Measuring Relative Poverty and Deprivation around the Globe

The Global Gridded Relative Deprivation Index is the first product of its kind to cover the world at a 1-kilometer resolution.
This image shows a map of the African continent. Sections of the map are colored yellow, orange, red, and purple. These colors correspond to a scale of relative deprivation wherein yellow designates the highest deprivation and purple the lowest.
The Global Gridded Relative Deprivation Index, Version 1 (GRDIv1) dataset provides a global index value that characterizes the levels of multidimensional deprivation in each pixel, where a value of 100 represents the highest level of deprivation and a value of 0 the lowest. This map displays deprivation values for Africa at a spatial resolution of ~1 km, with purple indicating low deprivation and yellow indicating high deprivation. Credit: NASA SEDAC.

In September 2022, the World Bank updated its global extreme poverty line from $1.90 to $2.15 per person per day. According to the institution, anyone living on less than this amount is considered to be living in extreme poverty; as of 2019, about 648 million people around the globe were in this situation. 

Yet, while research shows that income is strongly correlated with deprivation in impoverished people’s lives, the World Bank acknowledges that income alone is an inadequate measure of poverty. In fact, as the institution notes in its Poverty and Shared Prosperity Report (2022), nearly 40% of multidimensionally poor individuals are not captured by surveys of monetary poverty. To remedy this, the bank has created what it calls a Multidimensional Poverty Measure, which is designed to capture a more holistic view of poverty by accounting for the lived experiences of people and the multiple deprivations they face beyond lack of income.

The recently released Global Gridded Relative Deprivation Index, Version 1 (GRDIv1) dataset from NASA’s Socioeconomic Data and Applications Center (SEDAC) takes a similar approach by offering measures of relative poverty and deprivation for each 30 arc-second (~1 km) pixel in a raster image, where a value of 100 represents the highest level of deprivation and a value of 0 the lowest. Using inputs of subnational data on human development (the subnational Human Development Index), infant mortality, and child dependency, as well as satellite-derived data on built-up areas, nighttime lights, and the change in nighttime lights over time, it is the first product of its kind to cover the entire world at a 1-kilometer (km) spatial resolution—a far higher resolution than previously possible with census or survey data alone.

“[The GRDIv1 dataset] really gives a granular level of detail into what’s going on at a subnational level, because we’re not limited and solely constrained by the subnational units typically associated with the sources of our data on infant mortality, child dependency ratio, and so on,” said SEDAC Deputy Manager Dr. Alex de Sherbinin. “So, you really could never drill down below any level below which poverty-related metrics were reported for any given country.”

Beyond its global coverage and high resolution, the GRDIv1 dataset is a noteworthy addition to SEDAC’s existing collection of poverty-related datasets.

“SEDAC participated in the Millenium Development Project in the mid-2000s and disseminated a number of subnational, poverty-related metrics as a collection, but I’ve had a long-standing interest in growing our collection and updating some of the metrics we are using,” said de Sherbinin. “I was looking at the growth in new metrics that were combining multiple sources of data, including satellite data. There were groups at [the University of California, Berkeley] and Stanford [University] that produced an Africa-wide dataset and another dataset that covered all the low- and moderate-income countries, but it struck me that there was a potential need for something that would cover all countries.”

To address that need, de Sherbinin and his SEDAC colleagues identified what data should be used to measure relative poverty deprivation. Ten datasets were selected from the best-available data that either continuously vary across space or have at least administrative level 1 (provincial/state) resolution, and that have global spatial coverage. (For more information about the datasets that were chosen, see page 4 of the GRDIv1 documentation.)

Data Set Data Source Time Frame Format Resolution
Gridded Population of the World, Version 4 (GPWv4): National Identifier Grid, Revision 11 (NIDv4.11) CIESIN, Columbia University 2010 Raster 30 arc-second (~1 km)
Gridded Population of the World, Version 4 (GPWv4): Basic Demographic Characteristics,Revision 11 (BDCv4.11) CIESIN, Columbia University 2010 Shapefile Best-available Admin unit
Global Subnational Infant Mortality Rates, Version 2.01 (IMRv2.01) CIESIN, Columbia University 2015 Raster 30 arc-second (~1 km)
High-Resolution Settlement Layer (HRSL) Facebook Connectivity Lab and CIESIN, Columbia University 2015 Shapefile 30 meters (~1 km)
CanadianBuildingFootprints, Version 1.1 (MS) Microsoft 2019 Shapefile N/A2
Gridded maps of building patterns throughout sub-Saharan Africa, Version 2.0 (ECOP) Dooley et al., 2021. Ecopia 2020 Raster 3 arc-second (~100 m)
OpenStreetMap (OSM) Geofabrik, 2018. Accessed 20211 Shapefile N/A 2
Subnational Human Development Index (SHDI), Version 4.0 Global Data Lab 2018 Shapefile Sub-national regions
Gridded global data sets for Gross Domestic Product (GDP) and Human Development Index (HDI) over 1990-2015, Version 2.0 Kummu, M et al., 2020. 1990 - 2015 netCDF 5 arc-minute (~9.3 km)
Annual Visible Infrared Imaging Radiometer Suite [VIIRS] Nighttime Lights (VNL), Version 2.0 Elvidge et al., 2021. Earth Observatory Group (EOG) 2012 - 2020 Raster 15 arc-second (~500 m)

1. OSM data are updated regularly; as a result, no temporal frame was given.

2. Developed using satellite imagery and drawing polygon vectors that have no resolution.

The input data listed in this table were harmonized into six components using Esri ArcGIS geoprocessing and R (R Studio) tools, and eventually aggregated into the final GRDIv1 dataset.

Then de Sherbinin worked with Juan Martinez, senior research staff assistant, and Kytt MacManus, analyst and geographic information systems (GIS) developer, at Columbia University’s Center for International Earth Science Information Network (CIESIN) to devise an approach for marrying sub-national socioeconomic and satellite data to create an index of relative deprivation. Using the Vulnerability Hotspot Mapping Method developed by CIESIN in 2015, they created a transparent, reproducible methodology wherein data inputs were spatially harmonized, indexed, and weighted into six components:

  • Child Dependency Ratio, a metric defined as the ratio between the population of children (ages 0 to 14) to the working-age population (age 15 to 64); a higher ratio implies a higher dependency on the working population and a younger, faster growing population
  • Infant Mortality Rates, a metric defined as the number of deaths in children under 1 year of age per 1,000 live births in the same year; infant mortality rates are a common indicator of population health and higher infant mortality rates imply higher deprivation
  • The Subnational Human Development Index, which attempts to assess human well-being through a combination of three dimensions: education, health, and standard of living; lower subnational human development indexes imply higher deprivation
  • Built Area, a metric based on the notion that global rural populations are more likely to experience a higher degree of multidimensional poverty compared to urban populations, with other things being equal; the ratio of built-up area to non-built-up area is considered a dimension wherein low values imply higher deprivation
  • Visible Infrared Imaging Radiometer Suite (VIIRS) Nighttime Lights Intensity, where the average intensity of nighttime lights for the year 2020 is interpreted as a dimension of poverty with lower values implying higher deprivation
  • VIIRS Nighttime Lights Slope, which is the slope of a linear regression calculated from annual VIIRS nighttime lights data between 2012 and 2020; higher values (increasing brightness) imply decreasing deprivation and lower values (decreasing brightness) imply increasing deprivation

These six components were then used to determine the degree of relative deprivation in the final index raster. (For a more detailed explanation of how the dataset was created, see the dataset documentation.)

SEDAC, one of NASA’s Earth Observing System Data and Information System’s (EOSDIS) 12 Distributed Active Archive Centers (DAACs), is operated by CIESIN in Palisades, New York. SEDAC serves as an “information gateway” between the socioeconomic and Earth science data and information domains and has extensive data holdings related to population, sustainability, and geospatial data. Through its efforts to synthesize Earth science and socioeconomic data and information, its data products are useful to a wide range of decision-makers and other applied users.

The GRDIv1 dataset is no exception, and by offering measures of relative poverty and deprivation at a higher resolution and using more inputs than other multidimensional poverty datasets, it is well-suited to inform the work of non-governmental organizations, international aid organizations, government leaders, policymakers and planners, Earth scientists, and others around the globe, including those working to address the United Nations 2030 Agenda on Sustainable Development. It is also likely to benefit emergency responders and humanitarian organizations looking to target the distribution of resources in the aftermath of disasters.

Further, like all SEDAC data, the GRDIv1 dataset can be integrated with other Earth observation data on natural hazards and climate impacts to identify hotspots of vulnerability and risk. For example, for a recent NASA Earthdata webinar, Martinez examined the relationship between GRDI and fine particulate matter (PM2.5) to better understand the relationship between exposure to air pollution and deprivation, and how it varies both among and within nations.

“At the continental level, the first thing you notice is that PM2.5 levels in places like Asia are much higher than in other parts of the world. But PM2.5 levels vary by country and you can see a massive range in places like China,” Martinez said. “China conforms to the expectation that in rural, remote areas, relative deprivation is highest in areas with relatively low PM2.5 levels, but in other parts of the world like Afghanistan and India, there are really high levels of PM2.5 exposure in highly deprived areas, which may have something to do with crop burning or traditional means of clearing land. So, as the world becomes more aware of these fundamental underlying spatial inequalities, these data may help us better understand how they intersect with natural hazards.”

GRDIv1 data can also be used to investigate socioeconomic questions, such as whether poverty levels are higher near protected areas in low-income countries.

“There’s a hypothesis that protected areas somehow deprive local populations of access to resources, but there are other narratives that suggest they may provide additional revenues to local populations," de Sherbinin said. “So, this is another area in which the GRDI data might be applied.”

The image shows a page from the StoryMap that explains how the GRDI dataset was developed. The page shows a partial map of Africa with yellow and orange dots. Above the map is text that explains what information the GRDI dataset provides. In the bottom left corner of the image, is a legend pertaining to degrees of relative deprivation. Yellow and orange correspond to high deprivation, red and purple signal lower amounts of deprivation.
A page from "The Story of a Data Set: Documenting the Development of GRDIv1," a StoryMap that explains what the GRDIv1 data show, describes how the data product was developed, and lets users interact with its data in ways static maps don’t allow. Credit: NASA SEDAC.

In addition to the GRDIv1 dataset being the first of its kind to cover the world at a ~1-km resolution, it is the first SEDAC dataset to be accompanied by an interactive StoryMap that explains what the data show, describes how the data product was developed, and lets users interact with the data in ways static maps don’t allow.

“What I like about the StoryMap is that I can show it to anyone with a mobile phone, and I can show them the actual data. It’s not a picture. You can zoom-in and see the details,” said Martinez, who created the interactive web application. “So, the StoryMap provides ways to practice the open science ethos that NASA promotes and start bridging the gap between the scientific community and the public, who may not have the background or the geospatial tools to help them use the data.”

As a participant in NASA’s Transform to Open Science (TOPS) initiative, which aims to make the science community more inclusive, Martinez is actively involved in the efforts to shrink that gap, and the StoryMap provides an easy way for him to share the GRDI dataset with people who might not see it.

“This [StoryMap] is the first one and I hope that in the near future every SEDAC dataset has some sort of landing page or something that people can access and explore with just the internet and a phone,” he said. “We continually use it because it’s so accessible and easy for people to understand, and we’re not providing anything different. It’s the same information presented in the [dataset] documentation.”

How to Access the Data

The Global Gridded Relative Deprivation Index, Version 1 dataset is available through Earthdata Search as well as from the SEDAC website. The website also provides users with several maps of GRDIv1 data, including Africa, Asia, Europe, North America, Oceania, South America, and the world.

Additional Information

  • GRDIv1 is an index that represents relative levels of deprivation, rather than absolute levels of deprivation. That is, it measures places in relation to one another as being more or less deprived. (For more on data considerations and limitations, see the “Limitations” section on page 12 of the dataset documentation)
  • The data include the final GRDIv1 and six component inputs and are available in GeoTIFF format at ~1 km resolution and can be used directly in mapping and geospatial analysis in tandem with other spatial datasets; users also have the option of downloading separate zip files for the GRDIv1 and its individual components
  • The downloads are compressed zip files containing: 1) GeoTIFFs, 2) Readme.TXT file, and 3) PDF Documentation
  • Downloaded Zip files must be uncompressed in a single folder using either WinZip (Windows file compression utility) or a similar application; users should expect an increase in the size of downloaded data after decompression
Last Updated