Mapping China Aquaculture Ponds: Integrating a New Aquaculture Index With Machine Learning
Abstract Aquaculture Pond (AP) plays a vital role in ensuring food security, driving economic development, conserving resources, and maintaining ecological balance. Thus, accurately delineating the extent of AP is critical for effective policy‐making in aquaculture. However, existing methods for lar...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-06-01
|
| Series: | Earth's Future |
| Online Access: | https://doi.org/10.1029/2024EF005637 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849433270641491968 |
|---|---|
| author | JianChun Chen Chen Lin Kun Xue Ke Song ZhiGang Cao RongHua Ma DanHua Ma YiJun Tong |
| author_facet | JianChun Chen Chen Lin Kun Xue Ke Song ZhiGang Cao RongHua Ma DanHua Ma YiJun Tong |
| author_sort | JianChun Chen |
| collection | DOAJ |
| description | Abstract Aquaculture Pond (AP) plays a vital role in ensuring food security, driving economic development, conserving resources, and maintaining ecological balance. Thus, accurately delineating the extent of AP is critical for effective policy‐making in aquaculture. However, existing methods for large‐scale extraction of AP face challenges, such as difficulty in transferring segmentation thresholds and confusion with similar land features, which limits the accurate determination of their spatial distribution. This study focuses on AP in China, developing a tailored spectral index for AP extraction and creating an optimized classification method for large‐scale, automated AP extraction by integrating the WVndapi index with machine learning techniques. Using high‐resolution Sentinel‐2 data from 2023 and leveraging the Google Earth Engine, a nationwide AP distribution map was generated. The results indicate that: (a) The optimized WVndapi index extraction results indicate that the overall accuracy (OA) of AP identification across the nation reached 91%, with Cohen's Kappa of 0.88. (b) At the national scale, the spatial distribution of AP shows a pattern of higher density in the north and lower density in the south, with more AP in the east than in the west. Notably, inland AP account for 15% of the national total. (c) The contours and shapes of AP extracted used WVndapi index closely match the high‐precision results obtained through manual digitization (0.43 m), effectively distinguishing AP from confounding features such as gully, lake, river, and shadow. In summary, the establishment of the WVndapi index overcomes the limitations of confusion and misclassification among similar land covers, achieving the goal of adaptive threshold at a large scale. |
| format | Article |
| id | doaj-art-1cc6ac76e42044e4b03607e57771d532 |
| institution | Kabale University |
| issn | 2328-4277 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | Wiley |
| record_format | Article |
| series | Earth's Future |
| spelling | doaj-art-1cc6ac76e42044e4b03607e57771d5322025-08-20T03:27:07ZengWileyEarth's Future2328-42772025-06-01136n/an/a10.1029/2024EF005637Mapping China Aquaculture Ponds: Integrating a New Aquaculture Index With Machine LearningJianChun Chen0Chen Lin1Kun Xue2Ke Song3ZhiGang Cao4RongHua Ma5DanHua Ma6YiJun Tong7State Key Laboratory of Lake and Watershed Science for Water Security Nanjing Institute of Geography and Limnology Chinese Academy of Sciences Nanjing ChinaState Key Laboratory of Lake and Watershed Science for Water Security Nanjing Institute of Geography and Limnology Chinese Academy of Sciences Nanjing ChinaState Key Laboratory of Lake and Watershed Science for Water Security Nanjing Institute of Geography and Limnology Chinese Academy of Sciences Nanjing ChinaGeological Survey of Jiangsu Province Nanjing ChinaState Key Laboratory of Lake and Watershed Science for Water Security Nanjing Institute of Geography and Limnology Chinese Academy of Sciences Nanjing ChinaState Key Laboratory of Lake and Watershed Science for Water Security Nanjing Institute of Geography and Limnology Chinese Academy of Sciences Nanjing ChinaState Key Laboratory of Lake and Watershed Science for Water Security Nanjing Institute of Geography and Limnology Chinese Academy of Sciences Nanjing ChinaState Key Laboratory of Lake and Watershed Science for Water Security Nanjing Institute of Geography and Limnology Chinese Academy of Sciences Nanjing ChinaAbstract Aquaculture Pond (AP) plays a vital role in ensuring food security, driving economic development, conserving resources, and maintaining ecological balance. Thus, accurately delineating the extent of AP is critical for effective policy‐making in aquaculture. However, existing methods for large‐scale extraction of AP face challenges, such as difficulty in transferring segmentation thresholds and confusion with similar land features, which limits the accurate determination of their spatial distribution. This study focuses on AP in China, developing a tailored spectral index for AP extraction and creating an optimized classification method for large‐scale, automated AP extraction by integrating the WVndapi index with machine learning techniques. Using high‐resolution Sentinel‐2 data from 2023 and leveraging the Google Earth Engine, a nationwide AP distribution map was generated. The results indicate that: (a) The optimized WVndapi index extraction results indicate that the overall accuracy (OA) of AP identification across the nation reached 91%, with Cohen's Kappa of 0.88. (b) At the national scale, the spatial distribution of AP shows a pattern of higher density in the north and lower density in the south, with more AP in the east than in the west. Notably, inland AP account for 15% of the national total. (c) The contours and shapes of AP extracted used WVndapi index closely match the high‐precision results obtained through manual digitization (0.43 m), effectively distinguishing AP from confounding features such as gully, lake, river, and shadow. In summary, the establishment of the WVndapi index overcomes the limitations of confusion and misclassification among similar land covers, achieving the goal of adaptive threshold at a large scale.https://doi.org/10.1029/2024EF005637 |
| spellingShingle | JianChun Chen Chen Lin Kun Xue Ke Song ZhiGang Cao RongHua Ma DanHua Ma YiJun Tong Mapping China Aquaculture Ponds: Integrating a New Aquaculture Index With Machine Learning Earth's Future |
| title | Mapping China Aquaculture Ponds: Integrating a New Aquaculture Index With Machine Learning |
| title_full | Mapping China Aquaculture Ponds: Integrating a New Aquaculture Index With Machine Learning |
| title_fullStr | Mapping China Aquaculture Ponds: Integrating a New Aquaculture Index With Machine Learning |
| title_full_unstemmed | Mapping China Aquaculture Ponds: Integrating a New Aquaculture Index With Machine Learning |
| title_short | Mapping China Aquaculture Ponds: Integrating a New Aquaculture Index With Machine Learning |
| title_sort | mapping china aquaculture ponds integrating a new aquaculture index with machine learning |
| url | https://doi.org/10.1029/2024EF005637 |
| work_keys_str_mv | AT jianchunchen mappingchinaaquaculturepondsintegratinganewaquacultureindexwithmachinelearning AT chenlin mappingchinaaquaculturepondsintegratinganewaquacultureindexwithmachinelearning AT kunxue mappingchinaaquaculturepondsintegratinganewaquacultureindexwithmachinelearning AT kesong mappingchinaaquaculturepondsintegratinganewaquacultureindexwithmachinelearning AT zhigangcao mappingchinaaquaculturepondsintegratinganewaquacultureindexwithmachinelearning AT ronghuama mappingchinaaquaculturepondsintegratinganewaquacultureindexwithmachinelearning AT danhuama mappingchinaaquaculturepondsintegratinganewaquacultureindexwithmachinelearning AT yijuntong mappingchinaaquaculturepondsintegratinganewaquacultureindexwithmachinelearning |