<p>An ecosystem is a community of organisms interacting with its environment, and landscapes are spatially interactive ecosystems. Earth's burgeoning human population demands ever more from finite ecosystems; but if managed well, landscapes can sustain their provision of resources and services and adapt to fulfill the changing human appetite. Management relies on sound information, and managing landscape change requires reliable spatio-temporal databases of ecologically relevant information. Remote sensing technologies fill this niche, providing increasingly large and diverse datasets, but the algorithms to extract information from the data must be developed. I developed and compared three remotely sensed measurements of forest canopy height to one another and to in situ field measurements. Both the precision and the accuracy (as well as the cost) of the measurements sorted along an axis of spatial scale, with Light Detection and Ranging (lidar) measurements proving most reliable at fine scales but prohibitively expensive over large areas and various radar technologies more appropriate for larger areas, especially when calibrated to the more accurate and precise lidar measurements. I also adapted traditional, single-time landcover classification algorithms to extract dense time series of categorical landcover maps from archival multi-spectral satellite images. These measurements greatly expand the potential spatio-temporal scope of landscape ecology and management, facilitating a shift away from data-imposed reliance on "space-for-time substitution" and loosely connected case studies toward robust, statistical analysis based on consistent information.</p> / Dissertation
Identifer | oai:union.ndltd.org:DUKE/oai:dukespace.lib.duke.edu:10161/1143 |
Date | January 2009 |
Creators | Sexton, Joseph O. |
Contributors | Urban, Dean L. |
Source Sets | Duke University |
Language | en_US |
Detected Language | English |
Type | Dissertation |
Format | 2592096 bytes, application/pdf |
Page generated in 0.0094 seconds