Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


ucin1070488160.pdf (4.26 MB)

ETD Abstract Container

Abstract Header

AN ANALYSIS OF THE SPATIAL SCALE EFFECTS ON LANDSCAPE PATTERN METRICS IN A DEFORESTED AREA OF RONDONIA, BRAZIL

HAO, YONGPING
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1070488160

Abstract Details

2003, PhD, University of Cincinnati, Arts and Sciences : Geography.
Scale effects on the performances of landscape metrics have attracted a lot of concern recently. Most studies have focused on two scale effects: changes in spatial resolution and changes in spatial extent. So far, little empirical analysis has been done on the effects of spatial aggregation (filtering) on landscape metrics. This research analyzed the effects of three types of changes in scale on the performances of eighteen commonly used landscape metrics. These changes include (1) varying spatial aggregation, (2) varying spatial resolution, and (3) varying spatial extent. Six time series Landsat images of Ariquemes (path/row: 232/67), Rondonia, Brazil were used in the analysis. A subset of relatively homogeneous area of 1024 x 1024 pixels (30.72 x 30.72 km) was used for testing effectiveness of eighteen landscape metrics for varying spatial aggregation and varying spatial resolution. For effects of spatial extent, this area serves as a maximum extent. The texture analysis with a window size of 15 x 15 pixels was used to distinguish naturally barren areas, which were subtracted from images of 1984 - 2000 before further image processing. Results show that PPU (NumP), ED (TE) and MNN are good indicators of landscape fragmentation, with PPU (NumP) the most sensitive to slight changes. Similarly, Sq and LSI are good indicators of landscape shape irregularity or complexity, with Sq the most sensitive to slight changes. Results also showed that LPI, which has been commonly used as an indicator of patch aggregation, should not be used to indicate landscape fragmentation, though it is a good index of landscape composition. The analysis of the time series images show that the most dramatic landscape fragmentation in the study area occurred from 1984 to 1989. However, since 1984, deforestation in the urban area tended to occur around urban centers, which was characterized by aggregation with more regular patch shapes. Conversely, rural area experienced increasing trends in landscape fragmentation from 1984 to 2000. This study has demonstrated the importance of using landscape metrics that have little sensitivity to scale effects and most sensitivity to the actual landscape pattern changes for examining time series changes in deforestation on the landscape.
Dr. Robert C. Frohn (Advisor)
160 p.
Geography
scale effects; landscape metrics; time series images; deforestation; Rondonia

Recommended Citations

Citations

  • HAO, Y. (2003). AN ANALYSIS OF THE SPATIAL SCALE EFFECTS ON LANDSCAPE PATTERN METRICS IN A DEFORESTED AREA OF RONDONIA, BRAZIL [Doctoral dissertation, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1070488160

    APA Style (7th edition)

  • HAO, YONGPING. AN ANALYSIS OF THE SPATIAL SCALE EFFECTS ON LANDSCAPE PATTERN METRICS IN A DEFORESTED AREA OF RONDONIA, BRAZIL. 2003. University of Cincinnati, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1070488160.

    MLA Style (8th edition)

  • HAO, YONGPING. "AN ANALYSIS OF THE SPATIAL SCALE EFFECTS ON LANDSCAPE PATTERN METRICS IN A DEFORESTED AREA OF RONDONIA, BRAZIL." Doctoral dissertation, University of Cincinnati, 2003. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1070488160

    Chicago Manual of Style (17th edition)

ucin1070488160
3,641
© 2003, all rights reserved.
This open access ETD is published by University of Cincinnati and OhioLINK.