ai-leaf

This research advances knowledge and understanding of artificial intelligence (AI) to appropriately scale decision support systems for climate-smart agriculture and forestry. Creative and original ideas to be explored in this research include physics-constrained, domain-interpretable, and spatial variability-aware deep learning for:

• reliable out-of-sample predictions for un- or under-sampled fields and parcels,

• AI emulators of physical models, and

• multiscale and multicriteria decision support tools

This will be applied to evaluating tradeoffs between alternative agricultural and forestry practices for greenhouse gas mitigation and adaptation under current and future climate scenarios. ISRIC – World Soil Information was granted a sub-award by the University of Minnesota (consortium leader) to contribute on mapping soil properties and indicators and uncertainty assessments.

The overall objective of ISRIC’s work in the ai-leaf project is to support the consortium with knowledge on machine learning for digital soil mapping and uncertainty assessments.

• Develop a machine learning / artificial intelligence model that predicts soil properties in space and time in the Continental United States (US) at suitable spatial and temporal resolution, using publicly available soil information and static and dynamic remote sensing-based covariates.

• Test the model developed and quantify prediction accuracy using appropriate cross-validation techniques. Improve the modelling framework in collaboration with project partners by testing more advanced convolutional neural network modelling approaches, such as deep learning algorithms or other suitable AI methods. 

• Apply the models developed and collaborate with partners on using model outputs for AI-enabled calculation of soil indicators for sustainable soil management at various scales.

University of Minnesota, Cornell University, Colorado State University, Purdue University, North Carolina State University, Delaware State University.

The ai-leaf work is funded by the United States Department of Agriculture – National Institute of Food and Agriculture (USDA-NIFA)

Visit ai-lift website