Distribution of Mangrove Species Kandelia obovata in China Using Time-series Sentinel-2 Imagery for Sustainable Mangrove Management

Abstract

Mangrove management in China is converting from emphasis on occupied area to intrinsical quality. Mangrove species have varied ecological values, e.g., those having well-developed prop roots are more beneficial for coastal stabilization. The detailed distribution of mangrove species on a national scale remains a challenge, impeding species-specific applications in the ecology and management of mangroves, as well as evaluations of related Sustainable Development Goals. Although local-scale studies have explored various data sources, the variability of plant phenology is the key factor preventing their generalization from local areas to large latitudinal spans (e.g., the coast of China). The separability time period, defined by a separability metric for each scene of the image time series, provides a potential way to tackle the aforementioned problem. We conducted a case study on mapping Kandelia obovata in China based on Sentinel-2 time-series imagery, as it is a representative native mangrove species with the largest latitudinal span and tolerance to low temperatures. The proposed approach considered the separability between K. obovata and its typical co-occurring mangrove species, as well as that between K. obovata and salt marshes. The overall accuracy of the generated K. obovata map in China for 2020 reached 88.5% based on independently collected samples. The proposed approach is transferable to diverse mangrove species that inhabit a vast latitudinal span. The implications for sustainable mangrove management were discussed to reveal the benefits to precise management of mangroves, accurate biomass and carbon estimations accounting for species differences, and effective evaluation of mangrove ecosystem services.