Spatial extraction of sea-cucumber aquaculture ponds using remote sensing spectral and temporal features

The spatial distribution of aquaculture ponds plays a critical role in the layout, management, and evaluation of the aquaculture industry. While extensive research has been conducted on aquaculture pond extraction and monitoring, studies focusing on the differentiation of aquaculture ponds by species remain limited. The similar shapes and spectral features of water bodies associated with different species pose a challenge for extraction. A method for extracting sea-cucumber aquaculture ponds is proposed based on spectral temporal features using Sentinel-2 satellite imagery in this study. The method involves selecting the optimal sensitive spectral bands or combinations to construct two remote sensing indices of land-based aquaculture ponds. Using a time-series dataset of these indices, three key features—the mean and two slopes—are extracted. The corresponding time windows and thresholds for these key features are identified to develop a decision-tree algorithm for extracting sea-cucumber ponds. This method was applied to coastal aquaculture zones in Liaoning Province, China, to identify the spatial distribution of sea-cucumber aquaculture ponds in 2016 and 2023. The results showed that: (1) the proposed method achieved high extraction accuracy, with an overall accuracy of 79.24%; (2) Total area of sea-cucumber ponds in Liaoning Province was 931.08 km 2 , primarily located along the Huludao Xingcheng-Jinzhou Linghai and Yingkou Xishi-Dalian Zhuanghe coastal zones; (3) Over the past seven years, the area of sea-cucumber ponds increased by 624.57 km 2 , with expansion concentrated on the northwest coast of Liaodong Bay and both the eastern and western sides of the Liaodong Peninsula. These findings provide scientific support for the sustainable development of sea-cucumber aquaculture.