Recipient Organization
UNIV OF MINNESOTA
(N/A)
ST PAUL,MN 55108
Performing Department
Soil, Water, and Climate
Non Technical Summary
Simultaneously increasing crop yield, resource use efficiency, economic returns, soil health and reducing negative environmental impacts to achieve food security, soil health, climate mitigation and sustainable development is one of the most significant challenges of the 21st century. Precision agriculture has been regarded as a promising approach to meet this challenge. Currently, precision agriculture research has mainly focused on nutrient, water, pesticide, seeding, and tillage management to improve resource use efficiency, often without significantly affecting crop yield. Therefore, precision agriculture must move from optimization of single input or management practice into developing integrated precision crop management systems. There is an urgent need to take a multidisciplinary approach to combine crop growth modeling, proximal and remote sensing and artificial intelligence to systematically develop and evaluate related technologies to support the development and applications of integrated climate-smart precision crop management systems.Significant progress has been made with precision management of single agricultural inputs or practices, however, farmers manage their crops as a system. It is not as simple as adding different precision management practices together, because they have complex interactions. It is critically important to develop integrated climate-smart precision crop management systems that can be applied in Minnesota, Midwest US and the nation to make significant contributions to improve food security, resource use efficiency, economic viability, soil health, climate change mitigation and environmental sustainability. The project will form a multi-disciplinary team to combine crop growth modeling, proximal and remote sensing, machine learning to develop different integrated climate-smart precision crop management systems according to different farm conditions and conduct on-farm trials to evaluate and demonstrate the potential benefits of such systems in comparision with current farmer practices.
Animal Health Component
70%
Research Effort Categories
Basic
10%
Applied
70%
Developmental
20%
Goals / Objectives
The major goals of this project are to develop different integrated climate-smart precision crop management systems according to different farm situations that can improve crop productivity, resource use efficiency, profitability, soil health, adaptability to climate change and reduce negative environmental impacts (nitrate-N leaching,greenhouse gas emission, erosion and runoff, etc.). The specific objectives include:1) evaluate the potential of using artificial intelligence (machine learning) methods to improve management zone delineation, soil and plant nutrient and water status prediction using proximal and remote sensing technologies;2) develop different integrated precision crop(corn) management strategies by combining crop growth modeling, proximal and remote sensing, and artificial intelligence; and,3) evaluate the potential benefits of the developed integrated precision crop management systems as compared with current farmer practices and other management strategies through on-farm trials.
Project Methods
Multi-disciplinary Team FormationI will develop collaboration with faculty members and researchers with different backgrounds to form a multi-disciplinary team to develop integrated climate-smart precision crop management systems, including Carl Rosen (soil fertility, soil health, potato management), Vasudha Sharma (precision irrigation), Fabian Fernandez (nutrient management, environmental quality), Jeffrey Coulter (corn-based cropping systems), Anna Cates (soil health), Kyungsoo Yoo (soil-landscape analysis), David Mulla (precision agriculture, precision conservation, geospatial analysis, environmental quality, soil and crop modeling), Ce Yang (machine learning, hyperspectral remote sensing), Robert Koch (precision plant protection), and Solomon Folle (crop growth modeling).Objective 1:Intensive soil samples will be collected from different commercial corn fields. For selected field(s), in addition to soil samples, intensive corn leaf samples, EC sensor data and gamma ray sensor data will also be collected. A Niton pXRF sensor and an ASD FieldSpec 3 sensor will be used to scan corn leave samples as well as the soil samples to estimate plant and soil nutrients. Different machine learning models will be evaluated to predict different soil nutrient contents using soil sensing data as well as terrain attributes. The predicted soil nutrient maps will be used to make variable rate P, K, and lime applications.Field experiments involving different N and irrigation rates will be conducted in central Minnesota, different sensing technologies (Dualex leaf fluorescence sensor, Crop Circle Phenom, UAV-based multispectral and thermal sensing, and hyperspectral sensing, etc.) will be used to collect reflectance and canopy and air temperature data. Machine learning models will be developed using multi-source data fusion to predict crop nutrient and water status to guide variable rate nutrient(N) and irrigation management.Intensive soil, landscape, management, yield and satellite remote sensing data have been collected under the support of other projects. Different machine learning models will be evaluated to identify key soil and landscape variables influencing crop yield variability. After key variables are determined, different algorithms will be evaluated for MZ delineation. The best performing method will be determined and used to delineate fields into different MZs. The MZs will be evaluated for the potential of variable rate P, K, lime, S, variety and seeding based on sampling data or on-farm trial data.Objective 2:After the MZs are verified, the CERES-Maize crop growth model will be calibrated and validated using field-specific data, and then be used to determine suitable corn hybrid-specific optimum seeding rates and N rates for different MZs based on crop model simulations using the past 10-20 years of weather data, as demonstrated by Miao et al. (2006). Then a moderate amount (e.g., 1/3) of such model-based MZ-specific N rates will be applied before planting. An innovative multi-parameter Arable Mark 2 sensor system will be installed in the field right after planting. The data will be used for in-season calibration of crop growth model and together with historical or forecasted weather data for predicting management zone-specific optimal sidedress N rates (Wang et al., 2021). High spatial and temporal resolution PlanetScope satellite remote sensing images together with soil, weather terrain and management information will be used to develop machine learning models to predict crop biomass and plant N uptake. They will be used to fine tune zone-specific optimal sidedress N rates by considering within-zone variability. The Aral Mark 2 sensor data will be used together with UAV multispectral and thermal remote sensing images to determine site-specific irrigation rates. Other precision N management strategies, including the remote sensing and calibration strip-based N management strategy and remote sensing and machine learning-based strategies will also be evaluated integrated into different systems.In addition, we propose an innovative integrated climate-smart N management system (ICNMS). This system will include the following components and strategies: 1) Using manure to supply about 100% of estimated phosphorus (P) needs of corn; 2) Variable rate sidedress N application using the remote sensing and calibration strip-based precision N management strategy (RS&CS-PNM): This strategy was developed by the PI's group for corn in fields without manure application (Cummings et al., 2021b) and will be modified for fields with manure in this study. Before planting, different N rate strips will be implemented using commercial fertilizers, including 110% farmer's N rate (FNR, which is estimated total N need adjusted for estimated N supply from applied manure), 100%FNR, 80%FNR, 60%FNR and 40%FNR. These strips will run from one side of the field to the other side and be replicated across a field. The field strips will be delineated into grids, and five grids next to each other covering different N rates form a response block. High spatial (3.7 meter) and high temporal (daily revisit time) resolution PlanetScope satellite remote sensing images will be used to monitor corn growth and create response curves to N rates applied before planting, and determine optimal N rate in each response block. By deducting the N amount already applied before planting, site-specific sidedress N application rate will be determined for each grid at about V7-V9 growth stages to develop a prescription map. It will be used to guide variable rate sidedress N application; 3) Cover crops will be planted using a unmanned aerial vehicle (UAV) in late August and terminated in April the following year to take up residual N fertilizers in the soil to reduce leaching and provide the N to the crop in the following season.This ICSNMS technology is an integrated approach, and use year-, site- and hybrid-specific corn response to N information to make in-season N management decisions. High spatial and high temporal resolution PlanetScope satellite remote sensing images are used to monitor corn growth and responses to preplant N rates to make in-season N management decisions, and this is more practical than proximal sensors, UAV remote sensing or other satellite remote sensing platforms. It incorporated multiple N management strategies and is adaptive to and stable across different weather conditions than using one strategy. It is expected to increase corn yield, soil organic matter, N use efficiency, economic returns, improve soil health, reduce N runoff and erosion losses and nitrate leaching to protect surface and groundwater quality, reduce N2O emissions and increase carbon sequestration to mitigate climate change. Wide adoption of this system technology will make significant contributions to food security, environmental protection, and climate change mitigation.According to different field and farm conditions, different integrated precision crop management systems will be developed.Objective 3:On-farm strip trials will be conducted in different regions of Minnesota in 3 years to evaluate different integrated precision crop management systems in comparison of farmers' current management practices and regional optimum crop management systems (or best management practices) for yield, nutrient use efficiency, N losses, Greenhous gas emission, economic returns and soil health. The strips will run across the fields using farmers' or cooperatives' equipment. In addition to evaluations based on on-farm trials, crop growth models will also be used to simulate different crop management strategies and evaluate them under different environmental conditions.