Research into the Sensitivity of Fragmentation Metrics
Project Title: Research into the Sensitivity of Fragmentation Metrics
- Funding: DFG
- Duration: 2008-2011
- Coordination: Prof. Dr. Christoph Kleinn
- Project researcher: Paul Magdon
Forest fragmentation has lead to the disruption of ecological processes and in many cases to the degradation of the natural resources. This makes it relevant for forest and conservation policy as it is among the key indicators for sustainable forest management on the landscape scale, where a relationship between fragmentation and ecological, conservation and economic issues is assumed. Also within the framework of international policy processes, like the Convention on Biological Diversity (CBD) or the Convention on Climate Change, monitoring of forest fragmentation comes into the focus of decision makers. It has been subject of intensive research during the past decades.
To evaluate forest fragmentation and its link to ecological processes, quantification is in demand. While there is no intuitive or immediate approach, various fragmentation metrics have been proposed which are calculated from a combination of directly measurable variables such as patch area, edge length, number of patches, dominance, diversity, contagion, fractal dimension and others. Special software has been developed, such as Fragstats , APACK, Patch Analyst and r.le in the GRASS environment to calculate a variety of indices or metrics. This has lead to an increase in quantitative landscape analysis in the last years, since easy to use software is freely available.
However, the sensitivity of fragmentation metrics has attracted less attention in the scientific literature than the development of new metrics, although - as with any other indicator - this sensitivity is an important piece of information for comprehensive interpretation. Factors of relevance for the calculation and interpretation of fragmentation metrics and their sensitivity include:
Tropical and subtropical forest are under special interest of conservation and climate change policies as they are hot spots for both biodiversity and carbon sequestration and at the same time they are facing big threats to be converted to other landuse systems. Remote sensing is one of the key technologies for forest cover and thus forest fragmentation assessments as it provides the possibility to monitor large areas with a reasonable effort. Today a hugh amount of land cover classification system exists and are widely used but most of them do not apply a specific class or boundary definition based on quantitative criteria. Thus, a comparison of the land cover maps and derived fragmentation metrics in space and time from different providers is often not possible.
Overall objective of this study is to research into the sensitivity of fragmentation metrics to the factors described above and to contribute to a more focused and more differentiated interpretation of fragmentation metrics. The potential of two new German satellite systems TerraSAR-X and RapidEye for forest fragmentation assessments in the subtropics and tropics will be analyzed with special focus on implementing specific forest definitions during image classification. While this study takes forest and forest fragmentation as an example, it is a general methodological study and the results can immediately be applied to the fragmentation status of any other land use or land cover class. To achieve the described goals the project is structured into two main sections
- Simulation and Modelling: This part seeks to examine fundamental linkages between the mentioned factors and the fragmentation metrics. Based on fundamental geometric considerations the simulations will be performed on artificial landscape models utilizing stationary isotropic Gaussian random fields. The derived landscape models will be classified to binary forest / non-forest maps according to different forest and forest edge definitions. This landscape maps will be combined with a set of digital elevation models (DEM) in a geographic information system (GIS) to get insight into the influence of topography on fragmentation metrics. For calculation of the 3D patch metrics we developed the software Patch3d.
- Case Study : Costa Rica Fragmentation analysis will be carried out for two study sites (30x30km) in Costa Rica. This will involve a variety of remote sensing products including aerial images, RapidEye and TerraSAR-X data. Different segmentation and classification techniques will be tested to produce forest / non-forest maps representing various forest and forest edge definitions. Based on this maps fragmentation analysis will be carried out, involving the use of DEM's with different vertical and horizontal resolutions.