Washington State Department of Natural Resources uses Azure AI to automate aquatic resource monitoring, saving months of experts’ time

Created in 1957, the Washington State Department of Natural Resources (DNR) manages 5.6 million acres of public land. Its mission is to sustain and protect the health and productivity of Washington’s lands and waters to meet the needs of present and future generations.

The Aquatic Resources Division of the DNR fosters science-based stewardship of Washington’s unique underwater habitats and the diverse wildlife that they support. The division contains several specialized groups that together oversee 2.6 million acres of submerged aquatic lands and embedded resources. One group, the Nearshore Habitat Program, gathers hundreds of hours of underwater video footage each year that must be carefully analyzed to track the status and trends of marine vegetation.

The Nearshore Habitat Program saw an opportunity to glean useful data in ways that made better use of time and labor. In an effort to find greater efficiency, the team explored AI solutions using Azure Cloud Services for machine learning. 

Introducing AI for more efficient analysis of 350 hours of underwater video footage

As a government agency, the DNR stewards the state’s natural resources for environmental justice and long-term preservation while fostering the use of the resources for the benefit of the people of Washington. “There’s tension there,” says Peter Dowty, Application Developer at the DNR. “As scientists, our purpose is to get solid scientific information on the table for decision-making about those resources.”

To track the Eelgrass Vital Sign Indicator for the greater Puget Sound region, the DNR has specifically monitored the abundance and distribution of native eelgrass (Zostera marina) since 2000. Each summer, scientists in the DNR’s Nearshore Habitat Program use a research vessel to gather 350 hours of underwater video footage from 120 different sites. All of that video footage must then be reviewed manually by someone trained to identify and classify distinct types of seagrass, attaching each identification to precise GPS coordinates. This process is highly labor intensive, taking about three months of two scientists’ time.

In 2022, following conversations with a customer success account manager from Microsoft, the ecologists decided to try something different. “We’ve been trying to make this process more efficient, and that’s where Microsoft comes in,” says Bart Christiaen, Natural Resource Scientist at the DNR. AI might sound daunting to people without prior experience. But a catalyst for the eelgrass monitoring project came from the Washington State Commissioner of Public Lands, Hilary Franz. Franz encouraged innovation and creative problem solving in the DNR’s strategic plan and made AI part of the agency’s Snohomish Watershed Resilience Action Plan. Because the DNR’s Microsoft license included access to Microsoft Azure, the Nearshore Habitat Program decided the time was right to give AI a try. 

Saving months of specialists’ time with automated analysis

With the help of the Microsoft customer success support team, the ecologists used Azure Cognitive Services to pilot the DNR’s first AI project. “One of the things we’re working on is automated image recognition,” says Christiaen. “We upload a video file to the cloud, the video is split into individual frames, and then those frames are analyzed for the presence or absence of eelgrass.” Azure Machine Learning empowered the scientists to build and deploy high-quality data models, helping them gain the benefits of machine learning even without having computer science degrees. Meanwhile, Azure Data Lake Storage provides scalable and secure storage for the ever-growing hours of digital video footage and geographic information systems data.

The team is currently in a soft launch of this new method, using 2022 data to compare human and AI identifications as a test of how well AI stacks up. If the models are sufficiently trained, and the Nearshore Habitat Program team can demonstrate efficient monitoring of video footage of eelgrass using AI, the process could be fully automated in the coming years.

The team has achieved this technological advancement without any formal training in AI and machine learning. Using Azure, and with support from Microsoft, the DNR ecologists applied AI to their project quickly and with relative ease. “All of our backgrounds are in ecology. None of us had experience with AI or the Azure platform,” says Christiaen. “So, it was a steep learning curve for us, but the fact that we are where we are is to the credit of the Microsoft support team.”

The transition to AI will radically reduce the technician time needed for video analysis. What took several months for two people to do manually will now take weeks, with only one person preparing and uploading the videos to Azure and doing quality control. This is precious time that the team can use for other specialized projects, accelerating work in the field. “This technology promises to be a game changer for improving efficiency and replicability in our habitat monitoring efforts. It’s really exciting what we can accomplish,” says Cynthia Catton, Science Advisor at the Aquatic Resources Division. 

Envisioning new opportunities with AI

The DNR’s venture into what Microsoft CEO Satya Nadella has called a “golden age of AI” is turning its Aquatic Resources Division into an industry leader. The DNR’s lead in using AI for custom vision has other departments and divisions that are also working toward sustainability and greater analytical capabilities taking note. “AI is a growing field in natural resources because it expands what we can accomplish with the limited funding for this type of work. If we can streamline some processes using AI, that’s an opportunity that we should take advantage of,” says Jeff Gaeckle, Seagrass Ecologist and Co-Lead of the Nearshore Habitat Program at the DNR. “And we don’t have the resources for in-house support, so collaborating with Microsoft has been great.”

Looking ahead, the DNR has plans to add Microsoft Power BI to help to create streamlined reports of its results. Its investment in Azure today helps the DNR stay ready to meet the future’s inevitable changes with purpose and vision. “AI is opening doors for us and helps us to step up to where we can see new opportunities,” says Dowty. “The Microsoft folks have shown us how to approach processes in an automated way. It gave us a different perspective, and that was really important.”

Learn more about the DNR’s Eelgrass Vital Sign Indicator.