The Philippines has the fifth largest coastline in the world, which makes fishing a vital source of livelihood for Filipinos. With millions whose primary livelihood revolve around the aquaculture industry, it is imperative to ensure the security of their jobs while taking care of natural marine resources. 

Fish is a staple in the Filipino diet. Data shows that the average consumption of fish per person in 2018 is at 36.8 kilograms¹. In 2015, an average Filipino household allocates their third largest share of food expenses for fish and other marine products. Cereals and grains have the biggest share at 11.7%, while 5.4% is spent on meat and 5% is spent on fish². 

Every year, tons of fish and other marine products are acquired in our waters. In 2020 alone, the country produced an estimated amount of 4,403 thousand metric tons of fish and marine products³. The Philippines’ dependence on the aquaculture industry motivates the need to identify and map aquafarms such as fishponds, fish pens and fish cages nationwide. 

To achieve this, the DOST-Advanced Science and Technology Institute (DOST-ASTI) and the Philippine Statistics Authority (PSA) signed an agreement to collaborate on the use of satellite data for PSA’s 2022 Census on Agriculture and Fisheries (CAF)⁴. This census aims to build an inventory of agricultural and fisheries resources around the country. 

“The maps will be instrumental in building sustainable communities. Mapping aquafarm locations and structures will enable concerned government agencies, business sectors, and organizations to exert targeted efforts to boost the livelihood of the people dependent in aquafarming as well as to improve the aquafarm structures/practices.” PSA said. 

Fish pens in Bolinao, Pangasinan. Using Artificial Intelligence (AI), the AI4CAF collaboration was able to map fish pens using synthetic aperture radar (SAR) satellite images. 

From NovaSAR-1’s stripmap (6-meter resolution) radar images and through the Artificial Intelligence for Census of Agriculture and Fisheries (AI4CAF) Project, the teams were able to detect fish pens and fish cages in certain parts of the country. The satellite data from NovaSAR-1 provides the agencies with high-quality radar images used to map and identify unrecorded aquaculture structures. 

Using the trained convolutional neural networks (CNN) models developed by DOST-ASTI researchers, the teams were able to detect fish pens in SAR images taken from representative locations in Pangasinan and Laguna. The fish pens were digitized on calibrated and georeferenced SAR images and were fed to the CNN model as training data. This model can now be used to detect fish pen structures in radar images captured in other areas. 

Fishpens detected in Laguna using NovaSAR-1's images. 

Ground validation for the CNN model was performed in representative areas from Hagonoy, Bulacan. A model to detect fishponds along shorelines was generated and was able to identify majority of fishpond areas. Models for fish pens and fish cages were also created but additional training data and studies are currently being developed to improve outputs obtained from these models. 

The Philippines acquired 10% of tasking and data acquisition capabilities of NovaSAR-1 in 2019, a Synthetic Aperture Radar (SAR) satellite equipped with an Automatic Identification System (AIS) receiver designed by UK-based company Surrey Satellite Technology Ltd. 

References: 

¹ https://www.bfar.da.gov.ph/profile?id=2  

² https://psa.gov.ph/sites/default/files/ais_food_consumption_and_nutrition%202017.pdf   

³ https://psa.gov.ph/content/fisheries-situation-report-january-december-2020   

⁴ see recent Census on Agriculture and Fisheries here: https://psa.gov.ph/content/census-agriculture-and-fisheries-caf