"Africa's official maps are stuck in the past, often either outdated, incomplete-or both. But governments don't have the budgets to fix them, making it difficult to complete projects as complex as deciding where to put new solar plants to as simple as delivering a package. Now a new plan is underway to map the entire continent using satellite data and AI."
""Maybe 90% of African countries don't have access to an accurate current base map for their country," says Sohail Elabd, global director of emerging markets at Esri, the mapping company behind the Map Africa Initiative. At a United Nations event last year, Elabd met the heads of national mapping departments from around 30 African countries. Nearly all said they didn't have accurate base maps-the foundational maps that are critical for everything from urban planning to disaster response."
"On the plane ride home, he started thinking about what satellite data could make possible. Traditionally, mapping a country required flying specialized planes with sensors and cameras over the land for a month or more, along with extensive data collection on the ground. "Usually it's a very costly process and time consuming," says Elabd. Now AI can analyze satellite data and create detailed maps at a fraction of the cost."
Many African countries lack accurate, current base maps because national mapping agencies do not have sufficient funding. Traditional national mapping used expensive aerial surveys and extensive ground data collection, making updates slow and costly. A new initiative will train AI to analyze satellite imagery to identify landscape features, buildings, roads and geographic boundaries across diverse terrains. The effort will also produce additional layers such as agricultural fields, urban infrastructure and vegetation, enabling crop-type and tree-species analysis. Automated, satellite-based mapping promises faster, cheaper, and more detailed geospatial data to support planning, disaster response and infrastructure decisions.
Read at Fast Company
Unable to calculate read time
Collection
[
|
...
]