New Multi-temporal Global Population Grids – Application to

Transcription

Freire et al.
New Multi-temporal Global Population Grids and
Volcanism
New Multi-temporal Global Population
Grids – Application to Volcanism
Sérgio Freire
Aneta Florczyk
European Commission – Joint Research
Centre (JRC), Institute for the Protection and
Security of the Citizen
[email protected]
European Commission – Joint Research
Centre (JRC), Institute for the Protection and
Security of the Citizen
[email protected]
Martino Pesaresi
European Commission – Joint Research Centre (JRC), Institute for the Protection and Security
of the Citizen
[email protected]
ABSTRACT
Better and finer global analyses of human exposure and risk of natural disasters require improved
geoinformation on population distribution and densities, in particular concerning temporal and spatial resolution
and capacity for change assessment.
This paper presents the development of new multi-temporal global population grids and illustrates their value in
the context of risk analysis by estimating the worldwide distribution of population in relation to recent
volcanism. Results indicate that almost 6% of the world’s 2015 population lived within 100 km of a volcano
with at least one significant eruption, and more than 12% within 100 km of a Holocene volcano, with human
concentrations in this zone increasing since 1990 above the global population change rate.
The novel 250-m resolution population grids constitute the new state-of-the-art in terms of global geospatial
population data, with the potential to advance modeling and analyses at all stages of the emergency management
cycle.
Keywords
Built up, GHSL, population distribution, dasymetric mapping, volcanoes, spatio-temporal analysis.
INTRODUCTION
Evaluating potential or actual human exposure for global disaster risk assessment is as essential as it is
challenging, being limited by the availability and quality of geophysical and socio-economic data (Lerner-Lam,
2007; Peduzzi et al., 2009). As a result, estimations of human exposure are typically conducted for small areas
and without a temporal dimension, or merely approximated with coarse population geoinformation. However,
the rapid increase in exposure of population and assets is currently a major driver of growing disaster risk
(UNISDR, 2013), requiring adequate global geoinformation layers depicting the distribution and densities of
population. Such geoinformation is valuable at any stage of the emergency management cycle, but for timely
and efficient response it should be prepared and made available beforehand (Freire, 2010). After a disaster
event, the quality and level of detail of population data have a direct effect on response and lives saved (NRC,
Short Paper – Geospatial Data & Geographical Information Science
Proceedings of the ISCRAM 2016 Conference – Rio de Janeiro, Brazil, May 2016
Tapia, Antunes, Bañuls, Moore and Porto de Albuquerque, eds.
Freire et al.
Volcanism
2007).
Although global population grids have been produced since the 1990s, they suffer from some issues which still
limit their usability (Linard and Tatem, 2012; Mondal P. and Tatem, 2012). Among these are their coarse spatial
resolution, incomplete discrimination of settlements (especially smaller ones), and the use of dichotomous maps
of built-up (BU) in their construction. Also, for change detection and analysis, a coherent and robust spatiotemporal production model is required to allow objective comparisons.
Volcanic eruptions are among the most powerful and destructive natural hazards on the planet, capable of
causing extensive damages and harm to people from direct impacts and disturb livelihoods from indirect effects
(e.g. via disruptions of air transport, through climate change, etc.). Single events have the potential (and have
been known) to cause sudden global changes and mass extinctions of species. Concerning volcanic risk,
Andredakis and De Groeve (2015) highlight that risk elements are on the rise, most notably population
exposure. The shifting focus of urbanization from developed to developing countries taking place in recent
decades is also increasing global exposure to natural hazards, including volcanism (Chester, Degg, Duncan, and
Guest, 2001). Assessments of potential worldwide population exposure to volcanic hazard have been conducted
using globally-available population grids (e.g. Small and Naumann, 2001; GFDRR, 2011; Brown, Auker and
Sparks, 2015), but most such studies are now outdated or use coarse or otherwise inadequate population
geoinformation.
Small and Naumann (2001) have quantified human proximity to active volcanoes for 1990, but underline that
the spatial resolution of population data used (GPWv2) imposes limitations on the analysis and is not adequate
for detailed assessment of potential volcanic risk. The recent developments in global mapping of settlements and
built-up (Pesaresi and Ehrlich, 2009) enable production of improved global population grids, and these in turn
may allow conducting more updated and refined assessments of population potentially affected by hazards, such
as those posed by volcanoes.
This paper aims at (1) presenting the development of new and improved global population grids (the GHS-POP
model) and (2) illustrating their value by estimating globally the recent evolution and current distribution of
population in relation to historic volcanism.
DATASETS
Input data for this study has included geoinformation used to model population distribution, and data on global
volcanoes.
In order to model and map global population distribution, two geospatial datasets were combined. Their main
characteristics are listed in Table 1.
Data set
Source
GHS-BUILT
Population estimates
Dates
Data type
EC-JRC, IPSC
1990-2000-2014
Raster (38 m)
CIESIN-GPWv4
1990-2000-2015
Vector polygon
Table 1. Main input datasets used in modeling population distribution
The Global Human Settlement Layer built-up grids (GHS-BUILT), recently produced from Landsat imagery
collections (epochs: 1975, 1990, 2000 and 2013-2014), constitute a good covariate for population
disaggregation (Freire, Kemper, Pesaresi, Florczyk, Syrris, 2015). GHSL technology relies on automatic
analysis of satellite imagery to produce unprecedented fine scale maps quantifying built-up structures in terms
of their location and density (Pesaresi, Ehrlich, Ferri, Florczyk, Freire, Halkia, Julea, Kemper, Soille, and
Syrris, 2015) (Fig. 1). The image processing technology exploits structure (texture, morphology, and pattern) as
key information, outputting a texture-derived “built-up presence index” (Pesaresi, Huadong, Blaes, Ehrlich,
Ferri, Gueguen,, Halkia, Kauffman, Kemper, Linlin, Marin-Herrera, Ouzounis, Scavazzon, Soille, Syrris, and
Zanchetta, 2013). The distribution of built-up areas is expressed as their proportion in each cell. GHSL BU was
aggregated from its native output resolution (38 m) to that selected for modeling population (250 m), in World
Mollweide projection (Fig. 2).
Freire et al.
Volcanism
Figure 1. GHSL layer showing global built-up in 2014 (GHS-BUILT 2014), overlain on OpenStreetMap
Figure 2. GHSL 250m layers depicting built-up before 1990 (GHS-BUILT 1990), built-up added between 1990-2000
(GHS-BUILT 2000) and 2000-2014 (GHS-BUILT 2014) in the South of Surabaja, Indonesia
Population estimates were provided by CIESIN-GPWv4 (CIESIN, 2014) for matching periods. CIESIN collects
and integrates the most recent and finest resolution census data available, and for each census polygon estimates
resident population for given target years, adjusting figures to UN World Population Prospects. CIESINGPWv4 estimates build on official statistics and do not model or make inferences on the actual location of
population beyond the collected data, therefore providing spatio-temporal population datasets which are
especially suited for integration with other geoinformation. The CIESIN-GPWv4 input dataset comprises close
to 8 million populated census or administrative units in 2015, with average spatial resolution varying widely by
country.
Two datasets of volcanoes have been used, namely the Holocene Volcano List v 4.4.1 of the Smithsonian
Institution's Volcanism Program (GVP, 2013) and the NOAA Significant Volcanic Eruption Database (SVED),
this one also based on GVP. The Holocene Volcano List (HVL) is a global listing of over 1500 volcanoes which
includes information on the latitude, longitude, elevation, type of volcano, and last known eruption. Holocene
volcanoes are those believed to have been active during the Holocene epoch (the past 12,000 years).
The Significant Volcanic Eruption Database (NOAA, 2015) is a global listing of over 500 significant eruptions
which includes information on the latitude, longitude, elevation, type of volcano, and last known eruption. A
Freire et al.
Volcanism
significant eruption is classified as one that meets at least one of the following criteria: caused fatalities, caused
moderate damage (approximately 1 million USD or more), with a Volcanic Explosivity Index (VEI) of 6 or
larger, caused a tsunami, or was associated with a major earthquake. In this work, we excluded from the analysis
all submarine volcanoes if they are located below sea level, and volcanoes in Arctic and Antarctic zones. Table
2 outlines the data preparation steps, while Figure 3 illustrates final volcano datasets.
Holocene Volcano List (HVL)
Significant Volcanic Eruption
Database (SVED)
Input dataset
1,540 volcanoes
672 eruptions / 224 volcanoes
Data manipulation
(C - clean)
(C) Arctic/Antarctica (latitude ±70)
(C) Submarine, below sea level
Final dataset (tot. no. of
volcanoes)
1,423
220
Table 2. Preparation of volcano datasets
Figure 3. Distribution of final datasets of volcanoes (VD) and volcanoes with at least one significant eruption
(SVED), overlain on OpenStreetMap
MODELING AND ANALYSIS
Modeling global population distribution: 1990-2000-2015
To produce the new population grids, a simple model was developed and implemented. The approach is based
on raster-based dasymetric mapping (Wright, 1936) relying on GHS-BUILT as proxy to restrict and refine the
distribution of people and inform on the respective density.
For each period, population grids were produced following a simple volume-preserving dasymetric mapping
approach: given a census layer and a GHS-BUILT raster layer, for a populated polygon,
a) if the polygon generates 250 m cells and contains BU, disaggregate the population in proportion to BU;
b) if the polygon generates 250 m cells and does not contain BU, disaggregate population using areal weighting;
c) if the polygon does not generate its own 250 m cell, convert polygon to point (centroid within), sum all points
within a cell, and sum to mosaic of a) and b)
This approach allows preserving the full resolution of the GPW input population units and preserves the total
input population. The population distribution grids depict the distribution of resident-based population in builtup, in 1990, 2000, and 2015. Since these grids are produced and made available in Mollweide projection, values
Freire et al.
Volcanism
represent both population counts as well as densities.
Assessing global population distribution and volcanism
We analyze global population distribution as a function of distance to volcanoes in 1990, 2000, and 2015, using
the produced 250 m population grids. Using GIS software, population distribution is analyzed within a radial
distance of 100 km from volcanoes, by buffering each volcano in 5 km steps and conducting zonal analysis of
GHS population grids (illustrated in Figures 4 and 5 in Java, Indonesia).
A distance of 100 km is relevant for assessing direct effects of volcanic eruptions, since lethal pyroclastic flows
and surges (Nakada, 2000), and lahars (Rodolfo, 2000) may occasionally extend to these distances. Chester et al.
have estimated that among twelve destructive phenomena associated with volcanic eruptions, seven can
potentially reach such a distance (Chester et al., 2001).
Figure 4. Distribution of volcanoes in Java (Indonesia) and surroundings and the 5-km buffers (Buf VD), overlain on
OpenStreetMap
Figure 5. Distribution of volcanoes in the South of Surabaya, Indonesia, the respective 5-km buffers (Buf VD) and
the GHSL layer (GHS-BUILT 2014), overlain on OpenStreetMap
RESULTS AND DISCUSSION
Freire et al.
Volcanism
Global population distribution grids
These spatial raster datasets depict the distribution and density of resident-based population in each period,
expressed as the number of people per cell. Quality control was conducted to ensure that input totals were
preserved. Table 3 summarizes the results of population gridding to the disaggregation method applied in model,
in each period.
Disaggregation method
1990
2000
2015
Points
1.8
1.7
2.3
AW
4
2.8
2.2
BU
94.2
95.5
95.5
Total
100
100
100
Table 3. Results of population disaggregation according to method applied in model (in percent of total population).
Concerning the disaggregation method effectively applied in each grid, Table 3 shows that in all periods the
overwhelming majority of the global population (around 95%) was disaggregated into and in proportion to
mapped BU, and only small shares were rasterized via points (which may be in BU cells) or areal weighted
(AW). This shows that GHSL allowed for a very significant and effective increase in the spatial resolution of
input population layers.
Assessing global population distribution and volcanism
Overall global quantification of population in proximity to volcanoes is reported in Tables 4 and 5.
1990
2000
Change
90-2000
2015
Change
2000-2015
Pop. [106]
%
Pop. [106]
%
%
Pop. [106]
%
%
Within 100 km
648
12.2
768
12.5
18.5
939
12.8
22.3
Total
5,310
100
6,127
100
15.4
7,351
100
20.0
Table 4. Population within 100 km of Holocene volcanoes (VD)
Results show that a very significant proportion of the global population lives within 100 km of Holocene
volcanoes (VD) and that since 1990 this proportion has been increasing at a rate higher than that of the global
population change. In 2015, almost 13% of the global population is estimated to live within a range of potential
direct impact of volcanic eruptions.
1990
2000
Change
90-2000
2015
Change
2000-2015
Pop. [106]
%
Pop. [106]
%
%
Pop. [106]
%
%
Within 100 km
302
5.7
349
5.7
15.4
414
5.6
18.7
Total
5,310
100
6,127
100
15.4
7,351
100
20.0
Table 5. Population within 100 km of volcanoes with significant eruptions (SVED)
Concerning volcanoes with significant eruptions (SVED), the proportion of the global population living within
100 km has decreased slightly from 2000 to 2015 (to 5.6 %), although absolute values have increased by 18.7%
in this period to total 414 million people, following global population rise.
A more detailed depiction of variation of population as a function of distance from volcanoes is presented in
Freire et al.
Volcanism
Figs. 6 to 9, in the form of global distance profiles.
Fig. 6. Cumulative population as a function of radial distance to Holocene volcanoes (VD), in 1990, 2000, and 2015.
Fig. 7. Cumulative population as a function of radial distance to volcanoes in Significant Volcanic Eruption Database
(SVED), in 1990, 2000, and 2015.
Figs 6 and 7 show that substantial populations live in proximity to volcanoes, and that this presence has
increased from 1990 to 2015. Concerning total Holocene volcanoes, it is estimated that in 2015 more than 130
million people live within 20 km of volcanoes, up from 93 million in 1990 (a 45% increase). When volcanoes
with significant eruptions are considered, the number of people within this zone is about halved, having
increased 37% since 1990.
Freire et al.
Volcanism
Fig. 8. Population density as a function of radial distance to Holocene volcanoes (VD), in 1990, 2000, and 2015.
Fig. 9. Population density as a function of radial distance to volcanoes in Significant Volcanic Eruption Database
(SVED), in 1990, 2000, and 2015.
Figs 8 and 9 show that relatively high population densities occur in the vicinity of all volcanoes, especially of
those with significant eruptions, and those densities have been increasing considerably since 1990. However, it
is at a distance of 10 to 25 km from Holocene volcanoes that the absolute increase in population density has
been greatest (additional 45 people/Km2), from 1990 to 2015.
In all periods population density increases with proximity to volcanoes, peaking at around 15-25 km, and then
decreases sharply (perhaps due to steep slopes) in their immediate vicinity, but still maintaining relatively high
densities. This pattern is even more striking for volcanoes with significant eruptions (SVED), where overall
population densities are higher (up to 300 people/Km2), with obvious implications for exposure, risk analyses,
and potential impacts. This pattern is dissimilar from the monotonically decreasing densities reported by Small
and Naumann, 2001. This difference may be due to the finer characteristics of the new population grids,
enabling sharper analyses.
CONCLUSIONS AND OUTLOOK
Better and finer global analyses of potential human exposure and risk of natural disasters require improved
geoinformation on population distribution and densities. By combining new geospatial datasets, novel global
Freire et al.
Volcanism
population grids were produced for different periods. The use of a simple, globally-consistent approach for
population disaggregation into built-up areas mapped from satellite imagery enables assessment and comparison
of changes in time. These grids have also higher spatial resolution than those previously available, allowing
conducting more detailed assessments of population potentially exposed to hazards, both natural and man-made.
These new grids were used to analyze and quantify the distribution of population in relation to recent volcanism.
This analysis updates and refines that of Small and Naumann (2001), but also enables monitoring of patterns
and detection of changes with time. Preliminary results concerning the overall global situation indicate that
almost 6% (414 million people) of the world’s 2015 population lived within 100 km of a volcano with at least
one significant eruption, and almost 13% (939 million) within 100 km of a Holocene volcano. This last amount
has increased by 45% since 1990, above the global population change rate (38%) for the period. Population
densities are also high in vicinity of volcanoes, on average peaking around a distance of 15-25 km, and these
have been increasing with time in last 25 years.
These new population grids could benefit a global volcano warning system, by improving estimation of
potential human impacts (Andredakis and De Groeve, 2015). In case of volcanic eruption, such geoinformation
could also be used to plan evacuation and estimate resources needed for response. While this work has focused
on volcanism, these grids can be combined with any type of hazard, enabling advancing modeling and analyses
at all stages of the emergency management cycle.
Results of this work will be further explored, namely by including more spatially detailed analyses (e.g. at
continent level, by volcano or volcanic regions), and by refining assessment of population in the immediate
vicinity of volcanoes where highest population densities are observed.
REFERENCES
1.
Andredakis I. and De Groeve T. (2015) Towards a global humanitarian volcano impact alert model
integrated into a multi-hazard system. In: Harris, A. J. L., De Groeve, T., Garel, F. & Carn, S. A. (Eds)
Detecting, Modelling and Responding to Effusive Eruptions. Geological Society, London, Special
Publications, 426.
2.
Brown, S.K., Auker, M.R. and Sparks, R.S.J. (2015) Populations around Holocene volcanoes and
development of a Population Exposure Index. In: S.C. Loughlin, R.S.J. Sparks, S.K. Brown, S.F. Jenkins &
C. Vye-Brown (Eds) Global Volcanic Hazards and Risk, Cambridge: Cambridge University Press.
3.
Center for International Earth Science Information Network - CIESIN - Columbia University (2014)
Gridded Population of the World, Version 4 (GPWv4). Palisades, NY.
http://www.ciesin.columbia.edu/data/gpw-v4
4.
Chester D. K., Degg M., Duncan A. M. and Guest J. E. (2000) The increasing exposure of cities to the
effects of volcanic eruptions: a global survey, Environmental Hazards, 2, 3, 89-103.
5.
Freire, S. (2010) Modeling of spatiotemporal distribution of urban population at high resolution—Value for
risk assessment and emergency management. In Geographic Information and Cartography for Risk and
Crisis Management; Springer: Berlin, Heidelberg, Germany, 53–67.
6.
Freire S., Kemper T., Pesaresi M., Florczyk F., and Syrris V. (2015) Combining GHSL and GPW to
Improve Global Population Mapping, 2015 IEEE International Geoscience & Remote Sensing Symposium
Proceedings, 2541-2543.
7.
GFDRR (Global Facility for Disaster Reduction and Recovery) (2011) Volcano hazard and exposure in
GDRFF priority countries and risk mitigation measures, Bristol University Cabot Institute and NGI Norway
for the World Bank: NGI Report 20100806, Bristol.
8.
GVP (Global Volcanism Program) (2013) Volcanoes of the World, v. 4.4.1. Venzke, E (Ed.). Smithsonian
Institution. Downloaded 21 Jan 2016. http://dx.doi.org/10.5479/si.GVP.VOTW4-201.
9.
Lerner-Lam, A., (2007) Assessing global exposure to natural hazards: progress and future trends, Environ.
Hazards, 7, 10–19.
10. Linard C. and Tatem A. (2012) Large-scale spatial population databases in infectious disease research,
International Journal of Health Geographics, 11:7.
11. Mondal P. and Tatem A.J. (2012) Uncertainties in Measuring Populations Potentially Impacted by Sea
Level Rise and Coastal Flooding, PLoS ONE, 7(10).
Freire et al.
Volcanism
12. Nakada, S. (2000) Hazards from pyroclastic flows and surges. In: Sigurdsson, H. (Ed.), Encyclopedia of
Volcanoes. Academic Press, San Diego, CA, pp. 945–956.
13. NOAA (National Oceanic and Atmospheric Administration) (2015) National Geophysical Data Center /
World Data Service (NGDC/WDS): Significant Volcanic Eruptions Database. National Geophysical Data
Center, NOAA. doi:10.7289/V5JW8BSH.
14. NRC (National Research Council) (2007) Tools and Methods for Estimating Populations at Risk from
Natural Disasters and Complex Humanitarian Crises, Report by the National Academy of Sciences,
National Academy Press, Washington, DC, 264 pp.
15. Peduzzi, P., Dao, H., Herold, C., and Mouton, F. (2009) Assessing global exposure and vulnerability
towards natural hazards: the Disaster Risk Index, Nat. Hazards Earth Syst. Sci., 9, 1149–1159,
doi:10.5194/nhess-9-1149-2009.
16. Pesaresi M. and Ehrlich D. (2009) A methodology to quantify built-up structures from optical VHR
imagery, in Global Mapping of Human Settlement Experiences, Datasets, and Prospects, P. Gamba and M.
Herold, Eds. Boca Raton, FL, USA: CRC Press, ch. 3, pp. 27–58.
17. Pesaresi, M., Guo Huadong; Blaes, X., Ehrlich, D., Ferri, S., Gueguen, L., Halkia, M., Kauffmann, M.,
Kemper, T., Linlin Lu, Marin-Herrera, M.A., Ouzounis, G.K., Scavazzon, M., Soille, P., Syrris, V.,
Zanchetta, L. (2013) A Global Human Settlement Layer From Optical HR/VHR RS Data: Concept and
First Results, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 6, 5,
2102-2131.
18. Pesaresi M., Ehrlich D., Ferri S., Florczyk A., Freire S., Halkia M., Julea A., Kemper T., Soille P. and
Syrris V. (2015) Operating procedure for the production of the Global Human Settlement Layer from
Landsat data of the epochs 1975, 1990, 2000, and 2014. JRC Technical Report, EUR 27741 EN.
19. Rodolfo, K.S. (2000) The hazard from lahars and Jokulhlaups. In: Sigurdsson, H. (Ed.), Encyclopedia of
Volcanoes. Academic Press, San Diego, CA, pp. 973–996.
20. Small, C. and Naumann, T. (2001) The global distribution of human population and recent volcanism,
Environmental Hazards, vol. 3, 93–109.
21. UNISDR (2013) Global Assessment Report on Disaster Risk Reduction 2013 - From Shared Risk to Shared
Value: the Business Case for Disaster Risk Reduction, United Nations Office for Disaster Risk Reduction.
Geneva, Switzerland.
22. Wright J. K. (1936) A method of mapping densities of population, The Geographical Review, 26, 1, 103–
110.

New Multi-temporal Global Population Grids – Application to

Transcription

Similar documents

Bang for the Buck- Effective Vocabulary Instruction

featuring Dr. Nita Freire

Volcanic hazard in East Africa

Volcanoes

March 2011 - Azusa Pacific University

Newsletter April 2014

VOLCANOES

Active Volcanoes of the Kurile Islands

International finalists Sustainable energy building