Developing automated and improved coastline measurements so that Western Alaska communities can plan for change

View of a remote coastal bluff
Eroding coastline near Deering, Alaska. Photo courtesy of Frank Witmer/UAA.

Rapid warming is leading to accelerated coastal erosion and placing many Alaska communities at risk. Accurately forecasting future coastline changes is critical to communities so that they can mitigate and adapt to these changes. However, Alaska’s coastline is poorly mapped, which means that current erosion forecast models are limited in their predictive ability.

University of Alaska Anchorage associate professor Frank Witmer, associate professor Matthew Kupilik, and graduate student Kristopher Carroll are addressing this need in Alaska’s at-risk regions, with support from Alaska Sea Grant. They are using traditional statistical methods, as well as developing advanced artificial intelligence algorithms to process satellite imagery. 

Man in grey jacket standing on a rocky beach
Frank Witmer collecting GPS measurements along the coastline near Deering, Alaska. Photo courtesy of Frank Witmer/UAA.

“While walking coastlines to collect in-situ measurements offers the most accurate way to map coastlines, the method is time-consuming and puts a strain on resources,” said Carroll, a geospatial data science graduate student. “Developing ways to automate satellite image processing can provide more frequent coastline measurements over more coastline.”

The team is taking existing technology used to automate the analysis of lower resolution satellite images and adapting it to work with higher resolution and more detailed satellite imagery. The result is more accurate, detailed maps of Alaska coastlines. 

Looking down at a remote coast from a small plane
Aerial view of the coastline near Deering, Alaska. Photo courtesy of Frank Witmer/UAA.

The research team is currently applying the procedure to Deering, Alaska, a small coastal town on the Northern Seward Peninsula prone to the effects of coastal erosion. Initial results suggest the method generates more precise coastline measurements than existing methods. Once finalized, the approach will be scaled up and the resulting coastline model can be used to improve future erosion forecasts for communities on the Alaskan coast, specifically those along the western coastline that are experiencing coastal erosion. For more information, see the Climate-driven Arctic coastline modeling: improving erosion forecasts for communities project page.

“Storm events are large contributors to coastal erosion, so it’s important to capture coastline changes as close as possible to just before and after the storm,” said Witmer. “The improved spatio-temporal resolution also helps to capture images of Alaska coastlines that are frequently obscured by cloud cover.”