Date of Graduation

Spring 2016

Degree

Master of Science in Geospatial Sciences

Department

Geography, Geology and Planning

Committee Chair

Xiaomin Qiu

Keywords

object-based, land use and land cover, NAIP, random tree, Table Rock Lake

Subject Categories

Botany | Geology

Abstract

Land use and land cover (LULC) of Table Rock Lake (TRL) region has changed over the last half century after the construction of Table Rock Dam in 1959. This study uses one meter spatial resolution imagery to classify and detect the change of LULC of three typical waterside TRL regions. The main objectives are to provide an efficient and reliable classification workflow for regional level NAIP aerial imagery and identify the dynamic patterns for study areas. Seven class types are extracted by optimal classification results from year 2009, 2010, 2012 and 2014 of Table Rock Village, Kimberling City and Indian Point. Pixel-based post-classification comparison generated "from-to” confusion matrices showing the detailed change patterns. I conclude that object-based random trees achieve the highest overall accuracy and kappa value, compared with the other six classification approaches, and is efficient to make a LULC classification map. Major change patterns are that vegetation, including trees and grass, increased during the last five years period while residential extension and urbanization process is not obvious to indicate high economic development in the TRL region. By adding auxiliary spatial information and object-based post-classification techniques, an improved classification procedure can be utilized for LULC change detection projects at the region level.

Copyright

© Dexuan Sha Sha

Open Access

Share

COinS