Analyzing historical maps with AI to inform sustainable land use planning

Land use decisions – designing a city, building a highway, or designating land for conservation – can have long-term consequences on environmental and public health. But the long-term effects of these decisions are often poorly understood, in part because historical land use data is scarce. For example, the impact of increasing urbanization in East African cities on local ecosystems, energy use, and quality of life is largely unknown. Millions of hand-drawn and printed maps capturing historical land use exist worldwide, but extracting usable data from them currently requires hundreds of hours of manual labor per map. This project will develop the first AI foundation model capable of transforming scanned historical map images into structured geographic data — converting what is currently an inaccessible visual archive into an analytical resource that researchers and policymakers can use to understand the long-term consequences of land use decisions at global scale.

Above: Travel planning with a map. Image credit:  seb_ra / iStock