Monitoring landscape fragmentation and aboveground biomass estimation in Can Gio Mangrove Biosphere Reserve over the past 20 years |
| |
| Institution: | 1. VNU University of Science, 334 Nguyen Trai, Thanh Xuan, Hanoi, Viet Nam;2. Hanoi National University of Education, 136 Xuan Thuy, Cau Giay, Hanoi, Viet Nam;3. School of Biological and Marine Science, University of Plymouth, Plymouth, Devon PL4 8AA, United Kingdom;1. Global Change Research Institute of the Czech Academy of Sciences, Bělidla 4a, 60300 Brno, Czech Republic;2. IFER – Institute of Forest Ecosystem Research, ?s.armády 655, 254 01 Jílové u Prahy, Czech Republic;3. Masaryk University, Department of Geography, Faculty of Science, Kotlá?ská 267/2, 611 37 Brno, Czech Republic;1. Institute of Geosystems and Bioindication, Technische Universität Braunschweig, Langer Kamp 19c, D-38106 Braunschweig, Germany;2. Faculty of Biology – Biotechnology, Vietnam National University, Ho Chi Minh University of Science, 227 Nguyen Van Cu, District 5, Ho Chi Minh City, Viet Nam.;3. Institut für Geographische Wissenschaften, Physische Geographie, Freie Universität Berlin, Malteserstr. 74-100, 12249 Berlin, Germany;4. NIOZ - Royal Netherlands Institute for Sea Research and Utrecht University, Landsdiep 4, 1797 SZ Den Hoorn, Texel, the Netherlands.;5. Paleoecological Environmental Assessment and Research Lab (PEARL), Queen''s University, Dept. Biology, 116 Barrie St., Kingston, Ontario K7L 3N6, Canada |
| |
| Abstract: | Over the past 20 years, the mangrove landscape of Can Gio Mangrove Biosphere Reserve (MBR) has undergone drastic changes in space and time. However, we know very little about changes in mangrove landscape model characteristics from analysis of different aspects based on landscape fragmentation. In the present study, the temporal and spatial changes of landscape pattern of land use/land cover (LULC) over the past 20 years in Can Gio Mangrove Biosphere Reserve (MBR), southern Vietnam were analyzed based on remote sensing data, with high classification accuracy (overall accuracy >85%, Kappa >0.8). The present study selected representative landscape indexes and built an integrated landscape index to examine the spatial-temporal changes of landscape patterns. Overall, over the past 20 years, the degree of fragmentation has gradually increased, mainly occurring in the transition zone of MBR. These changes are intended to reflect the significant temporal variation of the MBR, where the ecosystem is strongly disturbed by the intensity of human activities. We then investigate the effectiveness of principal component analysis (PCA)-based machine learning techniques in estimating the mangrove AGB, and applying landscape indices to assess impacts in Can Gio MBR. It reveals that the ANN model obtained the highest prediction accuracy (R2train = 0.785), followed by GPR (R2train = 0.703), and SVM (R2train = 0.671). As a result of applying the ANN model, the predicted mangrove AGB in 2000 and 2020 in the study site ranged from 6.531 to 368.163 Mg ha?1, and 13.749 to 320.295 Mg ha?1, respectively. These results support the application of the model as a tool to support LULC management and protection in the study site, and to contribute insights into the future mangrove research in other regions of the world. |
| |
| Keywords: | |
| 本文献已被 ScienceDirect 等数据库收录！ |
|
