AI for Climate-Resilient Cropping Systems
Providing farmers and land managers with accurate and actionable crop suitability projections.
Many farmers and agricultural support specialists (e.g., university extension officers, soil district managers, co-operative farm leads) face barriers to effectively using relevant information from long-term climate projections. Existing climate/crop projection interfaces require users to know a priori which climate models and scenarios they are interested in, along with the technical details of how their crop of interest is represented in those models. Furthermore, oversimplified representations of climate change in existing crop models underestimate the impacts of extreme weather events on crop growth but these uncertainties are often hidden to end users and decision makers. To address these issues, we are developing use-inspired generative machine learning tools with user-centric web interfaces that communicate how climate change will impact future crop growth and suitability in the Conterminous United States (CONUS) in accessible and actionable ways. Our goal is to allow users to quickly visualize their own custom agro-climatic indicators in real time, while allowing them to opt-in to a richer presentation of the modeling details. This approach synthesizes insights across various state-of-the-art climate and crop models and focuses on the practical implications of their core assumptions, such as the probability of exceeding key temperature thresholds during the growing season. Our objectives are to: 1) Develop a prototype system for projecting key crop indicators in a changing climate trained on existing climate and crop-model simulations, focused on major crops over CONUS. 2) Iteratively redesign workflow based on results of structured interviews with food producers and county-level agricultural officials to determine user needs for: regionally-relevant crop types and agricultural indicators, spatial and temporal scale, and presentation formats (e.g. maps, reports, uncertainty assessments). 3) Deploy streamlined, indicator-focused climate and crop model emulators, integrated climate and crop model projection datasets, and an interactive web-application and analysis tool in partnership with local agricultural extension offices.
A Convergence Research Project
We are uniquely positioned to accomplish these tasks as a convergent team of climate, crop, soil, social, and data scientists with connections to university extension offices across the US.
How has the CORE Institute helped the team?
The incubator was crucial in helping us clarify the motivations and stakeholders of our project. Having the opportunity to collaborate in person with each other for an extended period of time enhanced our understanding of not just each other's expertise, but also our individual personalities and problem-solving approaches. This experience proved invaluable in forming a cohesive and collaborative team from the ground up. Our comprehensive discussions on stakeholders, customers, users, and use-inspired research have become key strategies we intend to implement in all our future proposals.
