Project : USDA ARS

ARS Home » Plains Area » Fort Collins, Colorado » Center for Agricultural Resources Research » Water Management and Systems Research » Research » Research Project #441925

Research Project: Improving Crop Performance and Precision Irrigation Management in Semi-Arid Regions through Data-Driven Research, AI, and Integrated Models

Location: Water Management and Systems Research

2025 Annual Report

Objectives
Objective 1: Identify crop physiological trait networks and soil nitrogen processes that improve the performance of agricultural systems under water and nutrient stress. Sub-objective 1.A: Identify physiological trait networks that advance process-based plant growth models, artificial intelligence (AI)/statistical models, and conceptual understanding of crop stress physiology. Sub-objective 1.B: Identify plant and soil processes that determine crop nitrogen requirements under varying water availability. Sub-objective 1.C: Develop rapid and cost-effective phenotyping methods to quantify complex physiological traits across genotypes. Objective 2: Develop methods to guide precision agricultural water management using remote-sensing, climate and soil data. Sub-objective 2.A: Develop algorithms and tools that integrate in-situ sensor and remotely sensed image data with soil and weather data to inform precision variable-rate irrigation (VRI) decisions. Sub-objective 2.B: Link multi-source remote-sensing data for detection of crop abiotic and biotic stress and estimation of crop water use using machine learning and AI techniques to support precision irrigation. Objective 3: Build better field- to farm-scale decision support datasets, tools, and models for stakeholders in water-limited regions to optimize water use, profitability, and sustainability.

Approach
Urban demand for water will increase ca. 80% over the next 30 years, independent of climate change (Florke et al. 2018). Considering the combined effects of urban demand and the changing climate, we can expect an increase in the needs for agricultural water and a decrease in the supply of agricultural water over the next several decades, resulting in decreased food security world-wide (Wallace 2000, Harmel et al. 2020, Hasegawa et al. 2020, Qin et al. 2021). There is therefore an urgent need to make crop species and agricultural practices more water efficient in the face of these challenges. The research proposed herein addresses key knowledge gaps and confronts these challenges with a multifaceted approach. Specifically, we aim to improve scientific understanding of which crop traits should be targeted to increase crop water productivity (crop production per unit water) and nitrogen use efficiency under limited water (Objectives 1.A, 1.B, & 1.C). This will be achieved through a truly broad multidisciplinary approach combining plant physiology, genetics, soil biogeochemistry, and process modeling. In parallel, we will develop novel irrigation scheduling techniques that will leverage newly emerging technologies (i.e., plant stress sensing, proximal sensing, airborne remote sensing, precision agriculture, machine learning) to improve the spatial and temporal application of both water and nitrogen (Objectives 2.A & 2.B). Lastly, these plant, soil, and irrigation data streams will be woven together to build new decision support datasets, tools, and models for stakeholders in water-limited regions (Objective 3).

Progress Report
Objective 1a: We discovered that Japanese millet (Echinochloa esculenta), as well as a very close relative of this species (Echinochloa crus-galli), are both capable of quickly and completely reversing (“refilling”) what is known to be the mechanism by which drought “kills” – the formation of gas (embolism) in the water-conducting xylem elements of vascular plants. Discovery in crus-galli was the first report of its kind (published in the high-profile journal Proceedings of the National Academy of Sciences). In search of this trait, we also quantified that neither maize nor sunflower possess this trait, in agreement with results from international colleagues made on other species (wheat, grapevine, several tree species), suggesting that “refilling” may be a rare but important physiological puzzle piece conferring drought tolerance in crop plants. The discovery of this trait is not just important because it expands our knowledge of how drought “tolerance” works, but if this trait can be transferred into other plants in the same genus or more distantly related species (e.g., sorghum, maize), and this trait proves successful, the monetary benefits to agriculture could be tremendous. Additionally, there was a 50-fold range in root pressure among the 20 sorghum lines tested. The distribution of this trait among the lines will be explored to develop follow up experiments that can be used to identify the genes involved in producing this pressure. Breeding sorghum hybrids with this trait has the potential to increase farm profitability tremendously when water is limited. Objective 1b: To identify plant and soil processes that determine maize nitrogen (N) requirements under varying water availability, we completed a third season of field data collection under full and near-dryland irrigation and a range of N fertilizer rates. As in prior years, high N inputs increased grain yield under full irrigation but reduced yield and water use efficiency when water was limiting—likely due to physiological feedbacks such as stomatal closure triggered by excess root N. Research confirmed strong water and N interactions, and these findings reinforce the need to tailor N inputs to water availability and plant uptake potential. Results were shared through peer-reviewed publications and conference presentations. Objective 1c: We developed and applied phenotyping methods to identify physiological traits affecting drought tolerance and growth across diverse grass species. We used high-resolution in vivo imaging and gas exchange to track embolism dynamics in maize leaves during drought and recovery, revealing coordination among embolism formation, transpiration, and photosynthesis. Anatomical scaling patterns across grasses and a global trait analyses revealed trade-offs and constraints on hydraulic traits. A transcriptomic study of root pressure in maize identified gene networks associated with water transport under low soil moisture. In contrast, a retrospective analysis of limited-transpiration traits showed that many widely-promoted traits do not consistently improve growth or drought tolerance, highlighting the need for rigorous trait validation. Together, these studies advance fast mechanistic phenotyping approaches to identify functionally meaningful traits and caution against relying on unvalidated or context-dependent traits when selecting for drought resilience. Objective 2a: Field research at the ARS Limited Irrigation Research Farm (LIRF) continued in 2024 to develop algorithms and tools that integrate in-situ sensor, remote sensing, soil, and weather data for precision variable-rate irrigation (VRI). This third year of data collection refined irrigation scheduling treatments using multiple evapotranspiration (ET) estimation methods, including reference ET as defined in the Food and Agriculture Organization Paper 56 (FAO-56), canopy temperature, soil water balance, and a crop model integrated with remote sensing approaches. Integration of remotely sensed data with a soil-crop-water model (RZWQM2) showed improved potential for timely irrigation decisions and irrigation water productivity. Data processing pipelines initiated in prior years were further streamlined to enable near real-time decision-making. Current efforts focused on revising treatment priorities for the 2025 season based on emerging insights into crop-specific water use patterns and data quality. Supporting publications advanced ET modeling for sunflower, expanded open-source Python tools for FAO-56 ET calculations, and improved prediction of non-stressed canopy temperatures for maize under semi-arid conditions. Together, these efforts support development of scalable irrigation decision tools for use by researchers, consultants, and producers. Project updates were shared with stakeholders through presentations and ongoing collaborations. Objective 2b: In 2024, we continued efforts to integrate multi-source remote sensing and machine learning for detecting crop stress and estimating water use to support precision irrigation. Building on two prior years of spectral and unmanned aerial vehicle (UAV) data collection in maize, we focused this season on data analysis and model refinement. Time-series UAV imagery—multispectral, thermal, and red-green-blue—was used to predict maize yield and assess stress under variable nitrogen (N) and water availability. Models developed using thermal data remained especially effective under water-limited conditions. Fiscal Year 2025 analysis emphasized optimizing model performance and scaling predictions from leaf- to canopy-level. Supporting studies demonstrated accurate estimation of chlorophyll contents in leaves using UAV multispectral data across maize growth stages and irrigation levels, and improved chlorophyll estimation in wheat across genotypes and stress conditions. Additional work compared the relative value of data quality versus model complexity for yield prediction, showing that data quality remains a primary constraint. These findings reinforce the value of thermal and multispectral UAV data in detecting crop stress and estimating yield. Results support development of scalable AI tools for precision irrigation and were shared in peer-reviewed publications and stakeholder meetings. Objective 3: In 2025, we advanced the development of field- to farm-scale decision support tools to help stakeholders in water-limited regions optimize irrigation, profitability, and sustainability. We analyzed evapotranspiration (ET)-based irrigation results from the RZWQM2 model in comparison with field-applied irrigation methods and continued refinement of the model, including modifications to improve simulation of leaf temperature—an important variable for linking crop water use and stress responses. We also worked to organize and standardize diverse datasets to support broader model application and integration. Using the Unified Plant Growth Model (UPGM), we developed updated regional planting date guidance across the Central Great Plains, improving assessments of production risk and seasonality across varied climate scenarios. Supporting studies highlighted the importance of topographic variability in yield outcomes, limitations in climate forcing datasets for hydrological modeling, and long-term changes in soil conditions across the Great Plains. Together, these efforts improve simulation accuracy and model usability, strengthening the foundation for precision water management at both field and regional scales. Project results were shared through peer-reviewed publications and collaborative stakeholder outreach.

Accomplishments
1. Remote sensing and machine learning improve irrigation management in maize. By integrating data from spaceborne, airborne, and proximal sensing platforms with both traditional and deep learning models, an ARS researcher in Fort Collins, Colorado, in collaboration with partners at Colorado State University, Oklahoma State University, and University of Texas, Arlington, developed scalable approaches for monitoring crop water use, predicting crop yield, and assessing crop stress across diverse irrigation scenarios. The research emphasized the value of high-resolution imagery and multisource data fusion in enhancing the accuracy and timeliness of critical crop indicators such as evapotranspiration, chlorophyll content, and canopy structure. Additionally, the development of user-friendly tools, including a web-based application for calculating the Crop Water Stress Index (CWSI), demonstrated practical pathways for translating scientific advancements into decision support systems for farmers and irrigation advisors.

2. Unified evapotranspiration language to enhance communication among water users, researchers, and stakeholders. ARS researchers contributed to national and international efforts to clarify and standardize evapotranspiration (ET) terminology, addressing a longstanding barrier to effective water management in agriculture and water resources. An ARS scientist in Fort Collins, Colorado, coordinated a team of 26 ET experts (including 4 ARS scientists across the country) to develop a comprehensive technical note to unify ET communication across disciplines. This collaborative document, developed with input from dozens of stakeholder groups and supported by more than 50 organizations worldwide, provides clear guidance on key ET concepts and calculation methods. By promoting consistent terminology and approaches, this work reduces confusion among researchers, practitioners, and policymakers, improving communication and facilitating the adoption of standardized ET methods and terminologies. These advances directly support the development of robust, transferable irrigation scheduling tools and decision support systems, which are essential for optimizing water use efficiency and sustaining agricultural productivity under resource constraints and variability.

3. Heavy nitrogen application decreases corn grain yield when water is limited. As we grapple with optimizing crop productivity where water is limiting, open questions remain on how to adjust nitrogen fertilizer, especially as the cost of fertilizer increases. ARS researchers in Fort Collins, Colorado, in collaboration with Colorado State University partners, found maximum yields with 37% less nitrogen fertilizer when water was limited. Many farmers, fearing the yield reduction that reduced water might cause to their crops, still apply high nitrogen rates in the hope that the extra fertilizer will partially offset the problems caused by the decreased water levels or apply extra fertilizer as insurance, in case of extra rain. In multiple experiments across multiple sites and years, scientists observed that even a little extra nitrogen hurts crop yield when water is limited and leaves behind a pool of nitrogen that will move into the environment. This is a double negative for the producer, who is paying for nitrogen, and as those costs go up, paying for an input that is hurting their bottom line. This research provides valuable considerations and critical information for policy formulation (e.g., Natural Resources Conservation Service Farm Bill programs) and management guidelines (e.g., State Extension programs and soil health non-governmental organizations).

4. Wild Millet relative “resurrects” itself after drought. ARS scientists in Fort Collins, Colorado, discovered a millet relative (Echinocholoa sp.) capable of reversing what is believed to be the primary cause of plant death during drought – embolism formation in the water-conducting tissues. This is the first direct evidence of complete and functional stem xylem “refilling” following severe drought stress. This breakthrough challenges long-standing assumptions about plant recovery after drought and has significant implications for crop resilience. This is an important discovery because if this trait can be transferred to maize, wheat, rice, or domesticated millet, this could greatly reduce the current economic impact of drought on agriculture, currently estimated at 9 billion dollars per year.

U.S. DEPARTMENT OF AGRICULTURE

Water Management and Systems Research: Fort Collins, CO