Research Themes
What is AI-MOD?
Artificial Intelligence for Multi-
Objective Decision Making (AI-MOD) helps people make better decisions when there is no single "best" solution exists. Using computer power, models, and data.
How it Works:
AI-MOD combines information from many sources, including satellite imagery, soil and weather data, and economic information. The AI compare many possible management options and identifies those that provide the best overall outcomes rather than optimizing for only one objective.
Why it is Important:
Researchers work to build tools that are transparent, easy to understand, and based on scientific knowledge. Many land management decisions involve trade-offs. For example, planting more crops may increase production but reduce wildlife habitat. AI-MOD helps decision-makers understand trade-offs before making important decisions by:
- Comparing conservation practices before implementation
- Evaluating different crop rotations or forest management plans
- Balancing food production with environmental sustainability
- Supporting regional land-use planning
- Informing agricultural and environmental policy
Finding Balance:
Instead of recommending one "perfect" solution, AI-MOD creates a range of high-performing options. These options show the best achievable combinations of goals, such as higher productivity, improved water quality, and strong economic returns, which in turn allows users to choose the solution that best matches their priorities.
Our Approach:
The research focuses on AI-guided multi-objective optimization for CSAF decision-making. This involves optimizing mitigation practices and bio-productivity across scales with multiple criteria, including equity and GHG reduction. The challenges involve complex problems necessitating fast response times, handling large areas and addressing spatial fragmentation. The research outlines three tasks: large-scale stochastic optimization for sequential decision-making, developing methods to compute Pareto frontiers for trade-offs and designing spatial optimization techniques to ensure contiguous solutions. Researchers are drawn on prior work in complex network optimization and spatial dependency modeling, particularly in the Amazon basin, showcasing their expertise in addressing CSAF challenges.
Dig Deeper:
- Download the printable AI-MOD fact sheet.
What is CLeAR?
Combined Learning and AI Reasoning (CLeAR) responds to the need for AI models that are transparent, explainable, and scientifically credible by embedding reasoning frameworks that clarify how predictions are generated. This in turn creates an AI that says “here is my answer”, as well as “here is why I think my answer is right”.
How it Works:
Researchers utilize data that is often noisy and incomplete. AI computer programs are fed this data, along with a set of parameters based on real-world observations and truths. The AI is then tasked with simulating the processes behind these observable outcomes in order to explain why something has happened.
Application in Agriculture & Forestry:
CLeAR is teaching AI to be both a smart learner and a careful reasoner, so it can solve agriculture and environmental problems while explaining how it reached its answers. The project is addressing problems such as:
- Understanding how soil carbon moves through soils
- Figuring out what molecules are present in an unknown soil sample
- Combining many kinds of information (maps, sensors, experiments, satellite data)
- Discovering hidden relationships in environmental systems
- Making AI tools more trustworthy for scientists and decision makers
Necessary Tools & Data:
Readily available datasets, including WoSIS, SoilGrids250, soil samples from the US and Europe, as well as satellite images are fed into programs that are trained to simulate the processes behind the data.
Our Approach:
AI-LEAF’s distinctive contribution is to unite advances in symbolic reasoning and machine learning within applied agricultural and forestry contexts, ensuring that technical innovation is matched with interpretability and trust. Our goals include:
- Advance foundational AI by combining machine learning with knowledge representation and reasoning to produce models that are interpretable, scientifically consistent, and scalable.
- Enable AI systems that can integrate diverse data types, infer causal relationships, and provide explanations that build trust with scientists, policymakers, and practitioners.
- Support use-inspired applications in agriculture and forestry by improving the interpretability and generalizability of AI models.
Dig Deeper:
- Computer scientists now understand that AI can be used to solve more complex problems if the machine can explain its reasoning. Read the Science article titles 'Artificial intelligence learns to reason' to learn more.
- Download the printable CLeAR fact sheet.
What is Computer Vision?
This research aims to improve estimates of how agricultural management practices affect soil health and sustainability. The goal is to detect areas vulnerable to degradation and optimize resilience and sustainable practices.
How it Works:
Computer Vision uses satellite data to assess the amount of crop residue cover and clods on farmland. These variables are widely used sources for measuring soil health and the efficacy of conservation efforts.
Clods & Crop Residue Cover:
Current conservation guidelines encourage 30%+ residue cover to optimize soil health. Computer Vision allows a quick and accurate estimate of current conservation efforts across broad areas to identify areas needing further soil conservation in order to reduce erosion, runoff, soil detachment and soil loss.
- Clods produced by tillage help to reduce erosion and improve water storage.
- Crop Residue Cover reduces erosion and improves soil infiltration.
Advanced Image Detection:
Using shortwave infrared light waves which are invisible to the human eye, Computer Vision is able to capture superior image details and assess them for cover and roughness with a greater degree of accuracy than current methods.
Our Approach:
AI-LEAF’s Computer Vision & Analysis (CV&A) theme applies multi-sensor imagery and advanced machine learning to transform the monitoring of crops, soils, and forests. By combining UAV, satellite, and hyperspectral data, CV&A detects drought, pests, and disease early and refines estimates of biomass, yield, and forest carbon. Co-designed with producers, foresters, and policymakers, these tools provide critical visual data to Digital Soil Twins, Agricultural Resilience, and AI-MOD, while releasing benchmark datasets, dashboards, and training modules to accelerate the adoption of improved agricultural and forestry practices.
Dig Deeper:
- Download the printable Computer Vision & Analysis fact sheet.
What is Digital Soil Twin?
This is an online digital twin of U.S. soils that delivers information regarding soil properties at different depths, topography, daily, field‑scale soil moisture maps, weather data, etc. It turns satellite, weather, soil survey, etc. into layers you can check on your phone or computer.
How it Works:
Digital Soil Twin uses knowledge-guided machine learning to fuse satellite, weather, land cover, soil surveys, and in-field sensors when available. This improves accuracy while keeping output scientifically consistent.
Available Layers & Forecasts:
- Soil texture & water‑holding capacity context at different depths
- Daily soil moisture map
- Crop irrigation requirements
Farming Applications:
- Timely tillage, planting, spraying, and harvest using trafficability and root‑zone moisture.
- Improve irrigation by reducing over/under‑watering with depth‑resolved moisture.
- Track effects of regenerative practices on soil carbon and explore carbon market readiness.
- Scout fewer acres by going straight to wet/dry problem spots.
- Document conditions for insurance, drought programs, or cost‑share reporting.
Our Approach:
The research focuses on AI-aided Digital Twins (AIDT) to enhance resilience planning for climate scenarios in CSAF. Digital twins are necessary for evaluating CSAF and AI concepts under diverse climate scenarios, considering impacts on agriculture and forestry. The digital twin is an in-silico representation of phenomena. One focus is on the creation of a digital for soils encompassing the contiguous United States. Soils play a critical role in agriculture settings including:
- providing essential nutrients to plants for their growth and development,
- supporting water retention and drainage, and thus helping to regulate soil moisture levels, and
- serving as a habitat for beneficial microorganisms that contribute to soil health.
Our work on digital twins involves modeling encompassing diverse meteorological and environmental phenomena to estimate critical soil properties such as moisture levels, salinity, and nutrient concentrations. By integrating diverse data sources alongside these models, we create a highly detailed and dynamic representation of real-world conditions. Users can interact with this digital twin through a visualization engine that enables the layering of multiple datasets, real-time animations, and the rendering of complex environmental processes, providing deeper insights and facilitating informed decision-making.
Dig Deeper:
- Check out the Soil Twin website.
- Download the printable Digital Soil Twin fact sheet.
What is FSAR?
Forest Sustainability & Agricultural Resilience (FSAR) uses artificial intelligence (AI) to help farms and forests adapt to changing environmental conditions.
How it Works:
By combining ecological science, economic information, and AI, researchers can better understand how drought, pests, wildfires, land-use changes, and other pressures affect working landscapes. These tools help people make informed decisions that support healthy ecosystems and productive agriculture.
Why it Matters:
Agricultural and forest systems face many challenges. Building resilience means preparing farms and forests to recover from these challenges while continuing to provide food, wildlife habitat, and other important resources. AI helps researchers answer crucial questions such as which areas are most vulnerable to drought or pests. These predictions allow producers, foresters, and policymakers to prepare for future challenges rather than simply reacting to them.
How AI Helps:
AI combines information from many sources, including:
- Satellite imagery
- Weather and climate records
- Soil and vegetation data
- Crop and forest measurements
- Economic and land-use information
By analyzing these large datasets, AI can identify patterns, predict future changes, and help land managers evaluate different management strategies before problems occur.
Predicting Future Changes:
By modeling soil health, crop productivity, forest conditions, water availability, and land-use change under different management and environmental scenarios, this research provides predictive insights that strengthen decision making.
Our Approach:
The AI-LAEF team examines how diverse climate-related stressors impact crop and forestry systems, as well as the effects of various adaptation efforts to minimize risks through shifts in system vulnerability and exposure. In crop systems, the team is building advanced machine learning models to better understand the impacts of hydroclimatic stressors on crop yields and examining adaptation-mitigation tradeoffs with irrigation. In forest systems, the team is building benchmark datasets, improving estimates of forest biomass, and examining risks from fires and other disturbances.
Dig Deeper:
- Download the printable FSAR fact sheet.
What is KGML?
Knowledge-Guided Machine Learning (KGML) combines traditional scientific knowledge with artificial intelligence. Instead of learning patterns from data alone, KGML incorporates rules of natural systems (such as plant growth or carbon cycling) to create more accurate and grounded predictions.
How it Works:
KGML uses large-scale environmental data, such as satellite imagery and weather observations, to pretrain AI foundation models that learn patterns across agricultural and natural systems.
Applications in Agriculture:
By leveraging the strengths of real world data and modeling, KGML can improve soil health and moisture predictions, carbon storage and air quality parameters. Through more accurate predictions, producers and land managers better understand field conditions, evaluate management practices, and make more informed decisions about irrigation, nutrient management, and conservation strategies.
Making it Available:
Innovations within KGML research (such as foundation models and benchmark datasets) allow researchers to build AI systems that learn from large-scale environmental data while remaining grounded in scientific knowledge about how natural systems work.
Our Approach:
Usability, trust, and co-design are essential for KGML’s success and require active integration of stakeholder needs and feedback. AI-LEAF Institute’s distinctive strength lies in its ability to connect advanced AI research with the practical needs of agriculture and forestry through intentional integration across science, policy, and education. The KGML theme embodies this value by combining technical innovation with stakeholder engagement to ensure that tools are not only accurate but also usable, trusted, and relevant. This integrated approach enables AI-LEAF to co-design solutions with partners, bridge the gap between research and policy, and prepare the next generation of practitioners who will apply these methods in the field.
Goals:
- Advance hybrid AI models that integrate process-based science with machine learning for improved estimation of soil carbon and atmospheric trace gases.
- Enable faster, more accurate predictions that are easier to interpret and integrate into on-farm decision support systems for estimating biogeochemical trace gases (COMET Planner, COMET Farm) at the regional and national scale.
- Position AI-LEAF as a national leader in KGML research, benchmark datasets, and stakeholder-ready applications.
Dig Deeper:
- Python Library for KGML GitHub
- Download the printable KGML fact sheet.