Quantifying landscape fragmentation

The features of land cover objects in digital raster images are often described by their pattern, connectivity, and fragmentation. While there are many quantitative measures for pattern and connectivity fragmentation is usually described in a qualitative way, and often for a specific species only living in the landscape under study. The notion of fragmentation comprises many aspects of image properties. A daunting task could be the bottom up approach to derive a series of indicators describing all kind of aspects and then trying to summarize them. In contrast, this study suggests a top-down approach, illustrating and comparing the use of three holistic concepts, based on geometric principles only, and resulting in normalized, quantitative fragmentation metrics describing both, the overall degree as well as the spatial distribution of fragmentation on any categorical land cover map. After providing the motivation for using the frameworks of contagion, complexity, and spatial entropy their algorithmic implementation is explained. The performance and features of the proposed three concepts are exemplified on a binary forest mask. Together with a batch-processing option these tools are available within the free image analysis software GuidosToolbox. The user-friendly provision of operational tools for a generic and especially quantitative assessment of fragmentation could contribute to an improved understanding and interpretation of landscape dynamics. Monitoring and especially quantifying the impact of human activities on our landscapes may also facilitate the design of efficient and assessable landscape resource policies and risk assessment studies.

VOGT Peter; 

2015-11-16

Instituto Nacional de Pesquisas Espaciais

JRC98242

978-85-17-0076-8,   

http://www.dsr.inpe.br/sbsr2015/,    https://publications.jrc.ec.europa.eu/repository/handle/JRC98242,