Publications

Gao, Y; Hu, ZL; Wang, Z; Shi, Q; Chen, D; Wu, S; Gao, YJ; Zhang, YZ (2023). Phenology Metrics for Vegetation Type Classification in Estuarine Wetlands Using Satellite Imagery. SUSTAINABILITY, 15(2), 1373.

Abstract
While the efficiency of incorporating phenology features into vegetation type classification, in general, and coastal wetland vegetation classification, in particular, has been verified, it is difficult to acquire high-spatial-resolution (HSR) images taken at appropriate times for vegetation identification using phenology features because of the coastal climate and the HSR satellite imaging cycle. To strengthen phenology feature differences, in this study, we constructed vegetation phenology metrics according to vegetation NDVI time series curves fitted by samples collected from the Linhong Estuary Wetland and Liezi Estuary Wetland based on Gao Fen (GF) series satellite images taken between 2018 and 2022. Next, we calculated the phenology metrics using GF series satellite imagery taken over the most recent complete phenology cycle: 21 October 2020, 9 January 2021, 19 February 2021, and 8 May 2021. Five vegetation type classifications in the Linhong Estuary Wetland were carried out using single images of 21 October 2020 and 8 May 2021, along with their combination and the further addition of phenology metrics. From our comparison and analysis, the following findings emerged: Combining the images taken in 21 October 2020 and 8 May 2021 provided better vegetation classification accuracy than any single image, and the overall accuracy was, respectively, increased from 47% and 48% to 67%, while the corresponding kappa was increased from 33% and 34% to 58%; however, adding phenology metrics further improved the accuracy by decreasing the effect of some confusion among different vegetation types, and the overall accuracy and kappa were further improved to 75% and 69%, respectively. Though some problems remain to be further dealt with, this exploration offers helpful insights into coastal wetland vegetation classification using phenology based on HSR imagery.

DOI:
10.3390/su15021373

ISSN:
2071-1050