Skip to main content


eCommons@Cornell

eCommons@Cornell >
Internet-First University Press >
Symposia, Workshops, Conferences >
Hydrologic Discovery Through Physical Analysis Honoring the Scientific Legacies of Wilfried H. Brutsaert and Jean-Yves Parlange >
Hydrologic Discovery - Posters >

Please use this identifier to cite or link to this item: http://hdl.handle.net/1813/29611
Title: I1. Inferring Species-­Richness and Species-turnover by Statistical Multiresolution Texture Analysis of Satellite Imagery
Authors: Convertino, M.
Lowry, N.C.
Linkov, I.
Mangoubi, R.
Desai, M.
Issue Date: May-2012
Publisher: Internet-First University Press
Abstract: The quantification of species-richness and turnover is one of the most important tasks in monitoring ecosystems. This is both for guaranteeing ecosystem function, and to understand the linkages between natural and human stressors with species patterns. Wetland ecosystems, particularly water-controlled subtropical wetlands, are extremely sensitive to external changes, for example in rainfall and water management. The effect of these changes at the metacommunity level in space and time are still not well understood. We analyze interseasonal and interannual average species-richness and turnover of the Arthur R. Marshall Loxahatchee National Wildlife Refuge (\Water Conservation Area 1» in the Greater Everglades Ecosystem) in South Florida as a case-study for the application of a novel multispectral image analysis technique. We use a texture augmented procedure to analyze high resolution satellite images (Landsat) in order to detect texture changes of vegetation, soil, and water components. α- and β-diversity, which are observed to be independent, are estimated for the green-band by the Shannon entropy and by the Kullback-Leibler divergence respectively. Validation with observations about the evolution of vegetation patterns shows that the analysis predicts 73 % and 100 % of species-richness and turnover within the study-area from 1984 to 2011. The KL divergence is a better metric than the difference of Shannon entropy which captures 85 % of the species-turnover. This is because the KL divergence takes in account the pairwise interactions between vegetation communities in time. α- and β-diversity are positively correlated, and diversity is strongly correlated to the average annual rainfall. We found that changes in vegetation, soil and water are positively correlated and that the fluctuations of the Shannon entropy for each component in the wet-season are smaller than in the dry-season. However, the KL divergence better predicts the species-turnover in the wet-season. The Gaussian density function in texture characterization and the use of the KL divergence constitute a promising technique for monitoring spatiotemporal ecohydrological patterns with particular focus on species-richness and turnover. We envision relevant applications of the KL divergence to infer species-dissimilarity, which is the diversity in space. This is particularly important when historical data or continuous monitoring data are not available in order to detect and potentially anticipate the effects of natural and anthropic changes on ecosystem structure.
URI: http://hdl.handle.net/1813/29611
Appears in Collections:Hydrologic Discovery - Posters

Files in This Item:

File Description SizeFormat
I1_Convertino_poster.pdf315.04 kBAdobe PDFView/Open

Refworks Export

Items in eCommons are protected by copyright, with all rights reserved, unless otherwise indicated.

 

© 2014 Cornell University Library Contact Us