Development of the Global Exposure Database (GED)

Development of the Global Exposure Database (GED)

Kishor Jaiswal, Synergetics Inc./USGS Golden COwith contributions from: P. Gamba, University of Pavia, Italy, C. Huyck and Z. Hu, ImageCat Inc.S. Vinay, R. Chen, and M. Becker, CIESINO. Odhiambo, G. Mboup UN-Habitat, Nairobi KenyaS. Ferri, E. Goldoni, D. Ehrlich JRC ItalyP. Henshaw, GEM FoundationD. Wald, USGS Golden CO

10NCEE, Anchorage Alaska

July 23rd, 2014

@GEMwrld #10NCEE

Objectives

• The aim of GED4GEM is to build a comprehensive, multi-scale andstatistically accurate database of population and buildings, to asses thephysical and economic exposure of a given area to earthquakes.

• The database had therefore to be:– state-of-the-art, i.e. including all existing (and freely available) data sets;– global, i.e. valid all for each country;– consistent, i.e. capable in providing statistical and spatial consistency (in

a region or a country);– easily upgradable through ad-hoc scripts.

• Including more information than just the building structural data, the GlobalExposure Database (or GED for short) could eventually be useful inmulti-hazard applications, e.g., earthquakes, floods, landslides,hurricanes and other disasters.

Consortium

Partners• University of Pavia (UNIPV)• The Center for International Earth Science Information Network (CIESIN)• Global Urban Observatory (GUO) of UN-HABITAT• ImageCat Inc.• The Joint Research Centre of the European Union (JRC)

Advisory Partners• US Geological Survey (USGS)• EUCENTRE• Geoscience Australia (GA)

Source Taxonomy Grid/Vector

Statistics Validation

Level 0 GPW, PAGER,GRUMP, UN-HABITAT, NERA

PAGERGEM

30” Country Internal:consistency with PAGER

Level 1 Sub-country db(Census, DHS, MICS, HAZUS, regional programmes)

GEMHAZUS

30” Region (Admin 1 & Admin 2)

Internal: test site informationat aggregated levels

Internal: quality of input data

Level 2 National/regional/local database(s)

GEM 30” Ad hoc

Level 3 Ground surveyBuilding database(s)

GEM vector Single building

External:regional and selected users

Global Exposure Database: levels

GED Level 0: data from PAGER

IMPROVED Level 0

• Additional information available from UN-Habitat to improve GED has been processed and sample results checked before ingestion into GED.

Census records

• Sample design– Systematic sample of every twentieth household.

• Sampling unit: Households• Sample fraction: 5%• Sample size (person records): 1,407,547• Sample weights: Self-weighting.

Expansion factor = 20.

Demographics and Health Survey records

• The Demographics and Health Survey (DHS) sample is designed torepresent each of the country’s administrative regions. In each region, astratified sample design was employed. Primary sampling units (PSUs) areselected with probability proportional to the estimated number of householdsfrom the Census.

Level 1: for the first time sub-national information

struct_code struct_ratio

W+WLI//R99 0.026

CR+CT99//RC+RC99 0.001

MUR+STRUB+MOM//R99 0.370

MUR+CL99//R99 0.043

MUR+CL99//RO 0.558

MUR+ADO//R99 0.002

struct_code struct_ratio

W+WLI//R99 0.020

CR+CT99//RC+RC99 0.00

MUR+STRUB+MOM//R99 0.375

MUR+CL99//R99 0.040

MUR+CL99//RO 0.558

Level 1: less coverage

• Check for region matching included (issues with GADM versus the population model versus national databases)

• GADM v.2 compliant

Level 2 data: room for detailed databases

struct_id struct_code struct_ratio

1 W+WLI//R99 0.026

9CR+CT99//RC+RC

99 0.001

12MUR+STRUB+MO

M//R99 0.370

19 MUR+CL99//R99 0.043

31 MUR+CL99//RO 0.558

102 MUR+ADO//R99 0.002

103MATO//RME+RM

E99 0.000

Stone 4.3%Cane/palm 0.7%Adobe / "tapial" 37.0%"Tabique, quinche" 0.1%Wood 1.9%Bricks 55.8%OTHER 0.2%

Level 2: aggregated data from existing GIS files

Guadeloupe: density of buildings + dwelling fractions from Level 0 (JRC + UNIPV)

Replacement cost data sources

• Published construction cost guides– Common in countries like

Europe, North America, Australia, et

– Available for a selection of countries in Africa, South America, Asia

• Purpose commissioned reports from local quantity surveyors

Example: Malaysia

Procedure

• It is proposed that the global range of GDPpc is subdivided into five bins and an index country (together with a full range of factors) is provided for each bin.

• Then, a factor for replacement rate is computed.

Current Replacement Cost Coverage

• A few countries with detailed information by expert opinion• Rest of the world (almost) with “default” values

A few examples of data in GED

A few question we can answer with GED 1.0

Level 0 (national) questions• Estimate of the total residential exposure of Russia

– X billions USD (computed using level 0 dwelling fractions from PAGER, average floor per capita, default replacement cost)

Level 1 (subnational) questions• Estimate of total wooden buildings that are present in the Saravan region of

Laos– Y (computed using level 1 dwelling fractions from 2006 MICS survey, the

average number of people per dwelling, default numbers for the numbers of dwelling per building)

Level 2 (local) questions• Estimate of total masonry buildings in a radius of 3 km around the center of

Brisbane – Lat. 153.03, Lon. -27.44, – or how much would cost to rebuild 60% of them?– Z and XX billion AUS (computed using level 2 data from NEXIS)

GEM Data and Models on the Platform

Thank you !

Except where otherwise noted, this work is licensed under: creativecommons.org/licenses/by-nc-nd/4.0/

Please attribute to the GEM Foundation with a link to -www.globalearthquakemodel.org

Questions?

• Just to help understanding the answers …