A review of methods for scaling remotely sensed data for spatial pattern analysis

Context: Landscape ecologists have long realized the importance of scale when studying spatial patterns and the need for a science of scaling. Remotely sensed data, a key component of a landscape ecologist’s toolbox used to study spatial patterns, often requires scaling to meet study requirements. Objectives: This paper reviews methods for scaling remote sensing-based data, with a specific focus on spatial pattern analysis, and distills the numerous approaches based on data type. It also discusses knowledge gaps and future directions. Methods: Key papers were identified through a systematic review of the literature. Trends, developments, and key methods for scaling remotely sensed data and spatial products derived from these data were identified and synthesized to detail the general progression of a science of scaling in landscape ecology. Results: Upscaling both continuous and categorical data can oversimplify data, creating challenges for spatial pattern analysis. Object-based and neighborhood approaches can help, and since patch boundaries are more likely to align with objects than pixels, these may be better options for landscape ecologists. Many downscaling methods exist, but these approaches are not being widely employed for spatial pattern analysis. Conclusions: A diverse range of scaling methods are available to landscape ecologists, but work remains to integrate them into spatial pattern analysis. Moving forward, advances in computer science and engineering should be explored and cross-disciplinary research encouraged to further the science of scaling remotely sensed data.