PCSAI: a phenological characteristics-based Spartina alterniflora index derived from a time-series of remote sensing images

The invasion of Spartina alterniflora (S. alterniflora), a non-native invasive salt marsh plant, seriously threatens the health of China's coastal ecosystems. The diverse vegetation of complex landscapes in coastal zones creates problems related to the phenomenon where the same object reflects light in different spectra and when different objects reflect light in the same spectra, leading to low extraction accuracy in the extraction of S. alterniflora. This study developed a phenological characteristics-based S. alterniflora index (PCSAI) using time-series satellite remote sensing data. Initially, cloud cover screening and median composites were generated from the time-series. Subsequently, the normalized difference vegetation index (NDVI) was computed, leading to the creation of an NDVI time-series dataset. Then, the Savitzky–Golay algorithm was used to filter the NDVI data, and a two-term Fourier function was combined to calculate the phenological characteristic coefficients. The phenological characteristic curves for various land cover types, along their distribution within phenological characteristic space were examined to construct the PCSAI. Finally, the PCSAI was segmented using the threshold segmentation method, followed by post-processing that combined digital elevation model data and mathematical morphology techniques to obtain the remote sensing extraction results for S. alterniflora. Based on Sentinel-2 satellite remote sensing data, experimental results on China's Zhoushan Archipelago show that (1) the PCSAI successfully delineates the spatial distribution of S. alterniflora, attaining an average overall accuracy of 93.74% and average Kappa coefficient of 0.88. (2) A 10-meter resolution spatiotemporal distribution map of S. alterniflora in the Zhoushan Archipelago from 2019 to 2023 was performed, revealing that its coverage expanded from 2.97 km² to a peak of 6.79 km², with approximately 5.50 km² of other landcovers converted into S. alterniflora. This study presents a practical approach for remote sensing monitoring of S. alterniflora, contributing valuable support for the high-quality development of coastal zone.