Providing Actionable Intelligence from Multi-Modal, Multi-Scale Data in Agriculture

As an industry, agriculture is rapidly adopting modern data collection and analysis procedures to help guide farming practices. Despite the increasing amounts of data in agriculture and almost universal adoption rates of major seed, synthetic fertilizer, and chemical innovations, worldwide crop yield growth rates have plateaued to less than 1% annually. This increased data on multiple aspects of cropping systems has not yet produced major improvements in total factor productivity, and in order to meet the United Nations anticipated food demand in 2050, crop yields must grow at 1.8% per year from a 2014 baseline. A new class of machine learning tools designed to create actionable intelligence will aid farmers in predicting yield, managing risks, achieving sustainability objectives, and optimizing input costs, while also enabling agricultural amendment producers to more precisely target their products such as fertilizers and biotic additives to the soil and seed regimes where they will be most effective. With this technology, total factor productivity can increase up to 2% annually for commodity and biofuel crops, while also improving water, nitrogen, and phosphorus use efficiency by 50%. Creating a multi-scale, multi-modal data framework using the latest machine learning techniques to aggregate, integrate, and analyze complex environmental sensing data into actionable intelligence will be the key to finding the biological drives for the next advancement in sustainable yield growth in biofuel feedstocks and beyond. By developing a graphical user interface linked to a cloud computing environment to visually represent the data, this technology will empower the farmer with understandable information and tailored prescriptions, reducing environmental impact and optimizing farming practice. These data integration tool sets will have utility throughout the agricultural and environmental sciences, including biofuel feedstock production, the cultivation of marginal lands, and the study of natural ecosystems. The development of reliable models of biogeochemical processes in cultivated regimes on monoculture-managed lands will be the building blocks and proof of principle for more sophisticated studies in natural contexts and emergent phenomena, and will be of broad interest to the larger molecular ecosystem biology research community.