The effects of war on land-use/land-cover change: An analysis of Landsat imagery for northeast Bosni...

مرکز و کتابخانه مطالعات اسلامی به زبان های اروپایی

منو

" The effects of war on land-use/land-cover change: An analysis of Landsat imagery for northeast Bosnia "
Frank D. W. Witmer J. V. O'Loughlin

Document Type	:	Latin Dissertation
Language of Document	:	English
Record Number	:	55667
Doc. No	:	TL25621
Call number	:	‭3256403‬
Main Entry	:	Frank D. W. Witmer
Title & Author	:	The effects of war on land-use/land-cover change: An analysis of Landsat imagery for northeast Bosnia\ Frank D. W. Witmer
College	:	University of Colorado at Boulder
Date	:	2007
Degree	:	Ph.D.
student score	:	2007
Page No	:	191
Abstract	:	The use of satellite technology by military planners has a relatively long history as a tool of warfare, but little research has used satellite technology to study the effects of war. This research addresses this gap by applying satellite remote sensing imagery to study the effects of war on land-use/land-cover change in northeast Bosnia. The war in Bosnia, 1992-1995, resulted in over 100,000 deaths, many more wounded, and the mass displacement of nearly half the population of 4.2 million. When combined with the destruction of much of the transportation infrastructure and housing stock, widespread mine placement, and loss of agricultural machinery, the impacts to both the people and land were dramatic. Though the most severe war impacts are visible at local scales (e.g. destroyed buildings), this study focuses on impacts to agricultural land, a larger scale visible to satellite sensors. Multispectral Landsat Thematic Mapper (TM) data (30m pixels) from before and during the war in addition to recent imagery from 2004/05 were used to detect abandoned agricultural land. The satellite images were co-registered to enable a perpixel analysis of changes based on the statistical properties of the pixels using multiple change detection methods. Ground reference data were collected in May of 2006 at survey sites selected using a stratified random sampling approach based on the derived map of abandoned agricultural land. Fine resolution (60cm) Quickbird imagery was also used to verify the accuracy of the classification. The remote sensing analysis results were then used to test two hypotheses related to war outcomes: (a) land abandonment is due to wartime minefields and (b) land abandonment is greater in pre-war Croat areas and areas where ethnic cleansing was heaviest. The effects of minefields on land abandonment was first tested in a geographic information system (GIS), and then by using multiple regression models that account for spatial autocorrelation among observations. The spatial regression analysis was conducted at the opstina (county) areal unit and used minefield locations, refugee returns and population change data as predictors of abandoned agricultural land. Results from these analyses show that a supervised classification of the Landsat TM data identified abandoned agricultural land with an overall accuracy of 82.5%. The GIS and spatial regression analysis of how war affects agricultural land showed that the presence of minefields and population declines are both associated with abandoned agricultural land. This research holds significance for both the remote sensing and civil war research communities. The use of freely available Quickbird imagery both as training data for the supervised classifier and as supplementary ground reference data suggest these methods are applicable to other civil wars (e.g. Darfur region of Sudan and the Horn of Africa) that may still be too dangerous for researchers to conduct field work in. By extending these methods to other war zones, comparisons of similarities and differences between such studies can then be made to draw broader conclusions of war impacts to land use and land cover.
Subject	:	Social sciences; Earth sciences; Bosnia and Herzegovina; Land cover; Land use; War; Geography; Armed forces; Remote sensing; 0799:Remote sensing; 0750:Armed forces; 0366:Geography
Added Entry	:	J. V. O'Loughlin
Added Entry	:	University of Colorado at Boulder