Electronic Theses and Dissertations

Identifier

779

Date

2013

Document Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Earth Sciences

Committee Chair

Esra Ozdenerol

Committee Member

Takuya Nakazato

Committee Member

Brian A waldron

Committee Member

Arleen Alice Hill

Abstract

Land-use land-cover change (LULCC) especially those caused by human activities, is one of the most important components of global environmental change (Jessen, 2005). This dissertation analyses the effects of geographic, biographical, and demographic factors on LULCC and how LULCC and geographic variables influence the spread of invasive Giant Hogweed in northeastern Latvia. Data sets used in this study include: remote sensing images (Landsat Thematic Mapper acquired in 1992 and 2007), global positioning system (GPS data), census data, and data from public participation geographic information systems (PPGIS). These data were processed and analysed in a geographic information system (GIS). Six categories of land-cover were studied to determine land-cover change (LCC) and the relationship to population change between 1992 and 2007. Classification and analyis of the 1992 and 2007 Landsat images revealed that land-cover changing to forest is the most common type of change (17.1% of pixels) followed by changes to agriculture (8.6% of pixels) and the least was changes to urban/suburban (0.8% of pixels). Integration of the census data and land-cover classification revealed interesting patterns, for example, that population density is positively correlated with percent change to forest, agriculture and urban. Modeling the spread of Giant Hogweed was achieved using logistic regression and a novel cluster analysis approach. The logistic regression model was used to model the spread of Giant Hogweed using presence and pseudo-absence data of Giant Hogweed, while cluster analysis used only Giant Hogweed presence data. Both models were run using data from a series of GIS layers including topographic and LULCC information. The results from logistic regression and cluster analysis show that Giant Hogweed is more likely to grow near roads, near rivers, in proximity to urban centers and in low elevation areas. Habitat suitability maps produced from both models indicate where Giant Hogweed is more likely to spread in the future and can serve as useful tools for policy makers and land managers to focus their efforts to manage weed invasions, and identify similar habitats where Giant Hogweed may occur in the future.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.

Download

Share

COinS