Affordable Access

Access to the full text

Using remote sensing to map degraded mountain peatlands with high climate mitigation potential in Colombia's Central Cordillera

  • Battaglia, Michael J.1
  • Lafuente, Angela2
  • Benavides, Juan C.3
  • Lilleskov, Erik A.4
  • Chimner, Rodney A.2
  • Bourgeau-Chavez, Laura L.1
  • Skillings-Neira, Patrick Nicolás2, 3
  • 1 Michigan Tech Research Institute, Michigan Technological University, Ann Arbor, MI , (United States)
  • 2 College of Forest Resources and Environmental Science, Michigan Technological University, Houghton, MI , (United States)
  • 3 Department of Ecology and Territory, Pontificia Universidad Javeriana, Bogota , (Colombia)
  • 4 Northern Research Station, Climate, Fire, and Carbon Cycle Sciences Unit, USDA Forest Service, Houghton, MI , (United States)
Published Article
Frontiers in Climate
Frontiers Media S.A.
Publication Date
Mar 12, 2024
DOI: 10.3389/fclim.2024.1334159
  • Climate
  • Original Research


Peatlands are the most carbon-dense ecosystems on earth. In tropical mountains, peatlands are numerous and susceptible to rapid degradation and carbon loss after human disturbances. Quantifying where peatlands are located and how they are affected by land use is key in creating a baseline of carbon stocks and greenhouse gas fluxes from tropical mountain peatlands. However, mapping peatlands in the páramo of the Northern Andes is difficult because they are in a topographically complex environment with nearly continuous cloud cover and frequent conversion to pastures or cropland. The goal of this effort was to identify the different types of páramo peatlands and their degradation patterns in the Colombian Central Cordillera. Moderate resolution cloud-free composites of optical imagery, temporal variance in ALOS- PALSAR L-band SAR, Sentinel-1 C-band SAR, and topography data were used as inputs in a machine learning classifier to identify was used to map 12 land cover classes including peatlands with natural vegetation and peatlands converted to pasture. Field data from 507 control points collected across the study area, including information on the vegetation and carbon content on the top 20 cm of the soil, were used to train and validate the classifier. Results show that the use of multiple platforms and image dates, including variance of the radar returns, is necessary for a clear separation of disturbed and undisturbed peatland classes. Peatland area varied across the study region, covering 7% and 20% of the landscape in the northern and southern portions of the study area, respectively. Disturbed peatlands with exotic grasses cover nearly 2% of the area. The overall accuracy of the peatland classes was 82.6%. Disturbed peatlands with exotic grasses had less carbon in the top 20 cm than undisturbed peatlands with natural vegetation. These results highlight the prevalence of peatlands in the tropical Andes and a promising approach to detecting peatlands converted to agriculture. Understanding the distribution and extent of these carbon dense ecosystems can facilitate the restoration and protection of peatlands in the northern Andes, with implications for the future trajectories of the national greenhouse gas inventory.

Report this publication


Seen <100 times