Level 1 exposure for the  

Please click on the map to show a distribution of costs, area and building count by construction type.

To download data for individual countries, please see the METEOR project data download page.

This Level-1 data set provides an estimate of building exposure data for the purposes of loss estimation and CAT modeling. It was developed on the METEOR project (Ghosh et al., 2018) for 47 ODA (Official Development Assistance) countries, as determined by the OECD-DAC (Organization for Economic Cooperation and Development, Development Assistance Committee). Without formal protocols for exposure development, risk management decisions in ODA countries may rely on loss modelling results suffering from less than adequate input data that may not be sufficient to address UN SDG and SFDRR targets and goals.

In this data set, for each 15 arcsecond (approximately 500 meters squared at the equator), the data provides an approximate value of the building stock in 2020 USD. A linked CSV file provides an estimate of the livable area in square meters, by structure type conforming to the GEM taxonomy (Brzev et al., 2013). A summary of the process of generating the data is provided below.

The building stock valuation estimates are generated using population figures from “Population Dynamics” provided by the UN Department of Economic and Social Affairs. Population is disaggregated from the national to the regional level primarily using LandScan 2012 data. The data is then spread to each 15 arcsecond cell using a combination of techniques. Digitized patterns corresponding to regional construction patterns are used in a machine learning process (Support Vector Networks (Cortez et al., 1995)) to develop a model for classifying grids. This model is applied to assign development patterns throughout each country, which informs the estimated building density, replacement cost per meter, and structural characteristics of the building stock, as described below.  Segmentation of development patterns in each county uses various EO data sets and weights developed moderate and high-resolution data sets, including:

1) NOAA night-time light annual composite (VIIRS) 15-arcsecond grid cell (Earth Observation Group NOAA-NCEI (2015))

2) Oak Ridge National Laboratory Landscan ambient population (LSCAN) resampled from the 30-arcsecond grid cell to 15-arcsecond (Oak Ridge National Laboratory. (2012))

3) JRC Global Human Settlement Layer (GHSL-Landsat) derived from Landsat imagery resampled from the 250-meter grid cell to 15-arcsecond (Corbane et al.,(1) 2018)

4) DLR Global Urban Footprint (GUF) resampled from 12-meter grid cells to a 15-arcsecond percentage of human presence raster grid (DLR Earth Observation Center. (2016))

5) JRC GHSL derived from Sentinel-1 SAR (GHSL-SAR) resampled from 20-meter grid cells to a 15-arcsecond percentage of human presence raster grid (Corbane et al.,(2) 2018)

6) CIESIN-Facebook High Resolution Settlement Layer (HRSL) resampled from 1-arcsecond grid cells to a 15-arcsecond percentage of human presence raster grid (Facebook Connectivity Lab and Center for International Earth Science Information Network - CIESIN - Columbia University. (2016))

7)  JRC mosaiced Sentinel-1 dual polarization bands (SAR B1, B2, B3) resampled from 20-meter resolution to a 15-arcsecond mean raster grid per band and a maximum mean value of the 3 bands (SAR-MaxMean). (Syrris, V., et. al (2018)

8) Gridded population from WorldPop 2020 resampled from 3-arcsecond grid cells 15-arcsecond (WorldPop(1))

9) An indicator of OSM data throughout the country- building count, area, and maximum building height were calculated from building footprint polygons and aggregated up to create 15-arcsecond grids. (ImageCat, Inc. (2019))

Population is then converted to an estimate of the number of households using the best available estimates of persons per household at the country level collected from various sources including IPUMS (Minnesota Population Center, 2019) and the UN. The area per household was estimated based on an extensive assessment of OSM building footprint data and screening based on likely size for buildings. Within a given country, the average size of single family households was often found to be reasonable, and an average was used for development patterns that are primarily residential. For some countries, there was scant data or observation bias. Thus, in countries where there was a value outside of one standard deviation from the regional mean, the regional average size was used per development pattern. Regions are: 1) Southeast Asia (Bangladesh, Bhutan, Cambodia, Lao People's Democratic Republic, Myanmar, Nepal, Timor-Leste), 2) Middle East and North Africa (Afghanistan, Sudan, Yemen 3) Eastern Africa (Burundi, Comoros, Djibouti, Eritrea, Ethiopia, Madagascar, Malawi, Mozambique, Rwanda, Somalia, South Sudan, Uganda, United Republic of Tanzania, Zambia), 4) the rest of Africa (Angola, Benin, Burkina Faso, Central African Republic, Chad, Democratic Republic of the Congo, Gambia, Guinea, Guinea-Bissau, Lesotho, Liberia, Mali, Mauritania, Niger, Sao Tome and Principe, Senegal, Sierra Leone, Togo), and Island nations (Haiti, Kiribati, Solomon Islands, Tuvalu, Vanuatu). For urbanized areas corresponding to capital cities where larger multi-family structures dominate and industrialized areas, regional averages were used per development pattern.

Replacement Cost was determined based on Huyck & Eguchi, 2017 and Huizinga et al., 2017. From Huizinga et al., the replacement cost of residential, commercial and industrial structures is estimated as a function of GDP.  These were updated to 2019 USD ( x 1.08) per meter squared and then adjusted based on development pattern based on Huyck & Eguchi, which found that in developing countries residential construction in urban areas was well represented, but that semi-engineered and engineered properties were over predicted when compared to Resettlement Action Plans and local expert opinion. Outer-city areas and smaller settlements are downscaled to 50% of the expected value, and rural construction is 50% of this figure. Urban areas consistent with capital city areas are consistent with the 2019 scaled values from Huizinga et al, 2017, as are industrial areas. These scaling factors are applied in accordance with the development patterns mentioned above.

The distribution of buildings by structure type are adjusted for each country. Housing census surveys and available ground imagery (Google Earth, Mapillary, ground photos, etc.) are first sourced to identify typical construction materials and structural systems prevalent within the country. Various engineering websites and standards (World Housing Encyclopedia [WHE], Prompt Assessment of Global Earthquakes for Response [PAGER] (Wald, et al. 2008), Global Earthquake Model [GEM] (Brzev et al., 2013)) are then sourced to validate and establish a preliminary structure type distribution, and these are refined by development pattern for level 1 data using expert judgement, a review of insitu and EO data for each country.

 

AECOM. (2017). Africa Property & Construction Cost Guide 2017, Offering global expertise and tailored local solutions in more than 150 countries.

Africa Property & Construction Cost Guide. (2017).

Breiman, L. Random Forests. Machine Learning 45, 5–32 (2001).

Breiman, L., Friedman, J.H., Olshen, R., and Stone, C.J. (1984). Classification and Regression Tree. Wadsworth & Brooks/Cole Advanced Books & Software, Pacific California

Brzev, S., Scawthorn, C., Charleson, A. W., Allen, L., Greene, M., Jaiswal, K., and Silva, V. (2013). GEM Building Taxonomy (Version 2.0) (No. 2013-02). GEM Foundation.

Corbane, C.,Florczyk, A., Pesaresi, M., Politis, P. and Syrris, V. (2018): GHS built-up grid, derived from Landsat, multitemporal (1975-1990-2000-2014), R2018A. European Commission, Joint Research Centre (JRC) doi: 10.2905/jrc-ghsl-10007 PID: http://data.europa.eu/89h/jrc-ghsl-10007

Corbane, C., Politis, P., Syrris, V. and Pesaresi, M. (2018): GHS built-up grid, derived from Sentinel-1 (2016), R2018A. European Commission, Joint Research Centre (JRC) doi: 10.2905/jrc-ghsl-10008 PID: Retrieved from http://data.europa.eu/89h/jrc-ghsl-10008

Cortes, C., Vapnik, V. Support-vector networks. Mach Learn 20, 273–297 (1995).

De Bono, A. & Chatenoux, B.  (2015). A global exposure model for GAR 2015. UNEP-GRID, GAR.

DLR Earth Observation Center. (2016). Global Urban Footprint (GUF). Available at https://www.dlr.de/eoc/en/desktopdefault.aspx/tabid-11725/20508_read-47944/

Earth Observation Group NOAA-NCEI (2015). Version 1 VIIRS Day/Night Band Nighttime Lights- 2015 Nighttime Light Annual Composite [dataset]: https://ngdc.noaa.gov/eog/viirs/download_dnb_composites.html

Esch, T., Bachofer, F., Heldens, W., Hirner, A., Marconcini, M., Palacios-Lopez, D., Roth, A., Üreyen, S., Zeidler, J., Dech, S., and  Gorelick, N., 2018. Where we live—a summary of the achievements and planned evolution of the global urban footprint. Remote Sensing 2018, 10.

Esch, T., Heldens, W., Hirner, A., Keil, M., Marconcini, M., Roth, A., Zeidler, J., Dech, S., Strano, E., 2017. Breaking new ground in mapping human settlements from space – The Global Urban Footprint. ISPRS Journal of Photogrammetry and Remote Sensing 134 (2017) 30-42.

Esch, T., Schenk, A., Ullmann, T., Thiel, M., Roth, A., Dech, S., 2011. Characterization of Land Cover Types in TerraSAR-X Images by Combined Analysis of Speckle Statistics and Intensity Information. IEEE Transactions on Geoscience and Remote Sensing, Volume 49, Issue 6, pp. 1911-1925.

European Commission, 2019. Global Human Settlement Layer (GHSL) data sets, available at https://ghsl.jrc.ec.europa.eu/datasets.php (last accessed August 20th 2019).

Facebook Connectivity Lab and Center for International Earth Science Information Network - CIESIN - Columbia University. (2016). High Resolution Settlement Layer (HRSL). Source imagery for HRSL © 2016 Digital Globe

German Aerospace Center (DLR), Earth Observation Center, 2016. Global Urban Footprint (GUF), available at https://www.dlr.de/eoc/en/desktopdefault.aspx/tabid-9628/16557_read-40454/ (last accessed 20 August 2019 ).

Ghosh, S., Huyck, C.K., Eguchi, R.T., Jordan, C., Smith, K.B., Silva, V., O'Hara, M.E., and Simon, C., 2018. Modelling Exposure through Earth Observation Routines (METEOR) for Developing Countries: Increasing Availability and Access to More Robust Risk Information. In AGU Fall Meeting 2018. 10-14 December 2018. Washington, D.C.

Henshaw, P., Silva, V., and O’Hara, M. (2018). GED4ALL Global Exposure Database for Multi-Hazard Risk Analysis. D1-Exposure Database Schema and Complementary Tools 2017-10, X pp., GEM Foundation, Pavia, Italy.

Huizinga, J., de Moel, H., & Szewczyk, W. (2017). Global flood depth-damage functions: Methodology and the database with guidelines (No. JRC105688). Joint Research Centre (Seville site).

Humanitarian OpenStreetMap Team (HOT). (2019). In-situ structural building type, height, and footprint area sampling polygons of Tanzania [dataset]

Huyck, C.K. and Eguchi, M.T. (2017). GFDRR Africa Disaster Risk Financing-Result Area 5 Exposure Development. Replacement Cost Refinements to the Exposure Data. Prepared for World Bank/GFDRR.

C. Huyck, Z. Hu, P. Amyx, G. Esquivias, M. Huyck, M. Eguchi (2019) METEOR: Exposure Data Classification, Metadata Population and Confidence Assessment. Report WP3.2.

ImageCat, Inc. (2019). OSM building footprint data aggregation to 3-arcsecond raster grid [dataset]. Unpublished

Jaiswal, K. and Wald, D. (2014). PAGER Inventory Database v2.0.xls. Golden, CO: United States Geological Survey (USGS).

Minnesota Population Center. Integrated Public Use Microdata Series, International: Version 7.2 [dataset]. Minneapolis, MN: IPUMS, 2019. https://doi.org/10.18128/D020.V7.2

Oak Ridge National Laboratory. (2012). LandScan High Resolution Global Population Data Set. https://landscan.ornl.gov/

OpenStreetMap contributors, 2019. OpenStreetMap data. Available under Open Database License, www.openstreetmap.org/copyright, available at https://www.openstreetmap.org/#map=4/38.01/-95.84  (last accessed 20 August  2019).

Parajuli, Y.K., Bothara, J.K. and Upadhyay, B.K. (2003). Housing Report-Uncoursed rubble stone masonry walls with timber floor and roof. Nepal. World Housing Encyclopedia–an Encyclopedia of Housing Construction in Seismically Active Area of the World Retrieved from https://www.eeri.org/lfe/pdf/nepal_uncoursed_rubble_stone.pdf

Pesaresi, M., Ehrilch, D., Florczyk, A.J. Freire, S., Julea, A., Kemper, T., Soille, P. and Syrris, V. (2015). GHS built-up grid, derived from Landsat, multitemporal (1975, 1990, 2000, 2014). European Commission, Joint Research Centre (JRC) [Dataset] PID. Available at http://data.europa.eu/89h/jrc-ghsl-ghs_built_ldsmt_globe_r2015b

Pesaresi, M., Ehrilch, D., Florczyk, A.J., Freire, S., Julea, A., Kemper, T., Soille, P. and Syrris, V. (2015): GHS built-up grid, derived from Landsat, multitemporal (1975, 1990, 2000, 2014)

Silva, V., Henshaw, P., Huyck, C., O’Hara, M., 2018. GED4ALL: Global Exposure Database for Multi-Hazard Risk Analysis; D5—Final Report; GEM Technical Report 2018-05; GEM Foundation: Pavia, Italy.

Syrris, V., Corbane, C., Pesaresi, M. and Soille, P. (2018). Mosaicking Copernicus Sentinel-1 Data at Global Scale. in IEEE Transactions on Big Data. doi: 10.1109/TBDATA.2018.2846265 URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8428406&isnumber=7153538