About Us

EarthDefine began in 2013 with a conviction about scale. We had spent years at the large geospatial firms, where every map product was made the same way: one small area at a time, on desktop software, at high cost. Our grounding is from the University of Michigan, home to the nation's first School of Forestry and Conservation, and it carried into everything that followed. We believed the same map work could be done at nationwide scale, with the lower cost passed along to the people using the data, and in 2013 we set out to build it that way.

Our first nationwide products were the National Tree Map and Building Footprints. We built the technology to make them possible, using advances in machine learning to map at a scale desktop workflows could never reach. Both are still available today, at greater precision and wider coverage (see Tree Map and Buildings API). The company grew from there, and today we map the entire United States several times a year.

We build our nationwide products from the best public imagery and elevation data available: NAIP aerial imagery, state orthoimagery, and USGS LiDAR, programs that represent billions of dollars of federal and state investment. The classification and quality control that turn those sources into finished, accurate data are our own work. Because the data is derived from public sources rather than bought from a commercial vendor, it carries none of the redistribution restrictions that commercial-imagery derivatives inherit, and none of the pass-through cost.

We refresh continuously, folding in the newest high-resolution sources as they are released: NAIP imagery as each state updates it, state orthoimagery whenever it appears (we check every state by hand, since the schedules are unpredictable), and new LiDAR as it becomes available. Release schedules vary by product, but the principle behind them does not, because our data is only worth trusting when it reflects the ground as it is now.

We also stay close to the research, testing new models as they are published, running our own results through human review to confirm they hold up on real samples, and keeping only the methods that measure up. Every product ships with published, product-specific accuracy metrics, validated against real ground conditions. Accuracy is the part of the work we treat most seriously, and it is where most of our care goes.

The product line grew by demand, each dataset answering a customer need no one else could meet. Today we offer nine nationwide datasets across the land surface, the built environment, and vegetation, with a tenth in development that maps driveways and their access points from the road. Federal agencies, state governments, and private companies rely on this data for consequential decisions, and partners such as PlanIT Geo and Regrid build on it. Nonprofits facing financial barriers receive all of it at half price, a practice that goes back to our founding.

We are a team of analysts, engineers, and data scientists, and alongside our off-the-shelf datasets we take on custom work for more complex geospatial problems.

If you want to see how the data holds up on the area you care about, tell us what you are trying to map and we will show you what is there.