A new study finds that coastal forest loss in North Carolina’s Albermarle-Pamlico Peninsula began accelerating after 2010 and may still be speeding up. The study also found that N.C. lost more than 20% of its coastal forest between 1985 and 2021.
Researchers used satellite imagery to track coastal forest loss in N.C. from 1985 through 2021 and found the state experienced a 21% loss of coastal forests, or approximately 64,220 hectares. Just over 40,000 ha of that land was lost to marsh, ghost forest and shrub, specifically.
That loss did not happen linearly, said Titilayo Tajudeen, lead author of a paper on the study and graduate researcher at North Carolina State University.
“Between 2010 and 2021, we saw 23,876 ha of forest were converted to marsh, ghost forest, and shrub,” Tajudeen said. “That is 1.5 times higher than the 16,968 ha lost to marsh, ghost forest, and shrub between 1985 and 2010.”
Sea level rise was the chief driver of forest loss. As sea levels rise, salty water inundates forest areas, killing trees and converting the area into ghost forest. Conversion into ghost forest has sped up even more than overall forest loss. Ghost forests grew by 7561 ha between 2010 and 2021, 2.5 times faster than between 1985 and 2010 (3087 ha).
Most of the highly affected areas were concentrated within one kilometer of the coast. Some had been subject to a series of extreme events.
“This area experienced severe drought from 2007 to 2011, and then also Hurricane Irene in 2011,” Tajudeen said. “While these events occurred long before 2021, some of the areas simply never recovered. Despite being protected, a combination of these extreme events along with rising sea levels has pushed them into new ecological states, including becoming ghost forests.”
To determine the rate of forest loss, researchers used two satellite imagery tools, known as Landsat 8 and Sentinel-2. These are image databases which scientists used to train an AI model, known as a convolutional neural network, which processes the type of grid-like data that Landsat and Sentinel provide. By doing so, the neural network can help identify which areas of the image are forest land, and which areas have been converted to marsh, ghost forest, and shrub.
The Sentinel-2 data has a resolution of 10 meters, which is smaller and sharper than the 30-meter Landsat 8 resolution. However, Landsat’s database covers a much larger span of time, which is what enabled researchers to examine forest loss going back to 1985.
Both Landsat 8 and Sentinel 2 had data for 2021, so researchers compared their performance in that year to determine which data set provided more accurate readings. They found that the sharper Sentinel-2 data outperformed Landsat images, but that Landsat was still an important tool because of its longer-term data record.
The paper, “Mapping coastal forest retreat using convolutional neural networks and different satellite imagery,” is published in PLOS One. Co-authors include Marcelo Ardon, Mirela Tulbure and Katherine Martin of NC State.
-pitchford-
Note to editors: An abstract follows.
“ Mapping coastal forest retreat using convolutional neural networks and different satellite imagery ”
DOI: 10.1371/journal.pone.0357346
Authors: Titilayo Tajudeen, Marcelo Ardon, Mirela Tulbure and Katherine Martin, NC State.
Published Sept. 15, 2026 in PLOS One
Abstract: Coastal forests are increasingly threatened by saturated soil and elevated salinity levels resulting from sea level rise, saltwater intrusion, and storm surges. In response to rising salinization and flooding, healthy coastal forests that rely on freshwater (both wetland forests and low-elevation upland forests) are transitioning into landscapes dominated by dead or dying trees, known as ghost forests. Situated among salt-tolerant shrubs and grasses, ghost forests eventually become marshes or open water. Here, our main objective was to quantify the dynamics and pathways of these forest landscape conversions, as well as the factors contributing to the changes, which is vital for understanding the progression of coastal ecosystem degradation and forecasting future changes. We focused first on identifying the best method to track forest landscape change by exploring the role of multiple remote sensing indices (i.e., multispectral, bi-seasonal, topographical, and phenological metrics) in enhancing the performance of deep learning models (convolutional neural networks, CNNs) for land cover classification in the coastal plain of North Carolina using surface reflectance of Landsat 8 and Sentinel-2 images. Then, we used the best available data (Landsat 8) to understand long-term change and identify patterns of land cover change from 1985 to 2021. Our study reveals that incorporating phenology and topographical indices enhances the separability of the ghost forests class from all other vegetation classes. In our assessment, the higher-resolution Sentinel-2 data (F1 Score = 96.3) outperformed Landsat images (F1 score = 93.4) for the 2021 co-available year. However, Landsat remains an important tool used due to its long-term data record. Therefore, we used Landsat to determine that 21% of forests were lost between 1985 and 2021, and that the rate of loss is increasing. Between 2010 and 2021, 23,876 ha of forest were converted to marsh, ghost forest, and shrub, which is 1.5 times higher than the 16,968 ha lost between 1985 and 2010. These conversions from forest to ghost forest and marshes were driven primarily by proximity to the channel, salinity, and the increasing rate of relative sea level rise (RSLR), which are the key environmental drivers of observed changes. By quantifying these changes, we highlight regions most vulnerable to environmental stressors, providing a basis for targeted conservation strategies.
PLOS One
Not applicable
Mapping coastal forest retreat using convolutional neural networks and different satellite imagery
15-Sep-2026
The authors declare no conflicts of interest