This is a project which is currently making use of HPC facilities at Newcastle University. It is active.
For further information about this project, please contact:
This project uses satellite datasets to detect paleochannels in the Middle East and North Africa (MENA region) that were foci for past human activity. This project develops cutting edge AI algorithms that integrate multi-modal earth observation datasets including very high-resolution datasets (VHR), to segment and classify arid landscapes.
This multidisciplinary project uses methods from computer science, geospatial data science, hydrology, archaeology and history. It develops large scale masks of paleochannels without the necessity of manual digitization and uses pre-processing methods such as edge detection to enhance paleochannels across vast landscapes in big satellite datasets.
Python-based geospatial and machine learning frameworks, including PyTorch, Rasterio, GDAL, HDF5, OpenCV, and Scikit-learn, will be used to process, analyse, and integrate large multi-modal Earth observation datasets. The workflow incorporates satellite imagery and terrain products, such as optical, topographic, and derived geospatial datasets, to support automated landscape analysis and palaeoenvironmental interpretation.
High-performance and parallel computing workflows will be developed for large-scale data handling, tiled image processing, and automated geospatial analysis across extensive study areas. Pre-processing pipelines will include the repeated parallel execution and evaluation of multiple feature-enhancement approaches, including edge-detection algorithms, texture filters, masking strategies, terrain derivatives, and image-transformation techniques, to optimise feature extraction from heterogeneous satellite and topographic datasets. Deep learning and computer vision approaches will then be applied to detect and reconstruct channels, while automated mosaicking and georeferenced stitching workflows will generate seamless large-area outputs suitable for GIS-based analysis and visualisation.