Deep Neural Network Regression for Normalized Digital Surface Model Generation with Sentinel-2 Imagery

Deep Neural Network Regression for Normalized Digital Surface Model Generation with Sentinel-2 Imagery

November 15, 2023

Our EAGLE student Konstantin Müller published together with our chairholder of the Deparment of Global Urbanization and Remote Sensing, Hannes Taubenboeck an article about DL for surface model generation. The article explores methods to extract high-resolution normalized digital surface models (nDSMs) from low-resolution Sentinel-2 data, enabling the creation of large-scale models. Leveraging the open access and global coverage of Sentinel 2, the study employs deep learning models, based on the U-Net architecture, with tailored multiscale encoders and conformed self-attention to achieve a mean height error of approximately 2m.

from the abstract: In recent history, normalized digital surface models (nDSMs) have been constantly gaining importance as a means to solve large-scale geographic problems. High-resolution surface models are precious, as they can provide detailed information for a specific area. However, measurements with a high resolution are time consuming and costly. Only a few approaches exist to create high-resolution nDSMs for extensive areas. This article explores approaches to extract high-resolution nDSMs from lowresolution Sentinel-2 data, allowing us to derive large-scale models. We thereby utilize the advantages of Sentinel 2 being open access, having global coverage, and providing steady updates through a high repetition rate. Several deep learning models are trained to overcome the gap in producing high-resolution surface maps from low-resolution input data. With U-Net as a base architecture, we extend the capabilities of our model by integrating tailored multiscale encoders with differently sized kernels in the convolution as well as conformed self-attention inside the skip connection gates. Using pixelwise regression, our U-Net base models can achieve a mean height error of approximately 2 m. Moreover, through our enhancements to the model architecture, we reduce the model error by more than 7%.

Link: https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10189905

you may also like:

Exploring the Interdisciplinary Potential of Arctic Research

Exploring the Interdisciplinary Potential of Arctic Research

The Arctic region, with its unique ecosystems and rapidly changing climate, presents a wealth of opportunities for interdisciplinary research.  Our colleague Prof. Marco Schmidt from the informatics department joined us in our recent field campaign. His research...

Exploring the Exposome: An Invited Talk at the DGG Convention

Exploring the Exposome: An Invited Talk at the DGG Convention

At the annual convention of the German Society for Vascular Surgery and Vascular Medicine (DGG) in Berlin, John Friesen from the EORC presented "Umweltfaktoren und kardiovaskuläre Gesundheit: Das Exposom in der modernen Gefäßmedizin" (Environmental Factors...

Arctic Ecology Research: Insights from the Recent Workshop

Arctic Ecology Research: Insights from the Recent Workshop

Our EORC staff members, Dr. Mirjana Bevanda and Jakob Schwalb-Willmann are currently participating in a workshop focused on Arctic ecology, organised by Prof. Larissa Beumer (UNIS). This workshop brought together international researchers dedicated to exploring the...

EORC Joins Forces to Organize 54th Annual GfÖ Conference

EORC Joins Forces to Organize 54th Annual GfÖ Conference

We are excited to announce that the Earth Observation Research Cluster is proudly co-organizing the 54th Annual Meeting of the Ecological Society of Germany, Austria, and Switzerland (GfÖ), scheduled for September 1-5, 2025, at the University of Würzburg. This event...

Privacy Policy

Lehrstuhl für Fernerkundung & Lehrstuhl für Urbane Fernerkundung

Erdbeobachtung an der Universität Würzburg