Type of Document	Dissertation
Author	Wang, Jialing
URN	etd-08142005-151446
Title	Multi-scale Forest Landscape Pattern Characterization
Degree	Doctor of Philosophy
Department	Geography, Department of
Advisory Committee	Advisor Name Title Xiaojun Yang Committee Chair Frances C. James Committee Member James B. Elsner Committee Member Jon Anthony Stallins Committee Member Mark W. Horner Committee Member
Keywords	MAUP Landscape Metrics Landscape Classification Scale The Red Hills Region
Date of Defense	2005-08-05
Availability	unrestricted
Abstract The purpose of this dissertation is to examine several important issues in landscape pattern analysis, including the identification of important landscape metrics, the impact of the modifiable areal unit problem (MAUP) in landscape pattern analysis, the linkage between pattern and process, and the application of landscape pattern analysis. A theoretical framework of hierarchical patch dynamics paradigm and a technical framework of GIS and remote sensing integration are employed to address these questions. The Red Hills region of southwestern Georgia and northern Florida is chosen as the study area. Land use/cover (LULC) and longleaf pine distribution maps were generated through satellite image classification. Sub-watersheds were used as the main analysis units. Principal component analysis (PCA) was conducted on 43 sub-watersheds at three hierarchical LULC levels to identify important landscape metrics. At both landscape- and class-levels, the measurement of fragmentation was identified as the most important landscape dimension. Other dimensions and important metrics varied with different scales. Hexagons were used as an alternative zoning system to examine the MAUP impact in landscape pattern analysis. The results indicated that landscape pattern analyses at class level and at broader scales were more sensitive to MAUP than at landscape level and at finer scales. Local-scale pattern analysis based on moving window analysis greatly reduced the impact of MAUP at class level, but had little effects at landscape level. An examination of the relationship between landscape pattern variables and biophysical/socio-economic variables was undertaken by using statistical analysis. The biophysical variables of soil drainage and mean slope and the socio-economic variables of road density, population density, distance to Tallahassee, Florida, and plantation amount were found to be closely correlated to the landscape patterns in this region. However, a large amount of variation in the landscape patterns remained unexplainable, suggesting that additional factors should be considered in the analysis of pattern and process relationship. The important landscape metrics identified by PCA were used in landscape classification and evaluation. Nine core longleaf pine patches were identified as having the first conservation priority. Eight sub-watersheds and thirteen plantations were considered the most important in management.
Files	Filename       Size       Approximate Download Time (Hours:Minutes:Seconds)     28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access    final-dissertation-submission.pdf 18.98 Mb 01:27:52 00:45:11 00:39:32 00:19:46 00:01:41