Improving ALOS Digital Elevation Model Using ICESat-2 Data and Random Forest

Document Type : Original Article

Authors

1 School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran

2 School of Surveying and Geospatial Engineering, Collage of Engineering, University of Tehran, Tehran, Iran

10.22059/eoge.2026.409807.1207

Abstract

Digital Elevation Models (DEMs) are fundamental analytical tools in environmental studies, urban development, natural change monitoring, and spatial analysis. These models are typically generated using remote sensing technologies, satellite imagery, and LiDAR measurements. ICESat-2, a state-of-the-art satellite developed by NASA, employs advanced LiDAR sensors and a unique orbital design to provide high-resolution elevation data. In this study, an integrated approach is presented to enhance the accuracy of ALOS-derived DEMs by incorporating ICESat-2 elevation data (ATL08 product) and the Random Forest machine learning algorithm. The research was conducted over Avajiq County in West Azerbaijan Province, using data from 2018 to 2023. ICESat-2 data were first used as a reference dataset. By extracting topographic features and applying the Random Forest model, the original DEM errors were corrected. Results indicate a significant improvement in elevation accuracy, with an RMSE of 0.0319, R² of 0.9727, and MAE of 0.0198. A comparison with elevation data from the National Cartographic Center of Iran further validated the performance of the proposed method, where RMSE, R², and MAE were reported as 2.35, 0.98, and 1.05, respectively.

Keywords

Main Subjects

Files

Share

How to cite

Statistics