Mapping Vegetation Communities using Convolutional Neural Networks

Artificial intelligence applied to multiple remote sensing datasets accurately represents shrub distribution across hillslopes at the Kougarok field site.

November 27th, 2019
The Science: 
  • Unique training dataset was collected at 30 field plots and the resulting vegetation products evaluated against them for accuracy.
  • Fusion of hyperspectral, multispectral, and terrain datasets was performed over a 343 km2 region.
  • High-resolution (5 m) vegetation classification map was generated.
The Impact: 
  • Produced one of the most accurate, high resolution, field-validated vegetation maps for tundra ecosystems.
  • Using map now to benchmark field predictions of plant distribution and dynamics across changing Arctic landscapes.
Summary: 
  • The authors aimed to develop a deep learning approach (convolutional neural network) capable of generating vegetation maps for the Seward Peninsula in western Alaska.

 

Langford, Z., Kumar, J., Hoffman, F., Breen, A. and Iversen, C., 2019. Arctic vegetation mapping using unsupervised training datasets and convolutional neural networks. Remote Sensing 11:69. doi.org/10.3390/rs11010069

Vegetation community types (e.g., shrubs) were mapped for the NGEE Arctic Kougarok watershed and the surrounding region using multi-sensor fusion of remote sensing data and topography with machine learning methods in a high performance computing environme

Contacts: 

Zachary Langford
Oak Ridge National Laboratory