Spatio-temporal land cover dynamics in the Pantanal assessed using lacunarity analysis on an ERS-1 SAR image time series

摘要

The Pantanal of Brazil, the largest wetland on the planet, is a disturbance-maintained ecosystem: an unusual topography coupled with a seasonal cycle of flooding and drydown creates a collection of landscapes that are environmentally heterogenous in space and time. Dominant land cover types include freshwater and saline lakes, periodically inundated grasslands, and forested corridors and patches. These cover types are highly heterogeneous in spatial arrangement and in response to inundation. Spatio-temporal analysis of land cover dynamics from synthetic aperture radar (SAR) image time series is relatively new research area but one that will expand given the increasing availability of SAR data. The Pantanal is well suited to microwave remote sensing because land cover types can exhibit great contrasts in backscattering. The authors have previously shown the efficacy of using lacunarity analysis with SAR imagery for quantifying land cover dynamics. In this presentation they extend that analysis to a total of seven ERS-1 SAR images from December 1992 to November 1993. This period includes both seasonal inundation followed by a significant climatic drought that transformed the spatial structure of backscattering across the landscape. Lacunarity analysis of the SAR image series captures the spatio-temporal rearranging and illustrates how complex land cover change can be quantified within a predictive framework.