Water Use Optimization
How might AI enable smarter water use and allocation across Washington’s agricultural system?
Washington State’s $12.9 billion agricultural economy depends on irrigation, with 80% of all freshwater withdrawals going to farms. That supply is under growing pressure. Droughts are becoming more frequent, driven by low snowpack and precipitation deficits that reduce summer water availability, when irrigation demand is at its peak.
2026 is the fourth consecutive year that Washington is under a drought emergency. In the Yakima Basin, water-right holders are forecast to receive just 44% of their normal supply. By the 2050s, projections show that seven out of every ten years will see snow droughts on average, compared to just one in five in the 1990s.
When water runs short, crops fail, orchards die, and farming families lose their livelihoods. Yet at every level of the water system, there are significant gaps in the tools, information, and incentives needed to manage this challenge effectively:
- On the farm, farmers are making irrigation decisions with incomplete information, often applying more water than crops need because the risk of under-irrigating is too high, and the evidence-informed precision tools that could help are out of reach for most small and medium-size operations.
- When it comes to water trading, farmers who have water to spare cannot easily transfer it to those who need it more. Water markets can be slow and costly to navigate driving willing buyers and sellers away, leaving water capacity underutilized while crops fail elsewhere.
- At the watershed scale, irrigation districts, tribal nations, and regulators are making allocation decisions without integrated, real-time data on how water moves through the system, leading to inefficiencies, allocation conflicts, and missed opportunities to distribute water where it is needed most.
- In policy and incentives, current policies and programs were designed to secure and protect water rights, and conservation incentives were not their original focus. As a result, farmers who reduce water use may see few direct rewards, and some may worry that conserving could affect their rights.
AI stands to address gaps in how water is managed, traded, allocated, and conserved across Washington agriculture.
Opportunity Areas
The opportunity areas below are meant to inspire, not prescribe. The best ideas may come from directions we haven’t anticipated. We welcome ideas that address one area or span several.
How might AI help farmers apply the right amount of water at the right time?
Many farmers lack the data and tools to make precise irrigation decisions, and the precision technologies that could help are too expensive for most small and medium-size operations.
Ideas might include:
- AI-driven irrigation scheduling linked to soil moisture sensing and satellite-based estimates of crop water use, integrated with farm management systems
- Cooperative models that pool costs to install sensors, data networks, and irrigation controls, making technology accessible to farms that cannot afford it individually
- Improved seasonal forecasting focused on the metrics most useful to irrigators. Seasonal timeliness, location specificity, and projection period duration all matter for crop and irrigation planning.
How might AI make it faster and less costly for farmers to lease or transfer unused water to those who need it most?
Water markets in Washington move slowly and cost a lot to navigate, so water often sits unused on one farm while crops fail on another.
Ideas might include:
- Digital platforms that reduce the time and cost of leasing unused water to other farms, cities, or industries
- Water banking tools compatible with Washington’s Trust Water Rights Program so farmers can conserve water without losing their rights
How might AI give irrigation districts, tribal nations, and regulators a shared, real-time picture of water capacity and flow through the system?
Irrigation districts, tribal nations, and regulators are making critical allocation decisions with fragmented data and no shared framework.
Ideas might include:
- Integrated platforms combining farm-level, district-level, and watershed-level data currently operating in isolation, including digital twins, forecast capabilities, and spatiotemporal products
- AI models for snowpack, streamflow, and groundwater recharge that give managers subseasonal visibility into water availability
How might AI help design and implement incentives that reward farmers for conserving water without putting their water rights at risk?
Existing policies and programs offer limited incentives for conservation and few rewards for farmers who reduce water use.
Ideas might include:
- Tools that help farmers identify and qualify for existing conservation incentives and programs, reducing the uncertainty and paperwork that discourage participation
- Policy analysis tools that help regulators understand system-wide impacts of water transactions, including effects on tribal reserved rights
