Fine-scale Offshore Aquaculture Boundaries and Optical Responses in Sansha Bay, China

Fine-scale aquaculture boundary mapping achieved an mIoU of 90.82% and an mF1-score of 95.11% using a Multi-Scale Feature Enhanced Vision Transformer (MSFE-ViT). The dataset, authored by Yujie Ma and updated in May 2026, includes 5.3 GB of TIF and ZIP files containing geospatial data for Sansha Bay, Fujian, China. It integrates chlorophyll and turbidity indices to construct an Aquaculture-Adjacent Optical Response Index (AORI) for distance–decay analysis.

Use Cases

• Mapping raft and cage aquaculture boundaries based on the MSFE-ViT segmentation model.
• Analyzing spatial variation of optical responses in waters adjacent to aquaculture facilities based on the NDCI and NDTI indices.
• Conducting distance–decay diagnosis of aquaculture-adjacent optical responses based on Euclidean Distance Transform analysis.
• Monitoring regional changes in aquaculture distribution based on multi-temporal snapshot results.
• Comparing pixel-aligned optical responses across inner-bay, estuarine, bay-mouth, and offshore settings.

Strengths

• The boundary extraction model achieved high accuracy metrics: mIoU of 90.82% and mF1-score of 95.11%.
• The dataset provides multi-temporal snapshots showing variation in aquaculture distribution.
• The analytical framework integrates fine-scale mapping with optical response indices (NDCI, NDTI) and distance–decay analysis.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The data is specific to Sansha Bay, Fujian, China, which may limit generalizability.

Provenance

Source

figshare

Collection Method

Developed using a Multi-Scale Feature Enhanced Vision Transformer (MSFE-ViT) for boundary extraction and integrated optical index analysis.

Freshness

Last updated 2026-05-22 21:27:21

Geography

Sansha Bay, Fujian, China