Duration: 10/01/2026 to 09/30/2031

Administrative Advisor(s):

NIFA Reps:

Non-Technical Summary

As the world population approaches 10 billion by 2050, agriculture must find sustainable ways to produce crops while using fewer resources, protecting ecosystems, and coping with increasing weather variability. To accomplish this, we need crops that produce more food, provide better nutrition, and withstand stress from abiotic and biotic factors. Although advances in genetic approaches have rapidly expanded, our ability to measure plant traits or phenotypes in the real world has not kept up. This gap—known as the “phenotyping bottleneck”—limits our progress in understanding plant adaptations and crop improvement under changing environmental conditions.

Advances in sensing— including imaging systems, drones, and automated platforms—allow rapid and accurate assessment of plant growth, stress responses, and resource use efficiency. However, these technologies require common standards, shared datasets, and better approaches (e.g. AI/ML, modeling) to interpret large volumes of generated data. The NC1212 project brings together plant scientists, engineers, and data experts to develop shared systems and create predictive tools that link sensor measurements to plant performance in major and underutilized crops. The NC1212 project activities will support diverse farming systems and train the next generation of researchers. The major objectives are to: (1) Advance sensor technologies and data acquisition systems for plant phenotyping in controlled and field environments; (2) Link phenotype to function through integrative, multiscale analyses; (3) Develop and apply artificial intelligence/machine learning approaches for decision support and predictive modeling of phenotypes; and (4) Foster education, training, and outreach in plant phenomics and data science.

Statement of Issues and Justification

Global agriculture faces an unprecedented challenge to sustainably feed, clothe, and fuel a projected population of nearly 10 billion people by 2050 while reducing inputs, protecting ecosystems, and adapting to weather variability. To achieve this, we must accelerate the development of crops that deliver higher yields, improved nutritional value, and enhanced resilience to environmental stressors under increasingly constrained resource conditions. Addressing these challenges requires an integrated understanding of how plant genotypes interact with their environments and management practices—knowledge that can only be gained through precise and scalable measurement of plant traits across biological, spatial, and temporal scales.

Over the past two decades, plant genomics has undergone a revolution. Advances in next-generation sequencing technologies have made it possible to rapidly and affordably characterize the genetic makeup of thousands of individuals across hundreds of species. Yet, despite the massive expansion in genomic data, the capacity to accurately and efficiently measure phenotypes—the observable outcomes of genotype × environment × management (G×E×M) interactions—has lagged far behind. This “phenotyping bottleneck” remains one of the greatest limitations to progress in crop improvement, particularly under stressful environmental conditions where dynamic responses over time are key determinants of performance.

The Phenotyping Bottleneck and the Need for Integration

Traditional plant phenotyping approaches rely heavily on manual, labor-intensive measurements, limiting both throughput and accuracy. Even in advanced breeding programs, traits are often measured at a single timepoint and under a limited range of environments. As a result, most selection decisions are based primarily on yield—a complex, low-heritability trait—rather than on physiological or morphological features that more directly reflect adaptive capacity.

The rise of high-throughput phenotyping (HTP) and plant phenomics offers a transformative opportunity to overcome this bottleneck. Advances in imaging, spectroscopy, robotics, and automation now enable continuous, non-destructive measurement of plant traits at scale. These tools allow researchers to capture complex, dynamic processes—such as canopy development, water and nutrient use efficiency, and stress response—in real time. However, fully realizing the potential of these technologies requires coordinated efforts in data acquisition, standardization, analysis, and interpretation across research sites, crop species, and disciplines.

Convergence of Technologies: Sensing, Modeling, and AI

The next frontier in plant phenomics is the integration of advanced sensing with artificial intelligence (AI)/machine learning (ML) and data-driven modeling. These computational approaches can extract biological meaning from vast, multidimensional datasets, enabling predictive frameworks that link spectral, morphological, physiological, and molecular traits to yield and resilience outcomes. Yet, the field faces several challenges:

• Heterogeneity in data collection protocols and calibration across sensors and platforms;
• Limited interoperability and metadata standards, impeding data sharing and reuse;
• Gaps in translating sensor-based and computational outputs into biological understanding; and
• A shortage inpipeline of researchers and young talent trained to work effectively at the intersection of plant science, engineering, and data science.

To address these challenges, NC1212 provides a coordinated regional and national platform for developing standardized, scalable, and interpretable phenomics pipelines. These pipelines will link trait measurement to underlying mechanisms (“phenotype to function”) and apply predictive modeling for breeding and management decision-making.

Expanding Impact: From Major Crops to Minor and Underutilized Species

While most phenomics research has focused on major food crops such as maize, soybean, wheat, and rice, expanding these methods to minor and underutilized crops—including small grains, specialty crops, and regionally important and native species—is critical for agricultural diversification and efficient production systems. Such crops often lack the genomic and phenotypic resources available for widely cultivated species but play vital roles in nutrition security, local economies, and ecosystem services. By adapting and validating phenotyping platforms across diverse species and growth habits, NC1212 will broaden the reach and impact of phenomics technologies, supporting the USDA’s strategic goals for sustainable intensification and equitable agricultural innovation.

The NC1212 Approach: Collaborative, Multidisciplinary, and Applied

The NC1212 project brings together plant biologists, breeders, engineers, computer scientists, and data analysts to collaboratively advance the field of plant phenomics. Building on a history of successful cooperation among member institutions, the project will focus on four integrated objectives: (1) developing and standardizing sensor and data systems; (2) linking biophysical traits or phenotypes to physiological function; (3) applying AI/ML for predictive modeling of phenotypes; and (4) enhancing plant phenomics education and training.

Each objective is supported by tangible, group-based research initiatives:

• The NC1212 Sensor and Data Standards Consortium to harmonize data acquisition protocols across field and controlled environments;
• The Phenotype-to-Function Network Pilot to coordinate cross-site trials linking sensor data with physiological and omics measurements in both major and minor crops;
• The AI Challenge Dataset for Phenomics to generate open, standardized datasets for training and benchmarking predictive models; and
• The Phenomics Course Series to expand interdisciplinary training and outreach.

These coordinated efforts will create a foundation for reproducible, scalable, and interpretable plant phenomics research that accelerates discovery, translation, and application across crops, environments, and institutions.

Expected Impact and Alignment with USDA Priorities

The outcomes of this project will directly contribute to USDA and NIFA priorities for sustainable agriculture, adaptation to changing environmental conditions, and biosecurity. Specifically, NC1212 will:

• Advance understanding of G×E×M interactions to improve crop resilience and productivity;
• Provide validated, transferable tools for data-driven breeding and management;
• Enhance workforce development through cross-disciplinary education and training;
• Expand the scope of phenomics to include minor and specialty crops; and
• Strengthen data infrastructure for long-term, multi-site collaboration.

By integrating advanced technologies, fostering training and outreach, and coordinating multi-institutional research, NC1212 will enable a new era of predictive, systems-level plant science—linking sensors to function, and data to decision-making for a more sustainable and resilient agricultural future.

Expected participating states: NE, MI, IN, MN, WA, SD, DE, NC, KY, AZ, WY, MO, LA

Expected non-land grant participating institutions: Danforth Plant Science Center

Related, Current and Previous Work

The NC1212 Multistate Research Project, Exploring the Plant Phenome in Controlled and Field Environments, builds upon more than a decade of coordinated regional efforts aimed at overcoming the plant phenotyping bottleneck and advancing technologies for crop improvement and understanding plant functions. The coordinated efforts across institutions began around 2016, with the establishment of the North American Plant Phenotyping Network (NAPPN) during an Inaugural Convening Event, co-hosted by Foundation for Food and Agriculture Research (FFAR) at Purdue University. In 2020, NC1212 was initiated and has served as a collaborative network linking plant scientists, breeders, engineers, and computational researchers across land-grant universities, federal research centers, and industry partners. These collaborations have enabled synergistic advances in field-based sensing, controlled-environment automation, and data analytics, now transforming plant research pipelines.

Progress and Impacts of Previous NC1212 Work

The previous iteration of NC1212 established a foundation for phenomics research through the development, validation, and dissemination of tools, platforms, and training materials. The major accomplishments include:

• Development and deployment of high-throughput field phenotyping systems, including UAV-based imaging, ground rovers, and proximal sensing platforms that integrate RGB, multispectral, hyperspectral, thermal, and LiDAR technologies.
• Advancements in image and data processing pipelines, improving data quality, calibration, and the extraction of biologically meaningful traits such as canopy temperature, vegetation cover, plant architecture, and vegetative indices.
• Integration of phenomics with genomics and breeding applications, resulting in predictive trait models that link sensor data to yield, stress tolerance, and resource-use efficiency.
• Training and workforce development, through workshops, symposia, and collaborative graduate training programs that introduced students and early-career scientists to the interdisciplinary tools of phenomics, data science, and computational biology.

These activities have catalyzed numerous publications, grant proposals, and shared datasets, demonstrating NC1212’s critical role in building community capacity and technological expertise within the region and beyond. For example, the community members are active participants of the NAPPN and International Plant Phenotyping Network (IPPN). The members have led various education and outreach (e.g. Webinar series on Remote Sensing Application In Agriculture Research at Purdue University), and traineeship (Washington State University, Michigan State University, University of Nebraska-Lincoln) programs. There have been efforts to generate and release large phenotyping datasets (University of Nebraska-Lincoln, Iowa State University, Purdue University). A detailed list of progress and impacts can be found in our annual reports. In recent years, the NC1212 community has grown stronger with increase in membership and coordinated efforts to advance phenomics.

Current Landscape and Emerging Opportunities

The pace of innovation in plant phenomics continues to accelerate, driven by advances in sensor technology, robotics, artificial intelligence, and machine learning (AI/ML). Modern phenotyping systems now generate terabytes of multimodal data per season, creating new challenges in storage, integration, and analysis but also unprecedented opportunities for biological discovery. Multi-sensor fusion and predictive modeling approaches have shown promise for improving yield prediction, nitrogen-use efficiency, and stress resilience across environments. At the same time, controlled-environment facilities allow precise manipulation of growth conditions and serve as testbeds for functional genomics and physiological studies.

However, realizing the full potential of these technologies requires cross-disciplinary collaboration and standardization—areas where NC1212 continues to play a leading role. By linking existing expertise and infrastructure across member institutions, NC1212 provides a framework for coordinated research that spans scales (leaf to canopy), environments (controlled to field), and species (major to minor crops).

Relationship with other multistate projects

There are several communities that members of this community interact with. These include NC1210 (Frontiers in on-farm experimentation), NCERA180 (Precision agriculture technologies for food, fiber, and energy production), W4009 (Integrated systems research and development in automation and sensors for sustainability of specialty crops), S1069 (Research and extension for unmanned aircraft systems (UAS) applications in U.S. agriculture and natural resources), S1090 (AI in Agroecosystems: Big data and smart technology-driven sustainable production), and NCCC170 (Research advances in agricultural statistics). NC1212 will further strengthen these interactions and knowledge across these projects will be integrated to advance phenomics.

Objectives

1. Advance sensor technologies and data acquisition systems for plant phenotyping in controlled and field environments
   Comments: Rationale: Reliable and interoperable sensor systems are critical for multi-environment phenomics. Shared calibration and data acquisition frameworks will enhance reproducibility across sites and species. Sub-objectives: 1.1. Develop, refine, and validate multisensor and imaging platforms for diverse crops and environments, including underutilized and specialty species. 1.2. Establish calibration protocols and metadata standards, including testing of new sensors, to ensure comparability and data quality across species, institutions, and environments. 1.3. Develop and disseminate best practices for metadata and data management following FAIR principles to enable data sharing and interoperability, utilizing collaborative infrastructure. 1.4. Develop open-source software and hardware interfaces for automated and high-throughput data collection. Expected outcomes: - Standardized calibration and acquisition protocols across multiple crops and environments. - Cross-platform data repositories and benchmark datasets. - Improved interoperability and reproducibility across major and minor crops.
2. Link phenotype to function through integrative, multiscale analyses
   Comments: Rationale: Understanding how measured phenotypes are related to plant processes (e.g. physiological, phenological) and plant functions (e.g. water use efficiency, nutrient use efficiency, resistance mechanisms, plant mechanics). Linking these phenotypes to plant functions is central to predicting and improving crop performance. Sub-objectives: 2.1. Integrate phenomics data with functional and environmental datasets, and other ‘omics data to uncover mechanisms of productivity, stress tolerance, and nutrient use efficiency. 2.2. Develop computational frameworks connecting phenotypes across temporal and spatial scales that lead to emergent plant functions. 2.3. Identify key measurable phenotypes that serve as proxies for functional processes in row and horticultural crops. Expected outcomes: - Integrated datasets linking phenotypes to underlying functions. - Mechanistic understanding of plant performance and resilience. - Validated trait-based predictors applicable across crop species.
3. Develop and apply artificial intelligence/machine learning (AI/ML) approaches for decision support and predictive modeling of phenotypes
   Comments: Rationale: AI/ML approaches enable efficient extraction of knowledge from complex phenomics datasets, facilitating prediction, discovery, and translation to breeding and management. Sub-objectives: 3.1. Apply and compare ML methods (e.g., convolutional neural networks, ensemble models, transformers) for predicting plant traits and performance. 3.2. Develop explainable and transferable AI frameworks to enhance interpretability and link data-driven models to biological insight. 3.3. Create shared, open-source tools and benchmark datasets for AI-based phenomics data analysis and visualization. 3.4. Explainable AI with data analytics and narrative visualization for transparent crop recommendation under dwindling natural resources and weather variability. Expected outcomes: - Predictive models for trait estimation, stress response, and yield. - AI tools and datasets accessible to the broader phenomics community. - Enhanced understanding of how machine learning outputs map to biological mechanisms.
4. Foster education, training, and outreach in plant phenomics and data science
   Comments: Rationale: Cross-disciplinary training is critical for building the next generation of scientists and professionals in phenomics and data-intensive agriculture. Sub-objectives: 4.1. Create modular, open-access training materials that equip different audiences with core skills in sensing, data management, and modeling. 4.2. Organize exchange programs across institutions to enhance cross-exposure and training for plant scientists, engineers, and data scientists. 4.3. Expand outreach and engagement with schools and other stakeholders in the field, to increase awareness in the neighboring associated communities. Expected outcomes: - Network-wide training programs and resource-sharing. - Increased data literacy and technical proficiency. - Stronger interdisciplinary and inter-institutional collaboration.

Methods

Objective 1: Advance sensor technologies and data acquisition systems for plant phenotyping in controlled and field environments

Sub-objective 1.1: Develop, refine, and validate multisensor and imaging platforms across more crops and environments, including underutilized and specialty species.

To address the need for robust multisensor platforms, we will engineer and deploy modular platforms capable of hosting RGB, hyperspectral (VNIR/SWIR), and/or 3D sensors simultaneously which can be benchmarked against existing infrastructure or sensors (Sankaran et al., 2015; Bao and Tang, 2016; Pauli et al., 2016; Bucksch et al., 2017; Masjedi et al., 2018; Bai et al., 2019; Bao et al., 2021; Kudenov et al., 2023, 2025). We will test these sensors across varying deployment vehicles, including high-throughput conveyor-based imaging systems for controlled environments at Core Facilities (e.g. University of Nebraska-Lincoln and/or Donald Danforth Plant Science Center) and unattended ground vehicles (UGVs), aerial vehicles (UAVs), or manual measurements for field environments. The platforms can also involve simple Raspberry Pi-based Internet of Things (IoT) systems (Sangjan et al., 2021; Chamara et al., 2023) or sophisticated fixed field infrastructure (e.g. Lemnatec Field Scanalyzer; Spidercam systems (Kirchgessner et al., 2017; Virlet et al., 2017; Burnette et al., 2018; Bai et al., 2019) capable of automated sampling at high spatial and temporal resolution. A key focus will be validating sensor co-registration and stability across diverse canopy architectures, ranging from row crops such as maize, soybean, and peanuts to structurally complex specialty crops such as grapes, strawberries, and cucurbits, as well as underground crops such as potato, sugar beets, carrots, and sweetpotato. We will utilize 3D printed, open-hardware mounts to minimize vibration and ensure consistent sensor geometry, which allows for the rigorous testing of sensor-to-sensor alignment.

Data validation will prioritize the assessment of decision-support capability, balanced with information content and a sensors’ cost (e.g., both at time-of-purchase and long-term data management) rather than sensor precision or raw sensor specifications (e.g., as based on the raw number of spectral channels or a focus on one specific capability over others – such as fluorescence versus hyperspectral). This will form a more generalized, interdisciplinary “taxonomy” of engineering tradeoffs spanning cost, data storage, spatial resolution, temporal resolution, spectral resolution, data transfer, data sharing, calibration (Reynolds et al., 2019; Kim, 2020) that are defined relative to information content, based on decision-making quality metrics like classification accuracy, year-over-year generalizability, fundamental knowledge generation, hypothesis testing, and statistical significance (Kuska and Mahlein, 2018; Zhi et al., 2025).

We will also work to develop optical simulation pipelines (Jacquemoud and Baret, 1990; Feret et al., 2008; Jacquemoud et al., 2009; Luo and Kudenov, 2016; Bailey, 2019; Krafft et al., 2024; Li and Xiang, 2024) to enable more robust data augmentation of optical sensor data when used for decision-support models. Sensor data modeling and synthesis efforts will be validated on how augmentation impacts “yearly splitting” of data, during model cross-validation in Objective 3. By training models on data from prior years, with testing on subsequent years, we can evaluate whether the high-dimensional data from hyperspectral sensors, or simulation-based data augmentation methods, offers a statistically significant improvement in model transferability and generalizability compared to more conventional (and accessible) RGB, RGB+Time of Flight (ToF) data, or training data-heavy pipelines. This type of testing will identify scenarios where the logistical burden of higher-dimensional data, which is complex to store and interpret, is unjustified because a simpler, broadly accessible sensor yields comparable decision-making accuracy - especially given deep learning’s ability to better leverage spatial (plant structural, textural, or environmental context) information (Su, 2020; Zhang et al., 2022). Important data collection considerations will also be defined, connecting the sensor platform and its deployment to down-stream modeling requirements. The number of replications, images, data storage, image format (raw or compressed), and their tradeoffs will be generalized. Post-processing protocols and example pipelines, needed for converting sensor data into calibrated outputs, will also be curated, where available from existing or ongoing studies, defined, and distributed from a common open-access platform.

Sub-objective 1.2: Establish calibration protocols and metadata standards, including testing of new sensors, to ensure comparability and data quality across species, institutions, and environments.

We will establish a network of “phenomics calibration sites” modeled after the vicarious calibration campaigns used for validating satellite radiometric calibrations (Thome et al., 1994; Thome et al., 2004). Just as desert playas were historically used to radiometrically calibrate satellite sensors (Slater, 1985; Slater et al., 1987; Teillet et al., 1990), we will utilize our highly instrumented crop fields as biological reference standards. We will coordinate ground-based data acquisition to coincide precisely with satellite overpasses from e.g. Landsat or TERRA’s MODIS sensors. This simultaneous collection allows us to use high-resolution ground data to validate the atmospheric correction and spectral fidelity of orbital platforms. We will develop standard operating procedures for these overpass campaigns that utilize downwelling irradiance sensors and large-format reference panels (Mishra, 2021; Swaminathan et al., 2024; Wang et al., 2024; Zhu et al., 2024) and existing open-source, ground-based image processing pipelines, such as PlantCV (https://plantcv.org, Gehan et al., 2025). Such efforts can be performed at different scales, from leaf to field levels. This ensures that the spectral signatures measured by our gantries and UAVs can be directly upscaled to validate the radiometric accuracy of satellite imagery used for broad-acre monitoring.

We will extend this framework to validate satellite-based decision support models through a spatial cross-validation approach. We intend to treat our distributed network of field sites as a validation grid where a single satellite-derived model, such as a soybean yield predictor or drought stress indicator, is tested across multiple distinct locations simultaneously. By benchmarking the satellite predictions against the high-fidelity "ground truth" phenotypes collected at each calibration site, we can determine the transferability of the model across different soil types and micro-climates. This process moves beyond simple sensor calibration to "decision calibration." It ensures that a remote sensing algorithm predicting crop status in one location is equally valid when applied to satellite imagery of fields in a different region. This results in a robust protocol for verifying that satellite-based information content is sufficient to render accurate management decisions at scale.

Sub-objective 1.3: Develop and disseminate best practices for metadata and data management following FAIR principles to enable data sharing and interoperability, utilizing collaborative infrastructure.

We will implement a metadata schema that extends the Minimum Information About a Plant Phenotyping Experiment (MIAPPE) standards to include granular details on calibration states (Krajewski et al., 2015; Ćwiek-Kupczyńska et al., 2016; Papoutsoglou et al., 2020). This includes recording specific sensor integration times, firmware versions, and the exact timestamps of white-reference or dark-current capture (Lobet et al., 2013). We will utilize collaborative infrastructure, such as modified BrAPI endpoints, to store raw sensor data alongside its associated calibration files and environmental metadata (Neveu et al., 2019; Selby et al., 2019). This structure will ensure that data is Findable, Accessible, Interoperable, and Reusable (FAIR), specifically enabling researchers to trace signal variances back to either biological changes or sensor configuration changes.

To facilitate the creation of robust, year-independent models, we will architect the data management system to support temporal partitioning by default (Sheridan, 2013; Roberts et al., 2017; Valavi, 2025). Datasets will be tagged and indexed by season and location to enable automated querying for Year Split validation sets. We will conduct workshops and create documentation demonstrating how to utilize these metadata tags to filter training sets from past years and test sets from the current year. This methodology ensures that shared repositories are not just data dumps, but structured archives that support the development of algorithms capable of handling the reality of historical versus future data application.

Sub-objective 1.4: Develop open-source software and hardware interfaces for automated and high-throughput data collection.

We will develop open-source, deployable sensor nodes that function as turn-key ground truth stations that can support the expansion of phenotyping calibration networks. Utilizing the accessible hardware identified in Sub-objective 1.1, such as high-resolution RGB and ToF modules, we will design low-cost, solar-powered units capable of autonomous data collection (Kassim, 2020; Saqib et al., 2020; Shahab et al., 2025). These units will be controlled by readily available microcontrollers, programmed to synchronize data capture with satellite overpasses, directly adhering to the protocols established in Sub-objective 1.2. By lowering the hardware cost and technical complexity of these stations, we enable the rapid deployment of additional ground calibration sites across diverse geographies, transforming a few isolated research stations into a distributed validation grid (Soussi et al., 2024). Additionally, it will better enable the ground truth stations to be used in individual high-resolution temporal and spatial phenotyping activities that can directly support objectives 2 and 3, as well as scaling extension activities to growers and producers with the developed decision support tooling (Pierpaoli et al., 2013; Lindblom et al., 2017).

The software deliverables will include a suite of edge-computing pipelines that automate the ingestion and pre-processing of data from these distributed nodes, focusing on using inexpensive low data-rate (500 megabyte) and long-term (10 year) cellular SIM cards specifically tailored for internet of things (IoT) off-grid usage (Lindblom et al., 2017; Friha et al., 2021). We will utilize containerized applications, such as Docker images, that come pre-configured with computer vision algorithms to detect in-scene calibration targets and perform radiometric corrections locally. These pipelines will include built-in validation modules that utilize our prior year split methodologies to benchmark the low-cost station data against high-fidelity reference standards. This ensures that as we expand the network using inexpensive tools, the data remains rigorous enough to serve as a valid ground truth for satellite-based decision support models.

Potential Group Research Project:
NC1212 Sensor and Data Standards Consortium – A coordinated effort among member institutions to standardize calibration targets and reference datasets for shared use across field and controlled-environment platforms, which lead to an agreed-upon sensor calibration process across facilities.

 

Objective 2: Link phenotype to function through integrative, multiscale analyses

Sub-objective 2.1: Integrate phenomics data with functional and environmental datasets and other ‘omics data to uncover mechanisms of productivity, stress tolerance, and nutrient use efficiency.

Datasets will be acquired on plant functional traits, which include photosynthesis and nutrient/water use efficiency, among others. Plant phenotypes will be acquired using ground-based robotics, drone imaging, static imaging and/or multi-spectral sensors. Sensing platforms developed in Objective 1 will be integrated as appropriate. All growth locations will be equipped with environmental monitoring equipment to collect weather data. Where applicable, other ‘omics data such as transcriptomics, proteomics or genomics data, will be collected. The types of data collected will be specific to different research questions but rely on a central framework of analysis that will be developed through this project.

A central framework and standard practices for integrating phenotypic, functional, environmental and ‘omics data will be developed. This framework will include spatiotemporal information for an integrated whole-plant approach.

Sub-objective 2.2: Develop computational frameworks connecting phenotypes across temporal and spatial scales that lead to emergent plant functions.

Process-based models such as crop growth (e.g. BioCro, Lochocki et al., 2022) or cropping system (e.g. DSSAT, Jones et al., 2003; APSIM, Keating et al., 2003) models that couple physiological, phenological, and canopy processes will be used alongside statistical and machine-learning approaches to capture trait dynamics across scales. These models will translate molecular and organ-level information into canopy functions (such as light use efficiency, water use efficiency, nutrient use efficiency) and whole-plant performance, providing a unified framework for identifying emergent functional outcomes.

Sub-objective 2.3: Identify key measurable phenotypes that serve as proxies for functional processes in row and horticultural crops.

The goal of this sub-objective is to identify key measurable phenotypes that serve as proxies for the designated functional outcomes, thus leveraging plant phenotypes to predict plant functions. Cross-scale computational tools will be used/developed to link phenotypes from molecular to canopy and ecosystem levels. Organ- and plot-level traits will be integrated with imaging-derived canopy traits and environmental time-series data.

Potential Group Research Project:
Phenotype-to-Function Network Pilot – Shared field and controlled-environment trials across NC1212 sites evaluating selected phenotypes and functions in row and horticultural crops using a standardized framework.

 

Objective 3: Develop and apply artificial intelligence/machine learning (AI/ML) approaches for decision support and predictive modeling of phenotypes

Objective 3.1: Apply and compare ML methods (e.g., convolutional neural networks, ensemble models, transformers) for predicting plant traits and performance.

Plants are living, growing organisms that continually bifurcate into new organs, adding complexity to their shape and architecture over time. Thus, analyzing traits over an extended period provides more crucial and insightful information about a plant’s vigor than traits extracted at a single time point in its life cycle. We will explore advanced computer vision techniques to extract morphological phenotypes from multimodal image sequences, and then apply deep learning, data analytics, and visualization techniques to analyze the temporal variation of phenotypes regulated by genotypes. We will develop novel methods for predicting plants’ traits based on environmental, soil, and weather characteristics using deep neural-network-based time-series modeling. 

Objective 3.2: Develop explainable and transferable AI frameworks to enhance interpretability and link data-driven models to biological insight.

Explainable AI (XAI) is a field of AI that uses a set of tools, techniques, and algorithms to provide interpretable, intuitive, trustworthy, and human-understandable explanations of AI decisions (Arrieta et al., 2000). SHAP (SHapley Additive exPlanations) is a popular XAI technique that assigns a contribution value to each input feature for a particular prediction. Phenomics leverages data-driven insights to optimize crop selection and resource utilization, improving agricultural sustainability and productivity. The application of XAI in phenomics is in its infancy and demands research attention for exploration. The proposed research will explore XAI techniques to link data-driven predictive and prescriptive models to biological insight for trustworthy farming decisions.

Objective 3.3: Explainable AI with data analytics and narrative visualization for transparent crop recommendation under dwindling natural resources and weather variability.

The proposed research will explore a data-driven crop recommendation system that integrates environmental and soil data to identify crops that share similar growth characteristics resulting in improved yield for a specific region. We will apply unsupervised clustering techniques such as k-means, hierarchical clustering, and density-based spatial clustering of applications with noise (DBSCAN) to group crops based on similar resource requirements, aiding in the identification of optimal crop clusters for varying environmental conditions. To enhance interpretability, we aim to integrate XAI techniques, specifically feature importance analysis, to provide transparent insights into how environmental and soil factors influence crop groupings. Furthermore, this research emphasizes the importance of data storytelling, where the results of clustering are translated into a coherent, human-centered narrative. This narrative helps farmers and stakeholders understand not only the data but also the underlying principles that drive the crop recommendations. By visualizing and narrating the relationships between soil and environmental factors, data storytelling bridges the gap between complex AI models and actionable insights. By combining clustering techniques, XAI, and data storytelling, this research advances the state-of-the-art in phenomics, providing a robust, transparent, and actionable crop recommendation framework.

Objective 3.4: Create shared, open-source tools and benchmark datasets for AI-based phenomics data analysis and visualization.

With advancements in imaging technologies and data acquisition systems, phenotyping has evolved into a data-intensive field, producing multimodal, multi-view, time-series complex datasets that require the power of computer vision, artificial intelligence, data analytics, and visualization techniques to extract meaningful insights, which are key to improving crop resilience, yield, and adaptability. To develop and evaluate new algorithms and perform uniform comparisons with the competing state-of-the-art methods, public availability of benchmark datasets adhering to FAIR principles (Sub-objective 1.3) is crucial (Wilkinson et al. 2016). High-throughput phenotyping systems (e.g. University of Nebraska Lincoln, Donald Danforth Plant Science Center) can uniquely enhance our understanding of plants’ life processes by capturing plants’ imagery at different life stages and in various modalities to generate massive datasets. Our previous works released several benchmark datasets, but they are small and targeted, e.g., at University of Nebraska-Lincoln - Component Plant Phenotyping for morphological analysis of maize structures (Das Choudhury et al. 2018), FlowerPheno for flower phenotyping (Das Choudhury et al. 2022) and 3D plant phenotyping (Das Choudhury et al. 2020) datasets, which typically involve a single image modality. To promote AI research in agriculture, University of Nebraska-Lincoln has recently developed a large-scale, open-source benchmark dataset called University of Nebraska-Lincoln ReproPheno (UNL-ReproPheno) dataset. This first-of-its-kind dataset, comprising multimodal, multi-view, and temporal images of plants maturing into developing flowers and fruits, will be publicly released with the intention of advancing reproductive-stage phenotyping research utilizing computer vision and artificial intelligence techniques. The project will further support such efforts, to generate multimodal, multi-view, temporal benchmark datasets, across environments and species.

Potential Group Research Project:
 AI Challenge Dataset for Phenomics – A collaborative NC1212 initiative to create an open, curated dataset across sites and crops (including minor species) for benchmarking and community development of AI models for trait prediction.

 

Objective 4: Foster education, training, and outreach in plant phenomics and data science

Objective 4.1: Create modular, open-access training materials that equip different audiences with core skills in sensing, data management, and modeling.

Across the NC1212 community, there is growing recognition that phenomics is advancing faster than the training available to students and early-career researchers. Many of the tools that now shape the field—new sensor systems, automated measurement platforms, large-scale data workflows, and AI-based analytical methods—require skills that are not taught consistently across institutions. Although several universities offer courses related to phenomics (e.g., Phenotyping for controlled environments – University of Wyoming, Eembedded systems – Noth Carolina State University, UAS applications in agriculture – South Dakota State University, Washington State University, Machine learning – Multiple institutions), these efforts remain scattered and vary widely in both scope and availability. A coordinated approach to training has now become essential. The community has been discussing the development of shared, open-access instructional materials that any instructor, student, or industry trainee can use. Open-access resources help level the playing field for institutions with limited course offerings or personnel, and they reduce redundant work among faculty who are independently building similar content, avoiding the squandering of resources. Modular training materials—ranging from introductory concepts to more advanced technical modules—would make it easier for programs to incorporate phenomics into their existing curricula or into focus summer or winter sessions, while giving learners clearer and more consistent entry points into the field. Beyond flexibility, a modular approach—both in the course architecture and in the training resources supporting it—also enables tailored configurations for different crops, environments, and research questions that can vary across institutions and audiences. In the classroom, this means instructors can assemble course pathways that speak to their students’ backgrounds: engineering students may engage more deeply with hardware and embedded systems, while plant science students may focus on experimental design, data interpretation, or modeling. The same principle ensures that the NC1212 offer can support diverse users and science drivers while still maintaining standardized workflows and comparability across sites. This emphasis on modularity is directly aligned with the technical philosophy driving our approach for Objective 4. Modularity is critical to overcoming persistent bottlenecks in phenotyping and agricultural resilience. By designing our effort as a system in which instruments, software, data workflows, and AI tools can be independently developed, updated, or replaced, each component can evolve without disrupting the overall cyber-physical platform presented. This flexibility allows rapid integration of new and future sensors, automation strategies, or analytical methods as they emerge across all participating institutions, helping address long-standing limitations in throughput, reproducibility, and data interoperability while enhancing education. Similarly, modular training materials will allow instructors to update content as the technology evolves, rather than rebuild entire courses from scratch.

Open-access resources also play a critical role in workforce development. As phenomics tools become more widely adopted across research and industry, there is an increasing demand for people who can collect, manage, and interpret complex datasets. Making these materials accessible to students, extension specialists, and industry partners—regardless of geographic location or institutional resources—will broaden participation and prepare a more technically capable workforce. Early efforts in this direction are already underway at the University of Wyoming that has developed the first interdisciplinary class in Controlled Environment Agriculture (CEA) as a deliverable of the Wyoming Innovation Partnership (WIP) program. WIP has been instrumental to throw the basis for a novel workforce development where students get to be exposed and interact with the actual stakeholder in the field. The existing class (CEA principles and applications), brings together faculty and business experts from plant science and production to engineering and phenotyping. This effort provides a useful starting point for expanding this initiative across the NC1212 network and expanding its focus to phenotyping and phenotyping for CEA. Currently, both the NC1212 community and the IPPN members have shown interest and have been discussing cross-institutional online education and building an online cross-state / international modular graduate program. Overall, combining modular cyber-physical platforms across the NC1212 institutions with modular, open-access training materials proposed here, creates a coherent and sustainable strategy for advancing phenomics research, enhancing agricultural resilience, and ensuring that the next generation of scientists is prepared to effectively use and innovate on these rapidly evolving tools.

Objective 4.2: Organize exchange programs across institutions to enhance cross-exposure and training for linking plant scientists, engineers, and data scientists.

A major bottleneck in advancing phenomics and next-generation agricultural research is the persistent separation of expertise across disciplines. Plant scientists often lack exposure to sensor design, robotics, or advanced analytics, while engineers and data scientists may have limited opportunities to engage with biological questions, experimental constraints, or agricultural systems. To bridge these divides, the NC1212 team proposes establishing structured exchange programs that connect students and early-career researchers across institutions and disciplines, creating intentional opportunities for cross-exposure and collaborative skill development. These exchange programs will involve students and trainees from plant science, horticulture, pathology, agricultural engineering, electrical engineering, computer science, mathematics, statistics, and related fields. By rotating participants across laboratories, research stations, and phenotyping platforms, the program will allow students to experience firsthand how biological questions, instrumentation, and computational approaches intersect. This exposure is essential for cultivating a workforce that is capable not only of operating phenomics technologies but also of designing, refining, and applying them to meaningful scientific and agricultural challenges. Where possible, the initiative will leverage existing federal training mechanisms—including NSF Research Experiences for Undergraduates (REU), USDA NIFA Research and Extension Experiences for Undergraduates (REEU), Research Innovation and Development Grants in Economics (RIDGE), and other relevant fellowship or internship structures. These programs provide a flexible framework for supporting student mobility, mentoring networks, and hands-on research experiences. However, the proposed exchange effort expands beyond traditional summer programs by fostering year-round coordination among faculty, shared research objectives, and community-driven curriculum development. The goals of the proposed exchange programs are threefold. First, they will create genuine interdisciplinary training pathways by embedding trainees in environments where they can learn complementary skill sets, such as plant imaging, sensor calibration, advanced data management, and machine-learning-driven phenotypic analysis. Second, the programs will strengthen professional networks across institutions, enabling students to build long-term collaborations and connections that persist well beyond their exchange period. Third, they will accelerate the adoption of phenomics approaches in both research and industry by preparing a diverse pool of scientists and engineers who can navigate the biological, computational, and technological dimensions of the field. Importantly, these exchanges will also support the broader goals of community-building and capacity development. By allowing students from smaller or under-resourced institutions to work with advanced phenotyping platforms and specialized instrumentation, the programs reduce inequities in training opportunities and broaden participation in emerging agricultural technologies. Institutions hosting participants benefit as well, as visiting students bring varied perspectives, fresh ideas, and experience with alternative research models and tools.

In combination with the modular training materials and the shared cyber-physical platform described elsewhere in the proposal, these exchange programs create an integrated strategy for cultivating the next generation of phenomics researchers. Together, they will help ensure that plant scientists, engineers, and data scientists are equipped not only with technical expertise but also with the interdisciplinary fluency required to tackle complex challenges in agricultural resilience and sustainability. Current efforts include the USDA NIFA REEU project on Phenomics big data management at Washington State University supported 38 undergraduate students from 16 states and 25 different institutions across five years, in Computer Science and Electrical Engineering, Horticulture, Crop and Soil Sciences, and Biological Systems Engineering across campuses, including research and extension centers. At the University of Wyoming a newly started NSF-REU site in 2025 brought 8 students from across the United States to work on integrated methods for data science in controlled environment agriculture and it will be expanded to phenotyping methods and AI implementation for the summer of 2026. Development of such programs for undergraduate and graduate students, and early career researchers (research/postdoctoral associates) will continue. Finally, international exchanges will be introduced utilizing the strong relationship of the NC1212 with the IPPN community. These exchanges will allow for the exposure to novel, advanced phenotyping techniques and data methodologies for students and research scholars and also create exposure to more diverse production and business models.

Objective 4.3: Expand outreach and engagement with schools and other stakeholders in the field to increase awareness in the neighboring associated communities.

To ensure that the benefits of phenomics research, education, and workforce development extend beyond academic institutions, each institution in the NC1212 team will significantly broaden outreach and engagement activities across local schools, community groups, regional stakeholders, and the broader scientific community. Increasing awareness of phenomics and its relevance to agriculture, environmental sustainability, and food systems is essential for building a broad talent pipeline and fostering long-term community support for agricultural innovation. Engaging K–12 students, teachers, and families are particularly important component of this effort. Many students today have limited exposure to plant science, engineering, or data-intensive fields and through targeted school visits, interactive demonstrations, and teacher support materials, our institutions will introduce students to sensor technologies, imaging methods, and real-world applications of plant phenotyping in ways that are accessible and engaging. In addition to K–12 outreach, we will deepen engagement with undergraduate and graduate students, early-career researchers, agricultural professionals, and industry partners.

Several strategies will be used to strengthen these connections:

1. Mentorship programs through NAPPN and other community-led initiatives.
   We will expand structured mentoring networks that pair undergraduate students with graduate mentors, and graduate students with phenomics and industry experts. This layered mentoring structure supports personalized guidance, fosters research identity formation, and helps trainees navigate academic and professional pathways.
2. Internship and experiential learning programs.
   Hands-on research internships—both during the academic year and the summer—will allow students to gain experience with sensors, data workflows, field-based measurement, and phenomics technologies. Partnerships with regional agricultural companies, plant breeders, and tech startups will further expand internship opportunities and strengthen ties between research activities and real-world agricultural applications.
3. Promoting networking opportunities at conferences and professional meetings.
   Networking is essential for retention and career development, particularly for students who may lack established professional connections. We will actively encourage and support attendance at meetings such as the NAPPN annual conference, providing structured opportunities for students to meet researchers, industry leaders, and peers working in phenomics and related fields.
4. Development of career advancement workshops.
   Workshops will address topics such as scientific communication, coding and data skills, grant writing, CV preparation, and career pathways in academia, industry, government, and non-profit sectors. These workshops will be designed to meet the needs of students at different stages and from a variety of disciplinary backgrounds.
5. Travel grants and financial support mechanisms.
   To ensure equitable access to conferences and training events, we will establish travel grants aimed particularly at students from smaller institutions, and first-generation college backgrounds. Travel support reduces financial barriers and exposes students to the broader phenomics community and emerging technologies.

Collectively, these outreach and engagement efforts will help build a more resilient phenomics community. By reaching students early, supporting their progression through higher education, and connecting them with professional networks, the program will cultivate a diverse next-generation workforce that is well-equipped to contribute to agricultural sustainability and technological innovation. These activities also strengthen community relationships and ensure that neighboring schools, families, and stakeholders can participate in—and benefit from—the scientific and technological advances emerging from this project.

Potential Group Research Project:
NC1212 Phenomics Course Series – “Portable” phenotyping course sessions and modules developed and recorded by individuals at member institutions to contribute towards plant phenomics literacy, spanning field and controlled environment phenotyping, data pipelines, and AI modules.

Measurement of Progress and Results

Outputs

• Established protocols and tools that follow FAIR principle to manage — store, process, and analyze — phenotypic data.
• Development of peer-reviewed publications (potentially highly collaborative across disciplines and/or institutions) and projects through grant opportunities.
• Open-source phenomics datasets.
• Open-source data/image software, processing algorithms, and models.
• Open-source protocols for sensor development and programming.
• Implementation of phenomics technologies and tools in breeding programs.
• Predictive models to derive agronomic, morphological and physiological responses of plants to the environment.
• Evaluation and enhancement of sensor capabilities.
• Number of undergraduate, graduate, and early career researchers trained or mentored.
• Number of participants reached through education, extension and outreach activities.

Outcomes or Projected Impacts

• Collaborative efforts to resolve gaps and challenges in the field of phenomics.
• Improved models to predict the performance of diverse germplasm in untested environments.
• The use of phenomics to empower plant biology, plant adaptation, and towards gene discovery to improve agriculture, agroecosystem, and environment.
• Phenomics driven crop improvement efforts to develop elite cultivars, which will improve farmer profitability and sustainability.
• Development of interdisciplinary next-generation workforce.
• Recognition of the field and policy changes to support further advancement of phenomics.
• Enhanced rural prosperity and economic development, and thus social stability.

Milestones

(2028):Years 1-2: Develop sensors and sensor systems with open-source hardware designs for phenomics applications. Establish collaborative projects to collect standardized datasets in one or more crops. Identify desired teaching modules to promote education and training in phenomics.

(2030):Years 2-4: Develop radiometric and calibration protocols and pipelines. Create benchmark datasets. Develop and release open-source post-processing pipelines through a shared repository. Develop modular, open-access training materials.

(2031):Years 4-5: Identify and validate key proxy phenotypes that define plant functions. Validate AI/ML approaches towards predicting crop performance and for crop recommendations. Establish mentorship and exchange programs, and workshops as a part of education, training, and outreach activities.

Projected Participation

View Appendix E: Participation

Outreach Plan

In terms of outreach, we aim to deliver phenomics science-based information through effective and efficient communication to ensure advancement of the field to promote sustainability and profitability in agriculture. Some of the key outreach activities during this project include: (1) publishing the results of our work in scientific journals, (2) presenting the results of this work at national and international conferences (e.g. NAPPN Annual Conference, International Plant Phenotyping Symposium, other disciplinary conferences/meetings), (3) presenting phenomics work in media releases, trade journals, and blogposts, (4) report phenomics work to local crop producers, representatives of local and national crop commodity organizations, and private sector partners interested in developing and commercializing new crop varieties, (5) incorporate research outcomes and tools into classroom learning, and (6) through virtual and in person training workshops.

Organization/Governance

There will be two committees: the technical committee and the executive committee. The technical committee will consist of project leaders for the contributing states, the administrative advisor, and NIFA representatives. Voting members include all the participants in this project.

In regard to the executive committee, there will be two leadership roles, a chairperson and a vice chairperson, each role to be served for a year. The roles of chair and vice chair will begin on the anniversary date of the project. The role of the chair will be to facilitate collaborative and other project activities, assist in finalizing the annual meeting (venue, dates, agenda), and prepare the annual report in a timely manner, with the assistance of the vice chair. The role of the vice chair will be to assist the chair and take meeting notes. The NC1212 will meet on the quarterly basis, with one in-person meeting annually. To be elected to the executive committee, the person should have been a member of this project for at least two years prior to being voted. The vice chair will resume the role as the chair in the following year.

The election will take place during the annual meeting. The meeting venue and potential dates will be finalized during the annual meeting. All members are eligible to vote. If a collaborative research, education or outreach activity or project is proposed and discussed within the NC1212 community, any active member of this community can take the lead, with the approval of its members. If two members want to lead a project, an anonymous vote will be set by the NC1212 chair to select the lead.

Literature Cited

Arrieta, A. B., Díaz-Rodríguez, N., Del Ser, J., Bennetot, A., Tabik, S., Barbado, A., ... & Herrera, F. (2020). Explainable artificial intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI. Information Fusion, 58, 82–115. https://doi.org/10.1016/j.inffus.2019.12.012

Bai, G., Ge, Y., Scoby, D., Leavitt, B., Stoerger, V., Kirchgessner, N., … Awada, T. (2019). NU-Spidercam: A large-scale, cable-driven, integrated sensing and robotic system for advanced phenotyping, remote sensing, and agronomic research. Computers and Electronics in Agriculture, 160, 71–81. https://doi.org/10.1016/j.compag.2019.03.015

Bailey, B. N. (2019). Helios: A scalable 3D plant and environmental biophysical modeling framework. Frontiers in Plant Science, 10. https://doi.org/10.3389/fpls.2019.01185

Bao, Y., & Tang, L. (2016). Field-based robotic phenotyping for sorghum biomass yield component traits characterization using stereo vision. IFAC-PapersOnLine, 49(16), 265–270. https://doi.org/10.1016/j.ifacol.2016.10.049

Bao, Y., Gai, J., Xiang, L., & Tang, L. (2021). Field robotic systems for high-throughput plant phenotyping: A review and a case study. In High-Throughput Crop Phenotyping (pp. 13–38). Springer International Publishing. https://doi.org/10.1007/978-3-030-77867-3_2

Bucksch, A., Atta-Boateng, A., Azihou, A. F., Battogtokh, D., Baumgartner, A., Binder, B. M., … Chitwood, D. H. (2017). Morphological plant modeling: Unleashing geometric and topological potential within the plant sciences. Frontiers in Plant Science, 8, 900. https://doi.org/10.3389/fpls.2017.00900

Burnette, M., Kooper, R., Maloney, J. D., Rohde, G. S., Terstriep, J. A., Willis, C., ... & LeBauer, D. (2018). TERRA-REF data processing infrastructure. In Proceedings of the Practice and Experience on Advanced Research Computing (pp. 1–7). ttps://doi.org/10.1145/3219104.3219152

Chamara, N., Bai, G., & Ge, Y. (2023). AICropCAM: Deploying classification, segmentation, detection, and counting deep-learning models for crop monitoring on the edge. Computers and Electronics in Agriculture, 215, 108420. https://doi.org/10.1016/j.compag.2023.108420

Ćwiek-Kupczyńska, H., Altmann, T., Arend, D., Arnaud, E., Chen, D., Cornut, G., ... & Krajewski, P. (2016). Measures for interoperability of phenotypic data: minimum information requirements and formatting. Plant Methods, 12(1), 44. https://doi.org/10.1186/s13007-016-0144-4

Das Choudhury, S., Bashyam, S., Qiu, Y., Samal, A., & Awada, T. (2018). Holistic and component plant phenotyping using temporal image sequence. Plant Methods, 14(1), 35. https://doi.org/10.1186/s13007-018-0303-x

Das Choudhury, S., Guha, S., Das, A., Das, A. K., Samal, A., & Awada, T. (2022). Flowerphenonet: Automated flower detection from multi-view image sequences using deep neural networks for temporal plant phenotyping analysis. Remote Sensing, 14(24), 6252. https://doi.org/10.3390/rs14246252

Das Choudhury, S., Maturu, S., Samal, A., Stoerger, V., & Awada, T. (2020). Leveraging image analysis to compute 3D plant phenotypes based on voxel-grid plant reconstruction. Frontiers in Plant Science, 11, 521431. https://doi.org/10.3389/fpls.2020.521431

Feret, J. B., François, C., Asner, G. P., Gitelson, A. A., Martin, R. E., Bidel, L. P., ... & Jacquemoud, S. (2008). PROSPECT-4 and 5: Advances in the leaf optical properties model separating photosynthetic pigments. Remote Sensing of Environment, 112(6), 3030–3043. https://doi.org/10.1016/j.rse.2008.02.012

Friha, O., Ferrag, M. A., Shu, L., Maglaras, L., & Wang, X. (2021). Internet of Things for the future of smart agriculture: A comprehensive survey of emerging technologies. IEEE/CAA Journal of Automatica Sinica, 8(4), 718–752. https://doi.org/10.1109/JAS.2021.1003925

Gehan, M. A., Fahlgren, N., Abbasi, A., Berry, J. C., Callen, S. T., Chavez, L., ... & Sax, T. (2017). PlantCV v2: Image analysis software for high-throughput plant phenotyping. PeerJ, 5, e4088. https://doi.org/10.7717/peerj.4088

Jacquemoud, S., & Baret, F. (1990). PROSPECT: A model of leaf optical properties spectra. Remote Sensing of Environment, 34(2), 75–91. https://doi.org/10.1016/0034-4257(90)90100-Z

Jacquemoud, S., Verhoef, W., Baret, F., Bacour, C., Zarco-Tejada, P. J., Asner, G. P., ... & Ustin, S. L. (2009). PROSPECT+SAIL models: A review of use for vegetation characterization. Remote Sensing of Environment, 113, S56–S66. https://doi.org/10.1016/j.rse.2008.01.026

Jones, J. W., Hoogenboom, G., Porter, C. H., Boote, K. J., Batchelor, W. D., Hunt, L. A., ... & Ritchie, J. T. (2003). The DSSAT cropping system model. European Journal of Agronomy, 18(3–4), 235–265. https://doi.org/10.1016/S1161-0301(02)00107-7

Kassim, M. R. M. (2020). IoT applications in smart agriculture: Issues and challenges. In 2020 IEEE Conference on Open Systems (ICOS) (pp. 19–24). https://doi.org/10.1109/ICOS50156.2020.9293672

Keating, B. A., Carberry, P. S., Hammer, G. L., Probert, M. E., Robertson, M. J., Holzworth, D., ... & Smith, C. J. (2003). An overview of APSIM, a model designed for farming systems simulation. European Journal of Agronomy, 18(3–4), 267–288. https://doi.org/10.1016/S1161-0301(02)00108-9

Kim, J. Y. (2020). Roadmap to high-throughput phenotyping for plant breeding. Journal of Biosystems Engineering, 45(1), 43–55. https://doi.org/10.1007/s42853-019-00072-y

Kirchgessner, N., Liebisch, F., Yu, K., Pfeifer, J., Friedli, M., Hund, A., & Walter, A. (2017). The ETH field phenotyping platform FIP: A cable-suspended multi-sensor system. Functional Plant Biology, 44(1), 154–168. https://doi.org/10.1071/FP16165

Krafft, D., Scarboro, C. G., Hsieh, W., Doherty, C., Balint-Kurti, P., & Kudenov, M. (2024). Mitigating illumination-, leaf-, and view-angle dependencies in hyperspectral imaging using polarimetry. Plant Phenomics, 6, 0157. https://doi.org/10.34133/plantphenomics.0157

Krajewski, P., Chen, D., Ćwiek, H., van Dijk, A. D., Fiorani, F., Kersey, P., ... & Weise, S. (2015). Towards recommendations for metadata and data handling in plant phenotyping. Journal of Experimental Botany, 66(18), 5417–5427. https://doi.org/10.1093/jxb/erv271

Kudenov, M. W., Bauer, L., Larsen, J. C., Locke, A. M., Doherty, C. J., O'Connor, B., & Balint-Kurti, P. (2025). Mueller matrix spectral and polarimetric imaging for high throughput plant phenotyping. In Photonic Technologies in Plant and Agricultural Science II (Vol. 13357, pp. 86–91). SPIE. https://doi.org/10.1117/12.3049631

Kudenov, M. W., Krafft, D., Scarboro, C. G., Doherty, C. J., & Balint-Kurti, P. (2023). Hybrid spatial–temporal Mueller matrix imaging spectropolarimeter for high-throughput plant phenotyping. Applied Optics, 62(8), 2078–2091. https://doi.org/10.1364/AO.483870

Kuska, M. T., & Mahlein, A. K. (2018). Aiming at decision making in plant disease protection and phenotyping by the use of optical sensors. European Journal of Plant Pathology, 152(4), 987–992. https://doi.org/10.1007/s10658-018-1502-6

Li, X., & Xiang, L. (2024). Photorealistic robotic simulation using Unreal Engine 5 for agricultural applications. arXiv. https://doi.org/10.48550/arXiv.2405.18551

Lindblom, J., Lundström, C., Ljung, M., & Jonsson, A. (2017). Promoting sustainable intensification in precision agriculture: Review of decision support systems development and strategies. Precision Agriculture, 18(3), 309–331. https://doi.org/10.1007/s11119-016-9491-4

Lobet, G., Draye, X., & Périlleux, C. (2013). An online database for plant image analysis software tools. Plant Methods, 9(1), 38. https://doi.org/10.1186/1746-4811-9-38

Lochocki, E. B., Rohde, S., Jaiswal, D., Matthews, M. L., Miguez, F., Long, S. P., & McGrath, J. M. (2022). BioCro II: A software package for modular crop growth simulations. in silico Plants, 4(1), diac003. https://doi.org/10.1093/insilicoplants/diac003

Luo, D., & Kudenov, M. W. (2016). Neural network calibration of a snapshot birefringent Fourier transform spectrometer with periodic phase errors. Optics Express, 24(10), 11266. https://doi.org/10.1364/OE.24.011266

Masjedi, A., Zhao, J., Thompson, A. M., Yang, K. W., Flatt, J. E., Crawford, M. M., … Chapman, S. (2018, July). Sorghum biomass prediction using UAV-based remote sensing data and crop model simulation. In IGARSS 2018 – IEEE International Geoscience and Remote Sensing Symposium (pp. 7719–7722). IEEE. https://doi.org/10.1109/IGARSS.2018.8517644

Mishra, P. (2021). Chemometric approaches for calibrating high-throughput spectral imaging setups to support digital plant phenotyping by calibrating and transferring spectral models from a point spectrometer. Analytica Chimica Acta, 1187, 339154. https://doi.org/10.1016/j.aca.2021.339154

Neveu, P., Tireau, A., Hilgert, N., Nègre, V., Mineau‐Cesari, J., Brichet, N., ... & Cabrera‐Bosquet, L. (2019). Dealing with multi-source and multi-scale information in plant phenomics: the ontology-driven Phenotyping Hybrid Information System. New Phytologist, 221(1), 588–601. https://doi.org/10.1111/nph.15385

Papoutsoglou, E. A., Faria, D., Arend, D., Arnaud, E., Athanasiadis, I. N., Chaves, I., ... & Pommier, C. (2020). Enabling reusability of plant phenomic datasets with MIAPPE 1.1. New Phytologist, 227(1), 260–273. https://doi.org/10.1111/nph.16544

Pauli, D., Andrade-Sanchez, P., Carmo-Silva, A. E., Gazave, E., French, A. N., Heun, J., … Gore, M. A. (2016). Field-based high-throughput plant phenotyping reveals the temporal patterns of quantitative trait loci associated with stress-responsive traits in cotton. G3: Genes, Genomes, Genetics, 6(4), 865–879. https://doi.org/10.1534/g3.115.023515

Pierpaoli, E., Carli, G., Pignatti, E., & Canavari, M. (2013). Drivers of precision agriculture technologies adoption: A literature review. Procedia Technology, 8, 61–69. https://doi.org/10.1016/j.protcy.2013.11.010

Reynolds, D., Baret, F., Welcker, C., Bostrom, A., Ball, J., Cellini, F., … Tardieu, F. (2019). What is cost-efficient phenotyping? Optimizing costs for different scenarios. Plant Science, 282, 14–22. https://doi.org/10.1016/j.plantsci.2018.06.015

Roberts, D. R., Bahn, V., Ciuti, S., Boyce, M. S., Elith, J., Guillera‐Arroita, G., ... & Dormann, C. F. (2017). Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure. Ecography, 40(8), 913–929. https://doi.org/10.1111/ecog.02881

Sangjan, W., Carter, A. H., Pumphrey, M. O., Jitkov, V., & Sankaran, S. (2021). Development of a Raspberry Pi-based sensor system for automated in-field monitoring to support crop breeding programs. Inventions, 6(2), 42. https://doi.org/10.3390/inventions6020042

Sankaran, S., Khot, L. R., & Carter, A. H. (2015). Field-based crop phenotyping: Multispectral aerial imaging for evaluation of winter wheat emergence and spring stand. Computers and Electronics in Agriculture, 118, 372–379. https://doi.org/10.1016/j.compag.2015.09.001

Saqib, M., Almohamad, T. A., & Mehmood, R. M. (2020). A low-cost information monitoring system for smart farming applications. Sensors, 20(8), 2367. https://doi.org/10.3390/s20082367

Selby, P., Abbeloos, R., Backlund, J. E., Basterrechea Salido, M., Bauchet, G., Benites-Alfaro, O. E., ... & BrAPI Consortium. (2019). BrAPI—An application programming interface for plant breeding applications. Bioinformatics, 35(20), 4147–4155. https://doi.org/10.1093/bioinformatics/btz190

Shahab, H., Naeem, M., Iqbal, M., Aqeel, M., & Ullah, S. S. (2025). IoT-driven smart agricultural technology for real-time soil and crop optimization. Smart Agricultural Technology, 10, 100847. https://doi.org/10.1016/j.atech.2025.100847

Sheridan, R. P. (2013). Time-split cross-validation as a method for estimating the goodness of prospective prediction. Journal of Chemical Information and Modeling, 53(4), 783–790. https://doi.org/10.1021/ci400084k

Slater, P. N. (1985). Radiometric considerations in remote sensing. Proceedings of the IEEE, 6. https://doi.org/10.1109/PROC.1985.13231

Slater, P. N., Biggar, S. F., Holm, R. G., Jackson, R. D., Mao, Y., Moran, M. S., ... & Yuan, B. (1987). Reflectance- and radiance-based methods for the in-flight absolute calibration of multispectral sensors. Remote Sensing of Environment, 22(1), 11–37. https://doi.org/10.1016/0034-4257(87)90026-5

Soussi, E., Zero, E., Sacile, R., Trinchero, D., & Fossa, M. (2024). Smart sensors and smart data for precision agriculture: A review. Sensors, 24(8), 2647. https://doi.org/10.3390/s24082647

Su, W.-H. (2020). Advanced machine learning in point spectroscopy, RGB- and hyperspectral-imaging for automatic discriminations of crops and weeds: A review. Smart Cities, 3(3), 767–792. https://doi.org/10.3390/smartcities3030039

Swaminathan, V., Thomasson, J. A., Hardin, R. G., Rajan, N., & Raman, R. (2024). Radiometric calibration of UAV multispectral images under changing illumination conditions with a downwelling light sensor. The Plant Phenome Journal, 7(1), e70005. https://doi.org/10.1002/ppj2.70005

Teillet, P. M., Slater, P. N., Ding, Y., Santer, R. P., Jackson, R. D., & Moran, M. S. (1990). Three methods for the absolute calibration of the NOAA AVHRR sensors in-flight. Remote Sensing of Environment, 31(2), 105–120. https://doi.org/10.1016/0034-4257(90)90060-Y

Thome, K. J., Biggar, S. F., Gellman, D. L., & Slater, P. N. (1994). Absolute-radiometric calibration of Landsat-5 Thematic Mapper and the proposed calibration of the Advanced Spaceborne Thermal Emission and Reflection Radiometer. In IGARSS ’94: IEEE International Geoscience and Remote Sensing Symposium (Vol. 4, pp. 2295–2297). https://doi.org/10.1109/IGARSS.1994.399718

Thome, K. J., Helder, D. L., Aaron, D., & Dewald, J. D. (2004). Landsat-5 TM and Landsat-7 ETM+ absolute radiometric calibration using the reflectance-based method. IEEE Transactions on Geoscience and Remote Sensing, 42(12), 2777–2785. https://doi.org/10.1109/TGRS.2004.839085

Valavi, R. (2025). blockCV [R package]. https://github.com/rvalavi/blockCV

Virlet, N., Sabermanesh, K., Sadeghi-Tehran, P., & Hawkesford, M. J. (2017). Field Scanalyzer: An automated robotic field phenotyping platform for detailed crop monitoring. Functional Plant Biology, 44(1), 143–153. https://doi.org/10.1071/FP16163

Wang, Y., Kootstra, G., Yang, Z., & Khan, H. A. (2024). UAV multispectral remote sensing for agriculture: A comparative study of radiometric correction methods under varying illumination conditions. Biosystems Engineering, 248, 240–254. https://doi.org/10.1016/j.biosystemseng.2024.11.005

Wilkinson, M. D., Dumontier, M., Aalbersberg, I. J., Appleton, G., Axton, M., Baak, A., ... & Mons, B. (2016). The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data, 3(1), 1-9. https://doi.org/10.1038/sdata.2016.18

Zhang, H., Ge, Y., Xie, X., Atefi, A., Wijewardane, N. K., & Thapa, S. (2022). High throughput analysis of leaf chlorophyll content in sorghum using RGB, hyperspectral, and fluorescence imaging and sensor fusion. Plant Methods, 18, 1–17. https://doi.org/10.1186/s13007-022-00892-0

Zhi, X., Chen, Q., Han, Y., Yang, B., Wang, Y., Wu, F., … Li, Y. (2025). Multi-modal feature integration from UAV-RGB imagery for high-precision cotton phenotyping: A paradigm shift toward cost-effective agricultural remote sensing. Computers and Electronics in Agriculture, 239, 111002. https://doi.org/10.1016/j.compag.2024.111002

Zhu, H., Huang, Y., An, Z., Zhang, H., Han, Y., Zhao, Z., ... & Hou, C. (2024). Assessing radiometric calibration methods for multispectral UAV imagery and the influence of illumination, flight altitude and flight time on reflectance, vegetation index and inversion of winter wheat AGB and LAI. Computers and Electronics in Agriculture, 219, 108821. https://doi.org/10.1016/j.compag.2024.108821

Attachments

Land Grant Participating States/Institutions

DE, KY, LA, MO, NC, NE, WA

Non Land Grant Participating States/Institutions

Danforth Plant Science Center