land, which are 27 and 49 people/km2
in 1975 and 2015, respectively); and these densities increase sharply away from volcanoes but in very close proximity, especially from SV, in whose vicinity densities are significantly higher. Moreover, while maintaining the overall pattern, those densities have been increasing considerably since 1975, particularly in the period 1975–1990. However, it is at a distance of 10 to 20 km from Holocene volcanoes that the absolute increase in population density has been greatest (additional 75 persons/km2
), from 1975 to 2015.
In all periods, population density generally increases with close proximity to volcanoes, peaking at around 15–20 km away, and then decreases sharply in their immediate vicinity, but still maintaining relatively high densities. This sharp decrease may be due to presence of steep slopes, volcanic activity, or enforcing of exclusion zones for settlement. This pattern is even more striking for volcanoes with significant eruptions (SV), where overall population densities are higher (up to 270 people/km2
), with obvious implications for exposure, risk analyses, and potential impacts. This pattern is dissimilar from the monotonically decreasing densities reported by [8
]. This difference may be the outcome of sharper analyses enabled by the characteristics of the new population grids, i.e., effective disaggregation to built-up areas and higher spatial detail.
These results further detail, revise and update previous estimates, such as that of reference [1
] that states “…at present, more than 800 million people in 86 countries live within 100 km of a volcano that could potentially erupt (GVM, 2014a)”, and that “…over 29 million people worldwide live within just 10 km of active volcanoes, and around 800 million people live within 100 km” [25
]. Likely reasons for such disagreement include their reliance on conceptually different population grids (e.g., LandScan versus GHS-POP), and their reliance on older census data. Also, LandScan does not depart from or aim to model distribution of residential
population, from which population living
in proximity can be estimated with more confidence, but instead maps ambient
Sources of Uncertainty
As with all global data, both the volcano datasets and the population grids are affected by error and uncertainty. This analysis is performed using a 250 m population grid, which also makes it sensitive to uncertainty in spatial precision. The uncertainties in GHS-POP mostly result from the quality and reliability of censuses and the capacity to fully detect built-up in diverse landscapes using heterogeneous Landsat image archives [32
]. Such issues typically worsen with temporal distance from the current period (for a more detailed discussion, see reference [27
]. Despite the mitigation procedures implemented in the production of the current population grids, some issues remain unsolved in global population data, which should not be used for pixel level assessments.
Issues of spatial accuracy and precision also affect the GVP datasets in which volcanoes are represented through points (coordinate pairs), and a single point frequently refers to a group of volcanoes or craters (e.g., volcanic fields) or giant calderas (e.g., Yellowstone, USA); for some of these, due to their size and extent, an areal feature (polygon) would be a better representation for the potential hazard source. Also, the geographical coordinates (decimal degrees) are often coarse with precision sometimes limited to the first decimal, which corresponds to a real world precision of about 5.5 km. This is clearly insufficient for detailed analyses in close proximity to volcanoes. This vagueness and low spatial precision reduce accuracy and discourage more detailed analyses. These and other uncertainties introduce limitations for studies: while these are usually mitigated when aggregated assessments such as this one are conducted, they can become apparent when analyses are carried out for individual volcanoes or at very fine scale (e.g., very close proximity).
3.2. Population Distribution from 1975 to 2015 in Relation to Volcanism, in Southeast Asia and Central America
This section analyzes in more detail results for two important volcanic regions, Southeast Asia and Central America. The first region comprises the countries of Indonesia, Philippines, and East Timor, while the second includes Mexico, Guatemala, Belize, Honduras, El Salvador, Nicaragua, Costa Rica, and Panama. Figure 6
shows location of Holocene volcanoes in these regions and the selected distances under analysis. SE Asia and Central America together represent 40% of the total global population living within 100 km of volcanoes, but with different contributions and patterns of population by radial distances (Table 5
In SE Asia the population within 100 km of HV has increased from 143 million in 1975 to 301 million in 2015, representing 84% of the region’s inhabitants. In Central America the population within this distance increased from 57 million in 1975 to 120 million in 2015, comprising 69% of the region’s inhabitants. Especially striking in SE Asia are the 10 and the 27 million people currently living within 10 km of SV and HV, respectively, accounting for almost half of the global population within this range of recently active volcanoes. This short distance is especially problematic in small inhabited islands which are completely contained within this range, where fast evacuation can be particularly challenging.
However, while in SE Asia the share of total population within 100 km of HV has increased (albeit slightly) between 1975 and 2000, in Central America this value has been decreasing, and even more so when SVs are considered.
Concerning the annual growth rates of population by radial distance of HV and SV, Table 6
shows important differences between and within these regions.
While in SE Asia the highest growth rates have occurred in close proximity to HV and SV volcanoes (within 10 km), in Central America the strongest rates are observed farther away (beyond 50 km), especially after 1990 and for HV. Also, whereas in SE Asia much of this growth has occurred at rates well above the regional ones between 1975 and 2000 (especially for HV), in Central America increases stayed below regional rates for all distances and periods, suggesting that areas more distant from volcanoes were favored for settlement.
While explaining the observed variations is beyond the scope of this paper, we can offer some hypotheses for the underlying causes:
Pressure for space in a context of relatively weak planning, causing settlement to encroach on volcanoes (overall average population densities are much higher in SE Asia than in Central America);
More recent dates of eruptions and variations in risk perception.
Additional research is needed to test these and other hypotheses that may explain settlement patterns.
Regarding variation of population densities in these regions in proximity of HV and SV, Figure 7
and Figure 8
show some differences. For both volcano sets, population densities are generally higher in SE Asia, especially in 2015 where these can be more than twice those in Central America (HV). In SE Asia, population densities increase faster with distance from HV and remain relatively high (>400 persons/km2
) up to 44 km away. Beyond 4 km, density surpasses the overall mean for the region (78 and 163 persons/km2
in 1975 and 2015, respectively). There is a pronounced peak in densities around a distance of 18–22 km that has intensified since 1975. Because this is essentially a mountainous landscape, this may also reflect the average distance of valley floors (or the mid-distance between consecutive mountains), where conditions for settlement are more favorable. Figure 9
illustrates the situation in 2015 in the Island of Java (Indonesia), showing the locations of Holocene volcanoes, main cities, and the population distribution.
Regarding SV volcanoes (Figure 8
), in Central America densities increase dramatically up to 6–8 km and remain high up to 16 km, decreasing markedly after that distance. SE Asia displays a similar proximal pattern for both HV and SV, with very high densities observed around 20 km from SV (700 people/km2
) and remaining high until 50 km away.
These results suggest that the overall settlement patterns observed, although specific for each region, were mostly defined prior to 1975, especially in Central America. These spatio-temporal patterns of settlement have consequences for DRM. This analysis confirms and characterizes increases in the exposure of population. Due to the large size, densities, and trends of population potentially exposed to volcanic activity, coupled with structural vulnerabilities, SE Asia and Central America constitute hot spots of exposure and risk. In both regions, the high densities of inhabitants in the immediate vicinity of volcanoes poses widespread challenges for DRM, in particular for evacuation though also the sheer scale of response and recovery operations can become quite daunting in case of large eruptions. In this context, effective monitoring of volcanic activity, emergency preparation and early warning can significantly reduce local risk and potential impacts from eruptions.
Furthermore, these new population data and results have implications for exposure indices, such as the VPI [19
] and PEI [24
]. Due to their reliance on and sensitivity to population distribution data, such indices should use the latest population grids available and be regularly updated to accurately reflect the current exposure and risk situation.