For nearly a century, two dams on the Elwha River blocked the natural flow of sediment and wood, leading to a highly altered river environment. Removal of the dams unleashed large quantities of sediment and wood that had been trapped behind the reservoirs. This debris was carried downstream, reshaping the river’s course and impacting its ecosystems.
In a new USGS-led study, scientists have leveraged cutting-edge remote sensing and Artificial Intelligence (AI) technology to measure the movement and storage of large wood along the Elwha River in Washington State. This research, which followed the historic removal of two major dams on the river, provides new insights into how natural materials like wood and sediment shape river ecosystems over time.
Researchers relied on aerial imagery collected from 2012 to 2017 as part of interdisciplinary before-after/control-impact studies of dam-removal response. The images were generated using structure-from-motion photogrammetry using a digital camera mounted in the wing of a small airplane.
Developing the Deep-Learning Approach
Using the aerial imagery, researchers created an AI model capable of identifying and measuring large wood deposits along the river corridor. The model tracked changes in the Elwha River over several years, providing highly accurate measurements of wood and sediment movement—within 15% of true values.
One of the study’s key findings was the relationship between large wood deposits and the formation of sediment bars. As wood accumulated on the river's banks, it helped trap and build up sediment, playing a critical role in reshaping the river channel. This natural process, which was disrupted by the dams, is now restoring the river’s dynamic riparian ecosystems.
The AI model developed for this study could have far-reaching applications beyond the Elwha River. The researchers suggest that this technology could be used to identify large wood and sediment deposits in other river systems worldwide, potentially aiding in the restoration and management of rivers affected by human intervention.
Moreover, the datasets and AI models from the study have been made publicly available to encourage further research into river dynamics and the interactions between water flow, wood, sediment, and riparian ecosystems.
For nearly a century, two dams on the Elwha River blocked the natural flow of sediment and wood, leading to a highly altered river environment. Removal of the dams unleashed large quantities of sediment and wood that had been trapped behind the reservoirs. This debris was carried downstream, reshaping the river’s course and impacting its ecosystems.
In a new USGS-led study, scientists have leveraged cutting-edge remote sensing and Artificial Intelligence (AI) technology to measure the movement and storage of large wood along the Elwha River in Washington State. This research, which followed the historic removal of two major dams on the river, provides new insights into how natural materials like wood and sediment shape river ecosystems over time.
Researchers relied on aerial imagery collected from 2012 to 2017 as part of interdisciplinary before-after/control-impact studies of dam-removal response. The images were generated using structure-from-motion photogrammetry using a digital camera mounted in the wing of a small airplane.
Developing the Deep-Learning Approach
Using the aerial imagery, researchers created an AI model capable of identifying and measuring large wood deposits along the river corridor. The model tracked changes in the Elwha River over several years, providing highly accurate measurements of wood and sediment movement—within 15% of true values.
One of the study’s key findings was the relationship between large wood deposits and the formation of sediment bars. As wood accumulated on the river's banks, it helped trap and build up sediment, playing a critical role in reshaping the river channel. This natural process, which was disrupted by the dams, is now restoring the river’s dynamic riparian ecosystems.
The AI model developed for this study could have far-reaching applications beyond the Elwha River. The researchers suggest that this technology could be used to identify large wood and sediment deposits in other river systems worldwide, potentially aiding in the restoration and management of rivers affected by human intervention.
Moreover, the datasets and AI models from the study have been made publicly available to encourage further research into river dynamics and the interactions between water flow, wood, sediment, and riparian ecosystems.