Mapping Poverty from Space with the World Bank
By our estimates, in a few years we will have so much daily satellite imagery that it would require the population of New York City — about eight million people — to stare at new imagery 24 hours, seven days a week, in order to just look at each photo, never mind analyze them. So as more industries turn to space to extract insights from satellite imagery, the question at the forefront is: How do we analyze all of that data?
Today’s rapid growth in satellite imagery reminds me of the internet boom decades ago. While satellite imagery used to be for the privileged, it is now increasingly accessible to companies and organizations across all industries. When internet data exponentially multiplied, we saw analytics, algorithms and data mining come into the conversation as solutions for understanding the data. Now, we’re seeing the same solutions arise with satellite data.
At Orbital Insight, we’re applying computer vision and deep learning algorithms to satellite imagery to create a big-picture understanding of the world that is quantitatively grounded in observation. We observe and monitor socioeconomic indicators, such as buildings, cars, planes, trains and ships, within millions of images, to see how the world is changing.
One of the reasons I joined Orbital Insight is because of the company’s clear commitment to humanitarian issues. Having spent some time in the nonprofit world, this was important to me. I’m proud of the fact that we’re working with nonprofit organizations to help them better monitor global issues like climate change and poverty.
The World Bank is a vital source of financial and technical assistance to developing countries around the world. For decades, the World Bank has relied heavily on data from household surveys to understand poverty and economic activity and to help policymakers focus interventions on the areas of greatest need. However, the process of manually acquiring this type of information is expensive and time-consuming, and censuses are typically only conducted once every ten years. Since economic conditions can change quickly, these valuable maps can become dated fast.
For this reason, we partnered with the World Bank to research whether at-scale geospatial analysis of satellite imagery could replace census data for generating more timely local poverty estimates. Together, we began to explore which features within satellite imagery could best supplement local poverty data in Sri Lanka. Orbital Insight automated the quantification and contextualization of several geospatial features, including measuring buildings and building height, counting car density and mapping agricultural land.
The theory was that all of these “signals” would provide useful indicators of economic activity and poverty at the local level. For example, changes in buildings and building height indicate development and construction, and therefore could be a proxy for population growth and migration. The number of cars can indicate which neighborhoods are more prosperous. Agricultural analysis can reveal information about food health and productivity, as well as help us gain a deeper understanding of rural and urban areas.
The results of Orbital Insight’s collaboration with the World Bank indicate that analysis of satellite imagery has great potential for quantifying poverty across countries. Of the “signals” and algorithms listed above, building density was the leading correlate of poverty in both urban and rural areas. For the full results of the World Bank’s findings, read the organization’s white paper.
Following the success of this initial project in Sri Lanka, we are expanding our work with the World Bank to Mexico. It is our hope that by making poverty data collection easier and more efficient, the World Bank will be able to more accurately respond to local needs across the globe. As we move forward with studying the various signals we can measure and their correlation to existing data, we hope to continue refining our technique.
As I look back to my professional experience before joining Orbital Insight, I had always been fascinated with how technology could strengthen humanitarian work — from securely communicating with my staff across borders as we helped North Korean refugees escape the country to analyzing human trafficking data within U.S. borders. At Orbital Insight, I’m excited to continue this journey as we explore how geospatial analytics can help with the many important global issues we face today.
For more information, read the full World Bank and Orbital Insight case study here.
