Analyzing Land Cover Change Detection Using GIS, Remote Sensing, and Machine Learning in the Colville National Forest, Northeast Washington, USA
Presentation Time: Tue, 08/10/2021 - 17:30
Keywords: Remote Sensing, Land Cover Change, Supervised Learning, Maximum Likelihood Classification, Normalized Differenced Vegetation Index, Colville National Forest
The Colville National Forest comprises 1.5 million acres of land in Northeastern Washington. With around 120,000 residents living in surrounding communities, this is a populated rural region where access to natural resources is highly vulnerable to land cover change. Over the last two decades, the forest has undergone changes in the spatial frequency and structure of land cover types, making the analysis of it a priority for the protection of the region and its communities. The digital storage of remote sensing data, coupled with Geographic Information Systems technologies and machine learning methods has made it possible to compare the current state of the land to prior states. Using Landsat 8 data and supervised learning, this project has generated land cover models for summer conditions from 2013 to 2020. The Normalized Differenced Vegetation Index has also been estimated to describe the summer trend of vegetation health in the forest. This study serves to describe how human activity and natural processes in Northeastern Washington may influence land cover over time, and provides a description of the nature, significance, and rate of change that can inform adaptive management planning in the federal lands of the region.
