An Open-Source Ecosystem for Building Weather Driven Agricultural Decision Tools

A great deal of agricultural research has been undertaken to optimize the management of crops in terms of productivity, efficiency, resilience, and cost. Under the hood, agricultural research often combines data and models on interacting processes such as plant development, pest pressure, plant-soil dynamics, crop management practices, economics, and the weather. Although highly applied, transforming agricultural research into practical tools and guidelines for growers has proven far from trivial. This session reports on an ongoing body of work to develop an ecosystem of open source libraries to transform agricultural research into practical tools for California farmers. Leveraging R’s robust features for package development and publishing, the libraries provide many of the core functions for implementing agricultural models, including computation of specialized metrics for plant and insect development, API wrappers for agricultural weather station data, climate projections, and databases for calibrated weather data. Open-source development has been essential in this effort, allowing the developers to build upon an extensive body of low level data processing tools, adopt best practices for code development, bundle documentation and guidelines, and make their work accessible and discoverable by other researchers. Outcomes of this work include more productive researchers who can focus on their experiments and domain expertise instead of reinventing core algorithms. The libraries are also being used to power a growing body of decision support web apps that provide recommendations for production practices including irrigation, pest management, and treatments for winter dormancy.