Capacity building and knowledge transfer in seaweed mapping in Indonesia

Early work in using satellite imagery to map seaweed production has shown there is potential for the information to be used to benefit the industry.

Partnership for Australia-Indonesia Research has used refining and enhancing satellite image processing and analysis methods to develop a seaweed production mapping and monitoring system.

This deep learning system will help the industry better understand the production dynamics of an area and support better coordination along the value chain.

This study uses special images to create a model that can automatically map seaweed farms in coastal areas. The images are processed and labelled using computer techniques to train a deep-learning model that accurately identifies seaweed in the images.

The deep learning model can be applied to time-series satellite images to produce multitemporal seaweed farm maps that can be used to monitor seaweed farm dynamic spatially.

A number of findings have come from the research including:

• The time-series map of seaweed production areas can be integrated with biomass production information collected from farmers of the local DKP office to estimate the multitemporal biomass production.
• The deep learning model can be applied to time-series satellite images to produce multitemporal seaweed farm maps that can be used to monitor seaweed farm dynamic spatially.

Click here to read the full report Capacity building and knowledge transfer in seaweed mapping in Indonesia

We propose a number of recommendations including:

• Our satellite-imagery based deep learning model should be used to automatically map seaweed farms in coastal areas.
• Seaweed production maps, based on the deep learning model, should be used to monitor the planting and harvesting cycles and to estimate the total biomass production in a specific area.

Feature image by PAIR.