The next frontier in irrigation: Decision-making with AI

As water resources tighten and pumping costs rise, irrigation efficiency is becoming an increasingly important driver of profitable crop production. In Nebraska, where irrigation depends heavily on the Ogallala Aquifer, producers face growing pressure to apply water more precisely than ever before.

Traditional irrigation scheduling tools such as soil-moisture sensors and evapotranspiration scheduling are highly effective but often operate independently, leaving producers to manually integrate multiple data streams without providing clear decisions. This is where emerging generative AI tools may offer new opportunities.

Generative AI models such as ChatGPT are designed to synthesize information quickly and communicate recommendations in a practical, conversational format. Unlike traditional decision-support tools that rely on fixed rules or dashboards, generative AI can integrate multiple inputs (such as sensor readings, weather forecasts, soil properties and crop stage) to provide a clear recommendation along with the reasoning behind it.

For irrigators, the potential value lies not in replacing agronomic expertise, but in helping interpret complex information more quickly and supporting more timely management decisions.

Testing AI irrigation decisions in Nebraska’s TAPS program

We evaluated the potential of generative AI for irrigation decision-making through the University of Nebraska-Lincoln’s Testing Ag Performance Solutions (TAPS) program. TAPS is a season-long farm management competition in which participants submit real-world decisions on irrigation, fertilizer and other inputs. Those decisions are implemented on university research fields, allowing teams to compare management outcomes under the same field conditions.

In the 2024 season, we used the ChatGPT-4o model to generate all management decisions, despite having no farming experience. The model was provided with key information such as crop type and growth stage, past and forecasted weather conditions, irrigation costs and previously applied irrigation. Soil-moisture sensor data were also included, either as direct soil water content values or screenshots showing moisture trends. We then asked the model specific questions. For example: “Considering all the data provided and a maximum irrigation depth of 1 inch, how much irrigation should be applied tomorrow?” The recommended irrigation depth was then implemented within the competition.

In 2025, the approach evolved further. Four custom GPTs were developed to compete in four separate TAPS contests in North Platte and Mead, Nebraska. Each model was configured with competition-specific information.

The AI-managed plots achieved the following grain yields in bushels per acre, along with their yield and profitability rankings among participating teams:

Overall, the results suggest that AI-assisted management can perform competitively with top human-managed systems under structured conditions. However, additional testing across more years, sites and management scenarios is needed before drawing broader conclusions.

How producers can use AI

We identified two practical ways generative AI tools could support irrigation management.

1. AI as an information and interpretation tool

One immediate benefit of generative AI is its ability to summarize and interpret irrigation-related data quickly. Producers can use AI to interpret weather forecasts, estimate crop water demand, analyze soil moisture trends and translate technical data into actionable insights. Instead of viewing sensor graphs in isolation, producers can ask AI systems to explain what the trends suggest about soil water depletion, stress risk or refill needs.

However, the quality of the response depends on the information provided. Users must clearly describe the situation and include relevant data. Prompts can be entered by typing or voice, and supporting files such as PDFs, images or documents can be attached.

2. AI as a decision-support tool

Generative AI can also assist with irrigation decisions when provided with accurate and sufficient information. For example, a grower might input soil texture, soil profile, current soil water content, growth stage, rooting depth and irrigation system limits (e.g., maximum application per event). Using this information, the AI can estimate soil water depletion and recommend an irrigation depth consistent with field and irrigation system constraints.

AI tools can also assist with more complex decisions. For example, when users include current grain prices and irrigation costs, the system can evaluate whether the expected yield benefit of an irrigation event is likely to cover the pumping cost.

Best practices and limitations

While generative AI tools show strong potential, they must be used carefully. Key considerations include:

• AI recommendations should always be checked against trusted agronomic principles, ET data and sensor measurements.
• These tools do not replace local expertise, crop consultants or field scouting.
• AI outputs are only as reliable as the data and assumptions provided.
• Fully automated irrigation control should not occur without safeguards, professional oversight and validated models.

Used appropriately, generative AI can function as an assistant, not a replacement for irrigation decision-making.

Opportunities for irrigation equipment and sensor manufacturers

The future impact of generative AI in irrigation will depend on stronger integration with high-quality data and existing irrigation technologies. Future opportunities may include:

• Linking AI tools with validated weather and ET networks.
• Integrating crop prices, fertilizer costs and pumping expenses into whole-farm optimization.
• Supporting AI-enabled irrigation control systems with built-in safety limits.
• Improving user-friendly decision platforms for producers.

With continued development, these tools may help improve both profitability and long-term water sustainability.

Nipuna Chamara, PhD, is a research assistant professor in the Department of Biological Systems Engineering at the University of Nebraska-Lincoln.

Saleh Taghvaeian, PhD, is an associate professor and irrigation engineer in the Department of Biological Systems Engineering and a faculty fellow at the Daugherty Water for Food Global Institute at the University of Nebraska-Lincoln.

Yufeng Ge, PhD, is a professor and advanced sensing systems engineer in the Department of Biological Systems Engineering and director of plant phenomics and a faculty fellow at the Daugherty Water for Food Global Institute at the University of Nebraska-Lincoln.