As daylight broke on Saturday morning, a satellite focused its long-range lens on Mandalay, Myanmar’s second-largest city, which had been hit hard by a powerful 7.7 magnitude earthquake the previous day. The aim was to snap images that, when analyzed with advanced artificial intelligence technology, could empower relief teams to quickly assess the extent of structural damage and locate areas where assistance was most urgent.
Initially, however, this sophisticated approach faced difficulties. “The clouds presented the biggest obstacle in this instance,” explained Microsoft’s chief data scientist. The technology could not penetrate the cloud cover, disrupting the view of the disaster-stricken area.
Thankfully, conditions improved, and hours later, another satellite operated by Planet Labs—a company based in San Francisco—captured the crucial aerial images and relayed them to Microsoft’s AI for Good Lab, based in Redmond, Washington. By then, even though it was almost 11 p.m. in Redmond, a team from Microsoft was eagerly awaiting the data.
Microsoft’s AI for Good Lab has previously employed AI-assisted damage assessment, monitoring events such as Libya’s severe flooding in 2023 or recent wildfires in Los Angeles. However, standard AI models couldn’t be used this time—they had to develop a specific model tailored for Mandalay due to the unique challenges each disaster presents.
“The geographic and disaster environments differ too much, and satellite imagery varies too widely to have one-size-fits-all solutions,” remarked Microsoft’s data scientist. Unlike fires that progress in more predictable patterns, earthquakes affect entire cities, complicating the task of immediately identifying where aid is essential.
Once the AI analysis wrapped up, it revealed that 515 buildings in Mandalay had sustained damage levels ranging from 80% to full destruction, and an additional 1,524 structures had suffered damage between 20% and 80%. This information not only underscored the disaster’s extensive impact but also helped pinpoint specific locations most in need of help. “For ground teams, this data is indispensable,” the data scientist noted.
While Microsoft stressed that this analysis should be deemed a preliminary guide requiring further on-site validation, it shared these findings with relief organizations like the Red Cross. Meanwhile, Planet Labs continues to capture images with its fleet of 15 satellites, documenting roughly a dozen sites across Myanmar and Thailand following Friday’s tremor.