SBIR Phase II: Creating high-quality, lower-cost soil maps using machine learning algorithms

Soilserdem Llc, Ames IA

Investigators

Abstract

The broader/commercial impact of this Small Business Innovation Research (SBIR) Phase II project will be to produce high-quality (accurate/high-resolution) soil maps for agronomists and farmers at scale. Accurate soil information is a fundamental driver of better, more-efficient crop/soil management. This new branch of technology will deliver developed map products across various cropping systems that exist in the continental U.S., intersecting economic and environmental sustainability. Making site-specific soil fertility mapping information accessible to a diversity of land stewards is the goal of this project. Expected outcomes include more environmentally responsible farm management and manure and nutrient-management planning, precision farming, land use planning, planting decisions, evaluating stressors on plants, field conditioning, crop rotation, and prediction/interpretation of yields. Other benefits are increased farm profitability and increased soil health. This technology will result in increased crop yield while allowing for decreased input costs, leading to higher profitability in an industry that chronically suffers from low profit margins. The anticipated project outcomes meet NSF goals by advancing science, improving the lives and health of U.S. citizens, and potentially generating increased tax revenues and jobs via increased farm success. This innovative technology has three components that differentiate it from the best current technologies used to produce maps of essential soil nutrients. The first is applying generalized landscape quantification to drive optimal soil sample collection accommodating landscape variability, thereby eliminating the need to collect unnecessary soil samples. The second component leverages advanced machine-learning algorithms that are able to use the small number of uniquely collected soil samples to produce accurate predictions. Finally, the technology is a transferable model that does not necessitate additional hardware to achieve its results. As envisioned, this technology can select appropriate covariate mosaics to capture relevant soil variability irrespective of cropping system and management practices. The scope of this project will be beneficial to row cropping system across the U.S., specifically targeting corn-soy, potatoes, wheat, and cotton production. Unlike currently available methods that produce inadequate data for challenging (cost-prohibitive) mapping targets, this new technology will render those targets accessible and cost-effective with reliable accuracies. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.