Researcher: Abraham Noah Wu

The current landscape of open-access DEM data in the world trades off quality with quantity. On one hand, free, open-access Global DEMs are only available in 30 m resolution. On the other hand, high-resolution, sub-2m DEMs are either locked behind a hefty paywall or are only available in selected regions of the world.

DEMGAN showed that the generated terrains are much more accurate than using traditional super-resolution algorithms because additional information is generated during the process. With much lower FID and RMSE compared to traditional super-resolution techniques, the results from DEMGAN could be used as substitutions where LiDAR-derived DEMs are unavailable