Header image

Arizona State University, Conservation International & Thinking Machines win Climate Change AI Innovation Grant to scale climate smart aquaculture

April 20, 2022 blog-post artificial-intelligence big-data climate-action climate-change computer-vision environment geospatial machine-learning open-data remote-sensing satellite-imagery southeast-asia sustainability development geospatial ngo non-profit

A hectare of mangrove absorbs up to 10 times as much carbon as a similarly sized area of terrestrial rainforest. Mangroves also protect coastal communities from storm surges, coastal erosion, and sea level rise, while helping fish populations to thrive. Source: Conservation International / Photo by Tim Novello

Mangrove forests are critical for both slowing the effects of climate change and helping vulnerable communities adapt to the impacts. Unfortunately, fish and shrimp farming has destroyed as much as 38% of mangroves globally and especially in Southeast Asia, home to at least 45% of the world's blue carbon stocks and producer of over 22% of global aquaculture production.

To address this problem, the global nonprofit Conservation International (CI) has piloted the “Climate Smart Shrimp” program, which helps shrimp farmers to produce more food using less land area, freeing up ponds to be replanted with mangroves. CI estimates that the CSS approach could restore as many as 1.7 million hectares of mangrove forests worldwide — while producing the same amount or even more shrimp.

Source: Conservation International

But one of the biggest hurdles to scaling-out the CSS program is locating the exact shrimp ponds where the CSS program would work. Publicly available aquaculture maps don't have all the information or level of detail that CI needs to determine if a site is suitable for CSS. And with hundreds of thousands of aquaculture sites across Southeast Asia, it is impossible to physically inspect every potential location.

This image shows aquaculture ponds in the Philippines, with “extensive” farms on the left and “intensive” farms on the right. The CSS program can only be implemented on extensive shrimp farms that are very large in size, low output, and don’t make use of specialized equipment that can intensify production.

This is where data science can help. Thinking Machines will work with CI to apply computer vision to region-wide satellite imagery to rapidly locate and classify shrimp ponds that are potential candidates for the CSS program. This data will then be combined with other criteria — such as proximity to current or historical mangroves, roads, coastlines, populated areas, and exposure to climate hazards like sea level rise and coastal flooding — to identify the top 40,000 most suitable hectares in Southeast Asia for the CSS program.

This heatmap ranks aquaculture areas as high, medium, or low suitability for the CSS program using criteria defined by CI’s aquaculture and mangrove experts, and data curated and derived from earth observation data.

The project was awarded a research grant by Climate Change AI (CCAI), a global nonprofit that catalyzes work at the intersection of machine learning and climate change. CCAI’s Innovation Grants Program supports research projects that leverage AI or machine learning to address problems in climate change mitigation, adaptation, or climate science. The grant program is funded by Quadrature Climate Foundation and Schmidt Futures with Canada Hub of Future Earth as the fiscal sponsor. Thinking Machines worked with Conservation International and the Arizona State University School of Sustainability on the proposal, which was one of 13 winners from almost 200 submissions to the grant competition.

The project will take place this 2022 and will produce open-source code, datasets, interactive maps, analyses, and research. It will also leverage Thinking Machines’ Eco-Intelligence Solution, a service that provides cutting-edge data analytics to organizations working to protect and restore nature. This solution helps project developers, funders, and landowners more effectively plan and monitor their environmental initiatives by applying end-to-end analysis to unique environmental datasets derived with AI from the latest earth observation data.

Stay tuned in the coming months as we share updates on the project!

Want to learn more? Contact us at [email protected]

Header Photo Credit: Waranont (Joe) via Unsplash.

MORE STORIES

Making Panel Discussions Memorable with Data Visualization

Minutes are the default method of documenting and summarizing meetings. But they’re also about as interesting to read as your social media's terms and conditions.

Understanding Data Storage Solutions: Databases vs. Data Warehouses vs Data Lakes

Understand the differences between OLTP Databases, Data Warehouses, and Data Lakes, and when to use which for your business

Building a Twitter Big Data Pipeline with Python, Cassandra & Redis

We use some of our favorite tech to build a data ingestion, storage, and analysis infrastructure.