A new fractal index to classify forest fragmentation and disorder

doi:10.1007/s10980-023-01640-y

Published April 7, 2023 | Version v.1

Journal article Open

A new fractal index to classify forest fragmentation and disorder

1. Research Center for Integrated Analysis and Territorial Management, University of Bucharest, Bucharest, Romania
2. GSRC, Division of Medical Physics and Biophysics, Medical University of Graz, Graz, Austria
3. Department for Experimental Oncology, Institute for Oncology and Radiology of Serbia, Pasterova 14, 11000, Belgrade, Republic of Serbia
4. USDA Forest Service, Southern Research Station, Research Triangle Park, North Carolina, USA
5. Department of Biomedical Engineering, Khalifa University, Abu Dhabi, UAE
6. Computational NeuroSurgery (CNS) Lab, Faculty of Medicine, Health and Human Sciences, Macquarie Medical School, Macquarie University, Sydney, NSW, Australia
7. Department of Life Science and Agriculture, Hokkaido University of Education, Asahikawa Campus, Asahikawa, Japan
8. Research Center for Integrated Analysis and Territorial Management, University of Bucharest, Bucharest, Romania Faculty of Administration and Business, University of Bucharest, 4-12, 030018, Bucharest, Romania View author pu
9. Research Center for Integrated Analysis and Territorial Management, University of Bucharest, Bucharest, Romania Faculty of Administration and Business, University of Bucharest, 4-12, 030018, Bucharest, Romania
10. Faculty of Administration and Business, University of Bucharest, 4-12, 030018, Bucharest, Romania
11. Department of Geosciences and Natural Resource Management (IGN), University of Copenhagen, Øster Voldgade 10, 1350, Copenhagen K, Denmark
12. Department of Integrative Biology, University of Texas at Austin, Austin, Texas, USA

Abstract

Forest loss and fragmentation pose extreme threats to biodiversity. Their efficient characterization from remotely sensed data therefore has strong practical implications. Data are often separately analyzed for spatial fragmentation and disorder, but no existing metric simultaneously quantifies both the shape and arrangement of fragments.Objectives. We present a fractal fragmentation and disorder index (FFDI), which advances a previously developed fractal index by merging it with the Rényi information dimension. The FFDI is designed to work across spatial scales, and to efficiently report both the fragmentation of images and their spatial disorder. Methods. We validate the FFDI with 12,600 synthetic hierarchically structured random map (HRM) multiscale images, as well as several other categories of fractal and non-fractal test images (4880 images). We then apply the FFDI to satellite imagery of forest cover for 10 distinct regions of the Romanian Carpathian Mountains from 2000–2021. Results. The FFDI outperformed its two individual components (fractal fragmentation index and Rényi information dimension) in resolving spatial patterns of disorder and fragmentation when tested on HRM classes and other image types. The FFDI thus offers a clear advantage when compared to the individual use of fractal fragmentation index and the Information Dimension, and provided good classification performance in an application to real data. Conclusions. This work improves on previous characterizations of landscape patterns. With the FFDI, scientists will be able to better monitor and understand forest fragmentation from satellite imagery. The FFDI may also find wider applicability in biology wherever image analysis is used.

Context. Forest loss and fragmentation pose extreme threats to biodiversity. Their efficient characterization from remotely sensed data therefore has strong practical implications. Data are often separately analyzed for spatial fragmentation and disorder, but no existing metric simultaneously quantifies both the shape and arrangement of fragments. Objectives. We present a fractal fragmentation and disorder index (FFDI), which advances a previously developed fractal index by merging it with the Rényi information dimension. The FFDI is designed to work across spatial scales, and to efficiently report both the fragmentation of images and their spatial disorder. Methods. We validate the FFDI with 12,600 synthetic hierarchically structured random map (HRM) multiscale images, as well as several other categories of fractal and non-fractal test images (4880 images). We then apply the FFDI to satellite imagery of forest cover for 10 distinct regions of the Romanian Carpathian Mountains from 2000–2021. Results. The FFDI outperformed its two individual components (fractal fragmentation index and Rényi information dimension) in resolving spatial patterns of disorder and fragmentation when tested on HRM classes and other image types. The FFDI thus offers a clear advantage when compared to the individual use of fractal fragmentation index and the Information Dimension, and provided good classification performance in an application to real data. Conclusions. This work improves on previous characterizations of landscape patterns. With the FFDI, scientists will be able to better monitor and understand forest fragmentation from satellite imagery. The FFDI may also find wider applicability in biology wherever image analysis is used.

Notes

Funding: Research was supported by a grant of the Romanian Ministry of Education and Research, CNCS–UEFISCDI, Project Number PN-III-P4-ID-PCE-2020-1076, within PNCDI III, grant of the Ministry of Research, Innovation and Digitization, CNCS/CCCDI-UEFISCDI, Project Number PN-III-P2-2.1-SOL-2021-0084, within PNCDI III, grant of the University of Bucharest "Spatial projection of the human pressure on forest ecosystems in Romania", University of Bucharest, (UB/1365) and grant "Development of the theory of the dynamic context by analyzing the role of the aridization in generating and amplifying the regressive phenomena from the territorial systems", Executive Agency for Higher Education, Research, Development and Innovation Funding, Romanian Ministry of Education Research Youth and Sport (UEFISCDI) (TE-2014-4-0835) to I.A., D.P., A.G., A.K.G., A.G.S., I.D.N., C.C.D., M.M., and D.C.D. K.K. was funded by the Japan Society for the Promotion of Science (KAKENHI Grant Numbers 18K06406, 19H02987)

Files