Project Number: 2036-61000-018-014-R
Project Type: Reimbursable Cooperative Agreement
Start Date: Sep 1, 2020
End Date: Aug 31, 2025
This five-year integrated FASE CAP project proposes to enhance the profitability and sustainability of irrigated agriculture in the western US by characterizing and managing system-wide interactions between crop genetics, environmental factors (e.g. climate and soil health), and farm management across spatiotemporal scales. The project will lay the foundation for a long-term shift (10-20 years after the project) to highly-automated mechanized farm management systems (irrigation, nutrient, salinity, pest), while improving currently used technology in the short-term (during the project) and medium-term (five/ten years after the project). Farmland within the Colorado River Basin (CRB) and the Salinas Valley, CA (SV) will be used as study areas. The project will disseminate current and emerging DA knowledge and management tools among farmers and other stakeholders, so that on-farm decision-making can take full advantage of big data computing and AI. The project will educate students to become the next generation of farmers, managers, consultants, and scientists applying data science tools to sustainable agriculture. Likewise, the project will empower stakeholders, extension personnel, and other change agents to bridge the digital divide across research and farming practices.
We will involve farmers, irrigation specialists, and agricultural consultants in the Colorado River Basin (CRB) and Salinas Valley (SV), state and local water agencies, and fellow scientists. Research and extension success will be evaluated by an external professional evaluator and by an advisory board of stakeholders (including Rio Farms [SV & CRB], Colorado Corn Growers [CRB], JV Farms [CRB], Top Flavor Farms [CRB], and others involved in defining project objectives). SO1. We will use daily, very high-resolution (<5 m) satellite imagery from Planet Labs, Inc. [PL] combined with medium resolution thermal imagery from Landsat, soil maps (e.g., POLARIS), gridded weather data, and deep learning (DL) to develop the proposed water, nutrient, salinity management tool. The DL models and algorithms will include Tensor Analysis methods, a class of powerful machine learning techniques that can extract actionable knowledge with as little human supervision as possible. Building on recent research, we will optimize existing crop water requirement (CWR) models using DL with an established training set of ground measurements from existing and future sites. For regional-scale CWR, we will use PL imagery along with ground information to model crop parameters (e.g., crop type, salt tolerance, growth stage), classify land use (e.g., farmland, roadways), irrigation system types (e.g., flood, sprinkler, drip), and refine soil maps. We will then provide irrigation recommendations using in-season CWR, soil water storage, and leaching requirements. Crop parameters and fertilization systems will be incorporated to predict the appropriate amount and timing of applications. We will integrate all modeling efforts into a single AI platform, which will be accessible via a web-based app (similar to smartirrigationapps.org/cotton-app). We will test the app both for uniform and site-specific management using field trials. SO2. We will use PL multispectral imagery to detect pest emergence using DL calibrated with controlled plot experiments (e.g., at UC Desert Research and Extension Center) and field surveys at organic farms. SO3. Based on the success of current ongoing programs by extension members of the team (e.g., UC Cooperative Extension Statewide Irrigation Training), we will establish a training program consisting of classes, workshops, seminars and on-farm demonstrations. These will be geared toward training farmers and other stakeholders on using current knowledge and DA tools to best manage farm operations in years 1-3. SO4. We will disseminate knowledge and tools developed in the project with training programs like those in SO3, in years 3-5.