D-SRGAN: DEM Super-Resolution with Generative Adversarial Networks

Digital elevation model (DEM) is a critical data source for variety of applications such as road extraction, hydrological modeling, flood mapping, and many geospatial studies. The usage high-resolution DEMs inputs in application areas improves the overall reliability accuracy raw dataset. goal this study to develop machine learning that increases spatial resolution DEM without additional information. In paper, GAN based (D-SRGAN), inspired by single image super-resolution methods, developed evaluated increase DEMs. experiment results show D-SRGAN produces promising while constructing 3 feet from 50 low-resolution It outperforms common statistical interpolation methods neural network algorithms.This shows it possible use power artificial networks also demonstrates approaches can be applied super-resolution.