A knowledge-based time series algorithm for robust mapping of on- and off-year Moso bamboo forests
Moso bamboo forests (MBFs) represented vital ecological and economic resources, necessitating dynamic monitoring of their on-year and off-year cycles to ensure sustainable management and effective carbon sequestration. To address classification challenges arising from regional phenological variation...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-07-01
|
| Series: | Ecological Indicators |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1470160X25006223 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Moso bamboo forests (MBFs) represented vital ecological and economic resources, necessitating dynamic monitoring of their on-year and off-year cycles to ensure sustainable management and effective carbon sequestration. To address classification challenges arising from regional phenological variations and the absence of key-period imagery, this study developed a knowledge-based on-year and off-year Moso bamboo forests (on/off-MBFs) classification algorithm (KB-OFBC) utilizing spectral feature analysis of Sentinel-2 imagery. First, by systematically analyzing the annual variation characteristics of NDVI and LSWI in MBFs, the Moso bamboo Time Series Index (MTSI) was constructed to extract MBFs. Second, by leveraging periodic spectral differences between on/off-MBFs in the NIR and red-edge bands, the Phenological Difference Index for bamboo (PDIbamboo) was established to distinguish these phenological phases. Finally, accurate spatial mapping of on/off-MBFs was achieved though the integration of MTSI and PDIbamboo. Results demonstrated that compared with existing indices, MTSI and PDIbamboo exhibited superior interclass separability, achieving an overall accuracy (OA) of 0.92 and a Kappa coefficient of 0.88. Application of KB-OFBC revealed that MBFs in Deqing County covered 18,036.56 ha in 2022, predominantly distributed in western mountainous regions, with off-MBFs accounting for 74.13%—significantly exceeding on-year coverage. In the validation areas of Anji County (Zhejiang Province) and Taojiang County, the algorithm achieved producer’s accuracy (PA) and user’s accuracy (UA) exceeding 0.85 and 0.80 respectively, demonstrating robust classification stability and spatial transferability. This algorithm provides a robust method for enhancing phenological monitoring, sustainable management, and large-scale mapping of MBFs. |
|---|---|
| ISSN: | 1470-160X |