Satellite measurements can bridge data gaps by providing wide spatial coverage and near-continuous observations, enabling researchers to study areas that are otherwise inaccessible. Remote sensing technology has long been recognised as an essential tool for monitoring hydrological changes, especially in regions with limited ground-based data. By identifying spatiotemporal patterns, this technology enhances the detection of long-term trends and impacts, making it invaluable for environmental research and management in data-scarce regions.
Monitoring basin water changes is critical for informed management, particularly when understanding how dams influence water resources and their subsequent impacts on the sustainability of downstream environments in arid regions, but quite often traditional methodologies can fall short in such areas where data may be scarce and ground-based measurements limited.
To help overcome this and investigate the environmental impacts of dam construction in southern Saudi Arabia, remote sensing and machine learning techniques were used in four basins all characterised by arid conditions – the Hali, Baish, Yiba, and Reem.
A recent study by Almalki et al in the Journal of Hydrology spans from 2003 to 2020, highlighting long-term trends and dam construction impacts on environmental variables. The study demonstrates that dam construction in arid regions can lead to significant environmental transformations, particularly in vegetation, groundwater, soil salinity, and runoff dynamics. Using a combination of space-for-time substitution, remote sensing, and machine learning techniques, the research revealed substantial changes in key environmental variables. The findings highlighted basin-specific disruptions, particularly in groundwater and runoff, underscoring the localised hydrological impacts of dams.
The authors say the research provides valuable insights for understanding the long-term environmental consequences of dam infrastructure, especially in data-limited regions. The use of remote sensing allowed them to overcome data limitations, providing a robust framework for assessing long-term environmental trends through the space-for-time substitution technique. This innovative approach is particularly valuable in regions like Yiba, Hali, Reem, and Baish, where ground observations are limited.
Ethiopian data
Although Ethiopia is recognised as having abundant water resources, technical constraints and inefficient management in the water sector means it hasn’t been fully exploited and translated into basin planning and development.
As timely monitoring of surface water resources and detection of spatio-temporal changes is critical for sustainable use and management of the country’s water resources, a recent study explored the potential of machine learning to detect and monitor surface water bodies using Landsat data in a cloud computing platform at four sites in Ethiopia from 1986 to 2023.
The sites were selected due to the diverse availability of water resources including lakes, rivers, reservoirs and wetlands, as well as their complex interactions with human activities across different geographical locations and basins of Ethiopia. These were:
- Addis-Ziway which includes Addis Ababa and its surroundings, Bishoftu lakes, Ziway and Koka reservoirs.
- Hayq-Hashenge.
- Tana site which covers mainly Lake Tana.
- Abaya-Chamo site includes Arba-Minch and the lakes of Abaya and Chamo.
The results confirmed that machine learning using Landsat data produces reliable results for surface water monitoring and provides spatio-temporal information to support surface water management and water policy in Ethiopia. Mathias Tesfaye and Lutz Breuer say the results of their study may have important implications for the country’s policy makers and water resource planners, particularly in the context of sustainable water management and achieving regional sustainable development goals.
Modelling dam failure
A casualty of the Ukrainian and Russian war, the Kakhovka hydropower dam on the Dnipro River in Ukraine collapsed in June 2023, causing widespread flooding in Europe’s greatest industrial and ecological disaster for decades.
Once a major water resource supplier for more than 700,000 people, the reservoir had covered an area of 2000km2. As reliable data is not readily available in this war-torn country, scant details exist about how the reservoir rapidly emptied. However, understanding how it drained is crucial for both scientific research and societal concerns. Quantification of the discharge process is a key factor in flood modelling and will contribute to policy formulation for post-disaster migration and reconstruction.
Recent studies have used three types of remote sensing satellite data (altimetry, SAR/optimal imagery, and gravimetry) to track changes in the reservoir level, area, and mass. These observations were then used in a discharge model to understand the drainage dynamics of the reservoir. This approach meant it was possible to determine details of the reservoir drainage process by estimating the size of the breach, initial volumetric flow rates, and total water loss.
Authors of the research published in Water Resources Research believe their study, which provides a paradigm of incorporating a variety of state-of-the-art satellite remote sensing observations into a discharge model, will not only shed light on this flood disaster but also help future relevant studies.
By leveraging remote sensing data, they say their methodology serves as a reliable tool for understanding and addressing similar challenges in the aftermath of catastrophic events.
Panoramic view of the lower pool of the Dnieper Hydroelectric Power Plant and rocks during a strong low tide after the destruction of the Kakhovka dam. Image: PhotOleh/Shutterstock.com
References
Remote sensing assessment of dam impact on arid basins in Southern Saudi Arabia: A machine learning and space-for-time approach. Raid Almalki, Mehdi Khaki, Patricia M. Saco, Jose F. Rodriguez. Journal of Hydrology: Regional Studies 58 (2025) 102221. https://doi.org/10.1016/j.ejrh.2025.102221
Remote sensing with machine learning for multi-decadal surface water monitoring in Ethiopia Mathias Tesfaye & Lutz Breuer. Scientific Reports (2025) 15:12444. https://doi.org/10.1038/s41598-025-96955-y
A Remote Sensing Approach of Land and Water Content Change Between 2014 and 2024 to the Porsuk Dam And Its Near Surroundings, Kübra Günbey, Harun Böcük. Eskişehir Technical University Journal of Science and Technology C -Life Sciences and Biotechnology Estuscience – Life, 2025, 14 [1] pp. 1-13.
Yi,S.,Li,H.‐s.,Han,S.‐C.,Sneeuw,N., Yuan,C.,Song,C.,et al.(2025). Quantification of the flood discharge following the 2023 Kakhovka Dam breach using satellite remote sensing. Water Resources Research
