Applying Unsupervised Learning in Exploring Local Variations of Vegetation Change

Authors: Tianqi Zhang*, Ohio State University, Desheng Liu, The Ohio State University
Topics: Remote Sensing, Polar Regions
Keywords: unsupervised learning, vegetation change, climate warming, tundra environment
Session Type: Poster
Day: 4/5/2019
Start / End Time: 1:10 PM / 2:50 PM
Room: Lincoln 2, Marriott, Exhibition Level
Presentation File: No File Uploaded


Vegetation in the tundra environment plays an essential role in global energy and carbon cycling processes. Previous investigations in the Arctic area have shown spatial variations in the correlations between vegetation change and climate warming. Although potential controlling factors in causing these spatially non-identical patterns have been discussed, which could originate from local variations (e.g., plant family, snow metrics, topography etc.), control strengths and senses (positive or negative) of these local variations have been seldom explored. Recently, machine learning algorithms have been applied in remote sensing studies. By applying an unsupervised clustering using the time series of spectral signature at per-pixel basis, which can be affected by the abovementioned factors, pixels categorized as the same cluster are similar to each other in terms of background and species information. The intra-cluster similarity and inter-cluster differences in vegetation change pattern relative to climate warming, therefore, could be detected.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login