Other pages in this section

Microsoft’s AI for Good Lab awards UW-TealWaters team for new wetland mapping technology

Microsoft’s AI for Good Lab awarded TealWaters one of 20 grants who will receive a combined $5 million in assistance over the next two years for “game-changing organizations and projects helping solve today’s challenges and paving the way for a better future.”

TealWaters is a multidisciplinary collaboration, co-led by UW scientists Monika Moskal and Meghan Halabisky, to transform Washington state’s water management capacity by providing tools that inform and guide wetlands planning, protection, and restoration. TealWaters plans to support AI model testing beyond the scope of its existing tools to increase communities’ resilience to climate change and environmental stressors. 

The School of Environmental and Forest Sciences and EarthLab connected with the team for a Q&A about this award.

Q: Can you tell us more about TealWaters? What are the goals of the project and how does artificial intelligence play a role?

A: TealWaters is a multidisciplinary team of UW and private sector researchers and community partners. Our mission is to map and model wetlands, their hydrologic connections, and their benefits to society so natural resource managers and communities can identify, govern, and restore wetlands and their essential functions. Our platform identifies where wetlands are, where they can be restored, and estimates their specific benefits (e.g. flood protection, water quality, food provisioning, biodiversity – to name a few). Our ultimate goal is to increase communities’ resilience to climate change and environmental stressors. 

The MS AI for Good grant will fund two critically needed datasets for Washington state: 1) an improved map of wetland inventory for Washington state at high resolution (1-5m) and 2) the first-ever statewide map of high carbon wetlands. It will support development of new models that provide insights on the hydrologic drivers of wetlands (e.g. groundwater, surface water) and their dynamics. In the process we will evaluate the capacity of advanced AI/ML wetland mapping and statistical carbon modeling approaches to improve upon TealWaters’ strong existing models. Any new approaches can be applied here in Washington and in other regions in the US and beyond to enhance wetland conservation, restoration, and climate mitigation efforts.

Q: How would accurately mapping wetlands support different communities in the region? 

A: Wetlands offer natural solutions to a lot of environmental problems. Accurate maps – of where wetlands are, where they used to be, and where they can be restored – therefore are essential for communities to manage wetlands for the things they care about most. Our team has strong partnerships across Washington State, including the WA Department of Ecology, WA Conservation Commission, Tulalip Tribes, The Nature Conservancy of Washington, King County, and national partners such as the US Fish and Wildlife Service. We also have an extensive stakeholder and user network, including tribal governments, federal and state agencies, conservation organizations, natural resource managers, researchers, private industry, landowners, and community groups. Through these partnerships, our products are deeply informed by the specific needs of diverse users across the state. 

Beyond identifying wetland locations, communities also need and want to know the specific functions and benefits different wetlands are providing, and their role within the larger hydrologic system. In particular, inland wetlands serve as critical carbon sinks, storing 10 times more carbon than coastal wetlands. However, in Washington, peatlands and forested wetlands remain unclassified with most completely missing from wetland inventories. Without accurate data, Washington risks losing some of our most effective natural carbon sinks, which are essential for meeting climate goals. These same wetlands virtually always provide a wide range of other, layered benefits such as improving water quality, reducing flooding, providing habitat and green space, cooling urban centers, and more – the focus of future work. Better wetland maps are also key to understanding the hydrologic drivers of wetlands and their connections to other waterbodies and water sources. This project will allow us to advance these tools as well.

Q: Your teammates have a wide range of expertise and you have worked with many different communities to develop this project. How does your interdisciplinary and community-engaged approach inspire the process and affect the outcome?

A: Our community-engaged, user-based, multidisciplinary process is the very heart of our work. Effective modeling of wetlands and their functions is a creative scientific process that requires convergence of the knowledge, perspectives, and analytical skills of many scientific disciplines. Guiding this development for real-world applications and to meet and respect community needs also requires collaboration across the sciences and humanities and the creative engagement and insight of our partners in their specific and diverse contexts. The combination allows our products to be informed, useful, and far more creative than they ever could be in the absence of each of these partnerships. 

Q: What’s next for TealWaters? How will this award support these goals?

A: From a technical standpoint, we are in the process of developing workflows (software) for mapping and modeling a wide range of wetland functions that eventually will extend beyond Washington state. This award will advance three specific models and datasets – wetland intrinsic potential, wetland carbon potential, and hydrologic drivers (surface water, groundwater) – for Washington state and support development of the relevant cloud-based data and computing architecture. In the process, we’ll test a new set of advanced models that will inform and guide product development now and in the future. 

From a mission standpoint, our work lands in helping natural resource managers and communities to identify, govern, and restore wetlands and their functions, and increasing communities’ resilience in the process. We aim for the tools developed through this award to be adopted by state and local agencies, with direct integration into Washington’s Climate Action Plan and wetland management strategies. We are partnering with the Washington Department of Ecology to ensure our data is transmitted to county-level planners and wetland staff. We will also collaborate with stakeholders to explore integration into carbon markets and state-level greenhouse gas monitoring. Across our partnerships and uses, the products developed through this award support our broader goals of helping communities effectively manage wetlands on the ground.

Q: In the project abstract your team writes, “Current maps of wetlands in the United States derive from an earlier generation of science and are limited, often inaccurate, and poorly linked to other kinds of spatial information.” How will AI, advances in wetland science, computing, remote sensing, and geospatial tools help your team better inform proactive conversations about water, water management, and wetlands policy?

A: AI is all about pattern recognition. Recent advances in AI, e.g. in modeling imagery and time series data, could make it possible to better recognize and understand patterns in wetland dynamics, like seasonal changes or long-term shifts in the dynamics of water on the landscape. Improved mapping and deeper understanding of the hydrologic drivers of wetlands will give a more complete, accurate view of wetlands as essential water resources and hydrologic connectors – i.e. as key components of how water moves and is stored across the landscape. The resources support community conversations and decision-making amidst the complex tradeoffs that exist everywhere between wetlands protection, restoration, and development, especially in the context of a shifting climate. These advances ultimately will help decision-makers and policymakers make more informed decisions about water and water management at the scales and in the contexts that are most relevant to them and their communities.

Q: How will emerging technologies like AI incentivize user participation and improve wetland models over time?

A: Wetlands are highly diverse in their structure and function, from small or ephemeral ponds to complex floodplains. Good tools need to adjust to local conditions and the priorities of different users who may care most about different functions that the wetlands are providing – e.g. carbon storage, flood control, wildlife habitat, water supply, and many more. Incentivizing different stakeholders to customize and inform AI mapping tools through their direct observations will hopefully lead to broadly useful models that use more domain-specific knowledge than would be possible otherwise.

Congrats to the TealWaters team!