Date of Completion

8-5-2020

Embargo Period

8-5-2020

Keywords

shadow, land cover, remote sensing

Major Advisor

Thomas Meyer

Associate Advisor

Daniel Civco

Associate Advisor

James Hurd

Associate Advisor

Chuanrong Zhang

Associate Advisor

Yeqiao Wang

Field of Study

Natural Resources: Land, Water, and Air

Degree

Doctor of Philosophy

Open Access

Open Access

Abstract

High-resolution imagery is becoming increasingly available for use in land-cover mapping; however, previous studies have found that urban shadows cast by elevated features in urban environments can cause substantial errors in land-cover classifications. Detecting and restoring land-cover information within urban shadows can help improve accuracy in land-cover mapping. Although a considerable number of studies have been conducted on both shadow detection and shadow restoration, few studies have focused on a complete urban shadow correction workflow that combines shadow detection and restoration for the purpose of land-cover mapping using high-resolution aerial imagery across large geographic extents s. Thus, the goal of this research was to develop a semi-automated approach to detect urban shadows and classify land-cover information within shadow areas for high-resolution aerial imagery. The specific objectives of this research were to: 1) develop and evaluate approaches that integrate multiple shadow detection methods, 2) evaluate the robustness of integrated shadow detection methods for a variety of landscapes and different forest canopy conditions, and 3) develop and evaluate a shadow correction algorithm to improve land-cover classification within shadow areas. This research will be beneficial to the remote sensing community working with high-resolution imagery by allowing them to mitigate the errors caused by shadows in land-cover classification at broad geographic scales with a low degree of human intervention.

Download
COinS