2023-03-15-ADB-Assessing_Spatiotemporal_Differences_in_Shrimp_Ponds_Using_Remote_Sensing_Data_and_Machine_Learning_Algorithms_20页_1mb

Summary of ADBI Working Paper 1366

Objective

This study maps and monitors spatiotemporal differences in shrimp ponds using remote sensing data and machine learning in Cai Doi Vam Township, Ca Mau Province, Vietnam.

Methods

Used high-resolution satellite imagery from DMC-3 (TripleSat) and Jilin-1, along with ground survey data. Applied unsupervised classification for land use/land cover mapping and supervised classification (maximum likelihood) for separating shrimp ponds. Accuracy assessment involved error matrices with a total of 56 validation samples.

Results

Achieved high classification accuracy (87.5% in 2019 and 89.29% in 2022). Found a 15.2% reduction in shrimp pond area between 2019 and 2022, with 66 hectares drying out. Vegetation cover increased significantly, indicating changes in land use and reduced aquaculture activity.

Conclusion

Remote sensing and machine learning enable effective monitoring of shrimp ponds for sustainable aquaculture, supporting decision-making and policy interventions.