IDENTIFYING ACTIVITY STATUS OF AMAZONIAN FISHPONDS USING REMOTE SENSING

Aquaculture in the Amazon has emerged as a solution for food security and income generation, addressing declining natural fish stocks. This study assesses the use of remote sensing to identify and monitor aquaculture ponds, determining whether they are active (stocked with fish) or inactive. We used in situ remote sensing reflectance (Rrs) and water quality data including chlorophyll-a concentration, turbidity and Normalized Difference Chlorophyll Index, to identify active ponds. Data were collected at Embrapa’s experimental ponds in Palmas-TO across an entire production cycle (Aug/23 to Feb/24) on different aquaculture systems. Rrs was simulated for PlanetScope’s SuperDove orbital sensor (Rrs-Dove) and classified using: Spectral Angle Mapper, Euclidian distance and Mahalanobis distance. Mahalanobis distance achieved the best performance with an overall accuracy of 83%. Notably, Rrs-Dove successfully identified 100% of inactive ponds and 75% of active ponds. This approach offers a valuable tool for sustainable and strategic aquaculture management in the region.