Assessment of optical vegetation indices for characterization of mangrove phenology

• Evaluated the performance of 26 vegetation indices from 6 categories for mangrove phenology. • Proposed the novel methods of time series reconstruction and VI assessment. • The chlorophyll- and water content-related VIs performed the best. • The structure-related VIs showed moderate performance. • The mangrove phenological patterns of different vegetation indices exhibited asynchrony. Mangroves are coastal ecosystems with salt-tolerant, evergreen trees and shrubs. Investigating mangrove phenology is significant for understanding their seasonal dynamics and ecological processes. To date, Normalized Difference Vegetation Index (NDVI) and Enhanced Vegetation Index (EVI) have been commonly used to monitor mangrove phenology from satellites, while other vegetation indices capturing physiological and biochemical characteristics of plants are rarely applied. With Sentinel-2 satellite imagery from 2019 to 2022 covering mangrove forests in China, we evaluated the performance of various optical vegetation indices from six classes related to structure, chlorophylls, carotenoids and anthocyanins, water content, dry matter, and photosynthesis, for mangrove phenology characterization, and compared the phenological patterns exhibited by them. We firstly selected 26 vegetation indices and improved the Harmonic ANalysis of Time Series (HANTS) algorithm for time series reconstruction. Then, we developed the Reconstruction Quality Index (RQI) and the Normalized Cumulative Reconstruction Success Rate (RSR nc ) to evaluate the capability of all vegetation indices in mangrove phenology characterization. Based on RSR nc , the five top-performing vegetation indices of different classes, including MERIS Terrestrial Chlorophyll Index (MTCI), Normalized Difference Infrared Index (NDII), Normalized Burnt Ratio (NBR), EVI and Near-Infrared Reflectance of vegetation (NIRv), were further employed for the extraction of phenometrics. Finally, the satellite-observed phenological patterns were intercompared with litter data. The results show that (1) chlorophyll- and water content-related vegetation indices outperformed the commonly used structure-related in characterizing seasonal patterns of mangroves, with EVI and NIRv showing earlier phenological patterns than MTCI, NDII, and NBR by 30–50 days. (2) An RQI-based model selection strategy enhanced reconstruction quality, while RSR nc effectively assessed the phenological characterization ability of different vegetation indices. (3) All VIs were negatively correlated with litter, with MTCI, NDII, and NBR achieving stronger correlations at shorter time lags than EVI and NIRv.