{
  "id": 2728246,
  "title": "Mapping Coastal Forest Retreat Using Convolutional Neural Networks and Different Satellite Imagery",
  "url": "https://urgent.news/2026/08/22/mapping-coastal-forest-retreat-using-convolutional-neural-networks",
  "topic": "science",
  "section": "Science",
  "published": "2026-08-22T00:00:00.000Z",
  "source": {
    "name": "bioRxiv",
    "slug": "biorxiv",
    "url": "https://www.biorxiv.org/content/10.64898/2026.08.18.745552v1?rss=1"
  },
  "original_language": "en",
  "account": "Coastal forests throughout the world face a growing threat as rising sea levels, saltwater intrusion, and storm surges lead to increased soil saturation and salinity levels. As a result, healthy coastal forests, consisting of both wetland and low-elevation upland forests, are transforming into ghost forests - landscapes dominated by dead or dying trees nestled among salt-tolerant shrubs and grasses. These ghost forests eventually evolve into marshes or open water, marking a serious degradation of coastal ecosystems.\n\nOur primary goal for this study was to quantify the dynamic changes and pathways of forest landscape conversions, as well as the contributing factors to these changes. To achieve this, we first evaluated the effectiveness of various remote sensing indices - multispectral, bi-seasonal, topographical, and phenological metrics - in improving the performance of deep learning models (convolutional neural networks) for land cover classification in the coastal plain of North Carolina using surface reflectance data from Landsat 8 and Sentinel-2 images.\n\nUsing the best available data from Landsat 8, we assessed long-term change and identified patterns of land cover change from 1985 to 2021. Our findings indicate that incorporating phenology and topographical indices enhances the separability of the ghost forests class from all other vegetation classes. The higher-resolution Sentinel-2 data (with an F1 Score of 96.3) outperformed Landsat images (with an F1 Score of 93.4) for the 2021 co-available year. However, Landsat remains a crucial tool due to its long-term data record.\n\nBased on our analysis, we determined that 21% of forests were lost between 1985 and 2021, with the rate of loss accelerating over time. From 2010 to 2021 alone, 23,876 hectares of forest were converted to marsh, ghost forest, and shrub - a figure that is 1.5 times higher than the 16,968 hectares lost between 1985 and 2010. These transformations from forest to ghost forests and marshes were primarily driven by factors such as proximity to channels, salinity levels, and the increasing rate of relative sea level rise (RSLR). These environmental drivers are the key contributors to the observed changes in coastal ecosystems.\n\nBy quantifying these changes, we can pinpoint regions most vulnerable to environmental stressors and provide a foundation for targeted conservation strategies aimed at mitigating further degradation of coastal ecosystems.",
  "summary": "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…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}