Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2025
|
| Online Access: | https://doi.org/10.5281/zenodo.17791173 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Table of Contents:
- <p><span>The TERRAVISION project aims to boost sustainability and efficiency in the mining industry by using advanced Earth Observation (EO) technologies. It is designing an all-encompassing platform, linking satellite, airborne, and ground-based remote sensing data to support the whole value chain of raw materials from exploration to post-closure activities. Important goals of the project include modernisation of exploration activities, improvement of risk and hazard management, optimisation of extraction rates and guaranteeing environmental compliance. The report outlines the initial design of the proposed super resolution modelling architectures for the spatial enhancement of spaceborne data. This enhancement is essential for improving the efficiency of Copernicus Sentinel and other spaceborne data products employed in mining-related applications, where precise spatial information is crucial for decision-making and operational efficiency.</span></p> <p><span>The primary goal of this report is to present the initial design of the Machine Learning (ML)/Deep Learning (DL) Super Resolution (SR) framework that aims to enhance the spatial resolution of spaceborne data beyond the capabilities of the original imaging systems. This framework will enable the cross-platform utilisation of data from the Copernicus Sentinel missions and the PRecursore IperSpettrale della Missione Applicativa (PRISMA) satellite. It will leverage the unique characteristics of spaceborne data products by incorporating ML methods and Deep Neural Network (DNN) architectures to enhance the spatial characteristics of the observed properties, particularly in supporting mining operations. Hyperspectral pansharpening will be applied to improve the spatial resolution of PRISMA hyperspectral (HS) images from 30 meters to 5 meters. This enhancement will be accomplished by integrating the high spatial resolution panchromatic (PAN) band, which is acquired simultaneously and is included in the PRISMA datasets. The spatial resolution of Sentinel-2 multispectral (MS) images, captured by the MultiSpectral Instrument (MSI), will be enhanced to a minimum common resolution of 10 meters while maintaining the integrity of its spectral resolution. An ML-based thermal sharpening method will enhance the spatial resolution of coarse land surface temperature (LST) data products by fusing them with high spatial resolution MS data. Additionally, an ML-based multisensory data fusion technique will be implemented to estimate near surface-level concentrations of air pollutants at a higher spatial resolution of 1 km. Data from various sources will be utilised as input, including spaceborne data acquired from the Tropospheric Monitoring Instrument (TROPOMI) onboard the Sentinel-5 Precursor (Sentinel-5P) and high-accuracy in situ data collected from ground-based air quality monitoring stations, along with meteorological, land cover, and topographic data.</span> </p>