About Site Language

WHDL is viewable in multiple languages. Use the pull-down menu to select a language to view the site.

I changed my language, but I’m still seeing resources in the other languages?

If a resource or text has not been translated into your selected language, it will appear in the initially added language. We are always looking for help translating these resources. If you can help, contact us!

WHDL - 00017634

Using Pixel-Based Classification and Change Detection to Map Burn Extent & Severity from Hyperspatial Drone and Very-High Resolution Satellite Imagery

Cole Edward McCall

Resource Types

Academic (Thesis, Dissertation, Etc.)

Resource Languages

English

Institutional Repository

Description / Abstract

Submitted to the Department of Mathematics and Computer Science in partial fulfillment of the requirements for the degree of Bachelor of Science

Over the past few years, the size and severity of wildland forest fires have continued to increase, causing more damage and destruction around the world. New methods have been developed to utilize machine learning algorithms to map forest fire burn extent and fire severity using aerial imagery. Algorithms such as the Support Vector Machine (SVM) can be used to classify pixels as either black ash, white ash, or unburned, while the Mask Region-Based Convolutional Neural Network (MaskR-CNN) can be used to find and map tree objects. The results from these classifications can be used to help local wildland fire managers assess the burned area and create a recovery plan.

This research has several steps: 1) improving the current method for mapping burn extent with hyperspatial drone imagery, 2) using the same (or similar) methods to determine if wildland fire burn extent can be mapped with high-resolution satellite imagery, 3) evaluating how spatial and spectral resolution impacts the accuracy of the classification, and 4) try to develop new methods that involve less (or no) training data, such as unsupervised change detection. The results of each step were promising, creating much more accurate classifications of wildland fire burn extent than can be obtained by other common products such as LANDFIRE. This research will continue, likely moving in a direction that further examines the use of unsupervised and self-supervised machine learning, which greatly reduces the training data needed.

Download this Resource

Download Free from WHDL

Download as a PDF: Using Pixel-Based Classification and Change Detection to Map Burn Extent
File Size Small - 936 KB

Go to Download Information

Subject Index

Copyright Notes

This material is used by permission of the rightsholder in accordance with the terms of the relevant content release. As a user, you have permission to use this content and download a copy for non-commercial use only. Downloading multiple quantities of this resource is expressly forbidden.

KeywordsMachine learning, MaskR-CNN, Burn severity, Support vector machine, Hyperspatial drone imagery, Multispectral satellite imagery, Burn extent

Degree AwardedBachelor of Science

Awarding InstitutionNorthwest Nazarene University

Committee MembersThomas McCall, Barry L. Myers

Supervisor or ChairDale A. Hamilton

Peer Reviewed?No

Keep Browsing