A multiscale framework for landscape analysis: Object-specific analysis and upscaling

Landscapes are complex systems that require a multiscale approach to fully understand, manage, and predict their behavior. Remote sensing technologies represent the primary data source for landscape analysis, but suffer from the modifiable areal unit problem (MAUP). To reduce the effects of MAUP when using remote sensing data for multiscale analysis we present a novel analytical and upscaling framework based on the spatial influence of the dominant objects composing a scene. By considering landscapes as hierarchical in nature, we theorize how a multiscale extension of this object-specific framework may assist in automatically defining critical landscape thresholds, domains of scale, ecotone boundaries, and the grain and extent at which scale-dependent ecological models could be developed and applied through scale.[PUBLICATION ABSTRACT]