Ran Goldblatt, Trevor Monroe, Sarah Elizabeth Antos, Marco Hernandez at the World Bank Data Blog: “The desire of human beings to “think spatially” to understand how people and objects are organized in space has not changed much since Eratosthenes—the Greek astronomer
The increasing availability of satellite data has transformed how we use remote sensing analytics to understand, monitor and achieve the 2030 Sustainable Development Goals. As satellite data becomes ever more accessible and frequent, it is now possible not only to better understand how the Earth is changing, but also to utilize these insights to improve decision making, guide policy, deliver services, and promote better-informed governance. Satellites capture many of the physical, economic and social characteristics of Earth, providing a unique asset for developing countries, where reliable socio-economic and demographic data is often not consistently available. Analysis of satellite data was once relegated to researchers with access to costly data or to “super computers”. Today, the increased availability of “free” satellite data, combined with powerful cloud computing and open source analytical tools have democratized data innovation, enabling local governments and agencies to use satellite data to improve sector diagnostics, development indicators, program monitoring and service delivery.
Drivers of innovation in satellite measurements
- Big (geo)data – Satellites in Global Development are improving every day, creating new opportunities for impact in development. They capture millions of images from Earth in different spatial, spectral and temporal resolutions, generating data in ever increasing volume, variety and velocity.
- Open Source –Open source annotated datasets, the World Bank’s Open Data, and other publicly available resources allow to process and document the data (e.g. Cumulus, label maker) and perform machine learning analysis using common programming languages such as R or Python.
- Crowd – crowdsource platforms like MTurk, Figure-eight and Tomnod are used to collect and enhance inputs (reference data) to train machines to identify automatically specific objects and land cover on Earth.
- High Quality Ground Truth –Robust algorithms that analyze the entire planet require diverse training data, and traditional development Microdata for use in machine learning training, validation and calibration, for example, to map urbanization processes.
- Cloud – cloud computing and data storage capabilities within platforms like AWS, Azure and Google Earth Engine provide scalable solutions for storage, management and parallel processing of large volumes of data.
…As petabytes of geo data are being collected, novel methods are developed to convert these data into meaningful information about the nature and pace of change on Earth, for example, the formation of urban landscapes and human settlements, the creation of transportation networks that connect cities or the conversion of natural forests into productive agricultural land. New possibilities emerge for harnessing this data for a better understanding