The global potato industry is experiencing a technological transformation, driven by precision agriculture, artificial intelligence (AI), and remote sensing technologies. These innovations are helping farmers make data-driven decisions to optimize nitrogen use and estimates tuber yields with increased accuracy.

A new study led by Bilal Javed (PhD Candidate in the Living Lab Initiative project at Agriculture and Agri-Food Canada – AAFC) and supervised by Dr. Athyna Cambouris (Scientist at AAFC), in collaboration with the School of Integrative Plant Science at Cornell University, NY, United States, demonstrates how drone-based multispectral imaging, satellite data, and AI-driven models can enhance nitrogen management and improve yield forecasting in potato farming.

Their research, has recently been published in Field Crops Research (Volume 324, 2025, Article 109794), provides insights into in-season nitrogen assessment and pre-harvest, yield estimation, and sustainability. The scientific paper is titled “In-season nitrogen status and pre-harvest potato yield assessment using air-spaceborne imagery with AI techniques.”

For farmers, agronomists, and researchers, these findings represent advancements in efficiency, sustainability, and profitability in potato production.

The Nitrogen Challenge in Potato Farming

Nitrogen plays a critical role in potato production, influencing canopy development, tuber formation, and overall yield. However, managing nitrogen effectively presents several challenges.

Insufficient nitrogen application results in stunted growth, premature leaf senescence, and reduced yields. Excess nitrogen promotes excessive vine growth, delayed tuber maturation, and lower starch content. Additionally, excessive nitrogen application can lead to nitrate leaching into groundwater, greenhouse gas emissions, and soil degradation.

Traditionally, nitrogen status is assessed using manual petiole nitrate sampling and laboratory analysis. While accurate, this method is time-consuming, labor-intensive, and impractical for large-scale monitoring.

Mr. Javed propose an AI-driven alternative, using remote sensing data from UAVs (drones) and Sentinel-2 satellites to monitor nitrogen levels in real-time.

Harnessing AI and Remote Sensing for Precision Agriculture

The study was carried out on four commercial potato fields in Prince Edward Island, Canada. Researchers focused on two critical crop stages: the early flowering stage (S1) for assessing nitrogen status and the pre-harvest stage (H) for estimating final tuber yield.

To collect high-resolution data, researchers used drones equipped with multispectral cameras to capture detailed field imagery at 5 cm resolution and Sentinel-2 satellite data providing wide-scale coverage at 10 – 20 m resolution.

The images were processed using AI-powered machine learning models, trained on multispectral imagery and ground truth data as actual petiole nitrate concentrations and tuber yield data. The objective was to determine how well these remote sensing techniques could predict in-season nitrogen levels and pre-harvest yield outcomes.

Key Findings

The study revealed that AI-driven models trained on drone and satellite imagery can predict nitrogen status and yield with high accuracy.

AI assessed in-season nitrogen status with an accuracy of over 85%. Tuber yield estimation had a 13.8% error margin, significantly better than conventional forecasting methods. Red-edge spectral bands and chlorophyll indices were the best indicators of nitrogen levels. Yield assessment was most accurate when using soil-adjusted vegetation indices and a newly developed Potato Productivity Index.

These findings highlight the potential for real-time, data-driven decision-making in potato farming.

Implications for Farmers, Agronomists, and Researchers

Faster, Non-Invasive Nitrogen Monitoring

Traditional nitrogen assessments require field sampling and lab testing, which can take several days including a lot of labour. With AI-powered remote sensing, entire fields can be scanned within minutes, identifying areas that need more or less nitrogen without disrupting crop growth.

More Accurate Yield Predictions for Better Planning

Potato yield forecasting has historically been difficult due to variable growing conditions, soil health, and weather fluctuations. AI-powered yield predictions allow growers to plan storage, logistics, and marketing strategies more effectively, helping to maximize profits and reduce waste.

Reducing Nitrogen Waste and Environmental Impact

Excess nitrogen not only increases production costs but also harms the environment. AI-based monitoring helps apply the correct amount of nitrogen at the right time, reducing nitrate leaching, greenhouse gas emissions, and soil degradation. By improving nitrogen efficiency, growers can cut costs while improving sustainability.

Scalable Solutions for Farms of All Sizes

Whether managing a small family farm or a large-scale commercial operation, remote sensing offers scalable solutions. Drones provide high-resolution data for precision applications at the field level, while satellites offer large-scale insights, making them ideal for regional and national agricultural monitoring programs. Farmers can choose the technology that best fits their operation and budget.

Additional Benefits of AI-Driven Precision Agriculture

Improved Crop Health Monitoring

Beyond nitrogen assessment, AI-powered remote sensing can detect signs of disease, pest infestations, and water stress earlier than traditional scouting methods. Early detection allows for timely interventions, reducing crop losses and improving overall productivity.

Better Resource Allocation

Precision agriculture ensures that inputs such as fertilizers, water, and pesticides are applied only where needed. This targeted approach reduces input costs and enhances environmental sustainability.

Enhanced Data Integration for Long-Term Decision-Making

By continuously collecting and analyzing field data over multiple seasons, AI models can identify long-term trends in soil health, crop performance, and environmental changes. This enables growers to refine their farming practices for sustained productivity.

Breaking Down the Technology: How AI Makes This Possible

Artificial intelligence is transforming how farmers assess crop health and predict yields. In this study, researchers tested several AI models—Random Forest, Bagged Trees, and Gradient Boosting Machines—to determine which provided the most accurate nitrogen and yield predictions.

These models analyze multispectral imaging data collected from UAVs and Sentinel-2 satellites, detecting subtle variations in plant health that traditional methods might miss.

By training AI models on real field data, researchers identified key indicators: the Canopy Chlorophyll Content Index (CCCI) and Red Edge spectral bands were most effective for nitrogen monitoring, while Optimized Soil-Adjusted Vegetation Index (OSAVI) and Potato Productivity Index (PPI) showed strong correlations with yield.

The study found that at the early flowering stage, the Bagged Trees Model using drone data had a 12.7% error margin for nitrogen assessment. For pre-harvest yield estimation, the Random Forest Model using Sentinel-2 data performed best, with a 13.8% error margin.

These results suggest that combining high-resolution drone imagery with broader satellite data can provide accurate, field-specific insights into crop performance.

AI-powered tools are becoming more accessible through integrated farm management software, allowing farmers to interpret data easily and apply insights without requiring advanced technical knowledge.

As AI models continue to improve, they will play an increasingly important role in helping potato growers make informed, data-driven decisions.

Looking Ahead: The Future of AI in Potato Farming

This study represents a step forward for precision agriculture. As AI-driven techniques become more widely adopted, they could soon become standard practice for automated nitrogen application systems, real-time crop health monitoring, and AI-powered predictive analytics for potato breeding and variety selection.

Enhancing adaptive nitrogen management

“Integrating AI with remote sensing technologies, such as drones and satellites, is transforming the way we manage nitrogen in agriculture. By providing real-time insights into machine learning algorithms, these innovations not only improve fertilizer efficiency but also enhance sustainability and optimize tuber yield,” says Bilal Javed.

“Instead of relying on static fertilization schedules, AI-driven insights can help farmers adjust nitrogen applications dynamically based on real-time field conditions. This could improve overall nitrogen use efficiency, reducing both costs and environmental risks.”

More precise decision-making

“Additionally, AI integration with climate and weather modeling will enable more precise decision-making. Predictive models can account for seasonal changes, precipitation patterns, and temperature fluctuations, helping growers mitigate risks associated with climate variability”.

Impact on breeding

AI-driven predictive analytics is advancing potato breeding by integrating crop phenotyping with genetic and environmental data, according to Bilal Javed.

“Machine learning models can analyze high-throughput phenotyping data—such as plant growth patterns, stress responses, and tuber characteristics—alongside genetic and environmental variables,” he explains.

“This enables breeders to identify promising traits more efficiently, accelerating the selection of high-yielding and disease-resistant potato varieties. While AI complements traditional breeding methods, its ability to process complex datasets enhances precision and reduces the time required for developing resilient cultivars.”

AI-powered automation

Furthermore, advancements in AI-powered automation could redefine labor-intensive aspects of potato farming. “In the near future, we may see autonomous drones and robotic systems equipped with AI capabilities for nitrogen application, weed control, and tuber size estimation, further enhancing operational efficiency,” according to Bilal Javed.

While the adoption of these technologies requires an initial investment in digital tools and expertise, the long-term benefits in cost savings, productivity, and environmental sustainability make AI a compelling solution for the future of potato farming.

Final Thoughts

For farmers, agronomists, and researchers interested in leveraging AI for better crop management, now is the time to explore investments in multispectral drone imaging for on-farm nitrogen monitoring, incorporating Sentinel-2 satellite data into farm management systems, and using AI-driven analytics platforms to improve nitrogen use efficiency and yield predictions.

“The successful adoption of these technologies will depend on knowledge-sharing and collaboration across the agricultural sector,” Bilal Javed points out.

“Extension services, agronomic researchers, and technology providers must work together to ensure that farmers have the training and support needed to integrate AI-based decision-making into their daily operations.

“Moreover, policymakers and agricultural organizations should encourage incentives and funding opportunities for farms looking to adopt AI-powered precision agriculture. Subsidies or cost-sharing programs for drone imaging and AI-based software platforms could accelerate adoption, making these innovations accessible to more growers”.

As AI technology continues to evolve, its accessibility and ease of use will improve, making it a standard component of farm management. The more data farmers collect and analyze over time, the more refined and valuable AI-driven insights will become, leading to continuous improvements in crop health, input efficiency, and overall farm profitability.

By embracing data-driven decision-making, potato growers can increase profitability, reduce environmental impact, and secure the future of their farming operations. The future of potato farming is digital, with AI playing a leading role in shaping a more efficient and sustainable industry.

Article Author: Lukie Pieterse, Editor/Publisher, Potato News Today
Source: Bilal Javed (PhD Candidate in the Living Lab Initiative project at Agriculture and Agri-Food Canada – AAFC)
Journal Source: Field Crops Research. Volume 324, 1 April 2025, 109794.
In-season nitrogen status and pre-harvest potato yield assessment using air-spaceborne imagery with AI techniques
DOI: https://doi.org/10.1016/j.fcr.2025.109794
Contact:
Bilal Javed would welcome responses from anyone with an interest in this research and will respond to any comments and queries. He can be reached at bilal.javed@agr.gc.ca
Cover image: Potato field during the full flowering stage of crop development. (Photo credit: Bilal Javed)
Acknowledgements:
Mr. Javed wishes to credit Dr. Athyna Cambouris for her invaluable mentorship and scientific contributions to the study, as well as all co-authors, whose expertise was instrumental in this work. Dr. Louis Longchamps (Cornell University) provided key insights into soil and crop science, while Dr. Parminder S. Basran (Cornell University) contributed to advanced analytical approaches. Dr. Marc Duchemin and Dr. Noura Ziadi (Agriculture and Agri-Food Canada) offered expertise in soil fertility and agronomic research. Dr. Antoine Karam (Université Laval) provided valuable input on soil and nutrient management. Dr. Stephanie Arnold and Dr. Adam Fenech (University of Prince Edward Island) contributed their knowledge of climate change and environmental adaptation. Their collective expertise significantly strengthened this research.