A FRAMEWORK FOR IMPROVED DATA FLOW AND INTEROPERABILITY THROUGH DATA STRUCTURES, AGRICULTURAL SYSTEM MODELS, AND DECISION SUPPORT TOOLS
The agricultural data landscape is largely dysfunctional because of the industry’s highvariability in scale, scope, technological adoption, and relationships. Integrated data andmodels of agricultural sub-systems could be used to advance decision-making, but interoperability challenges prevent successful innovation. In this work, temporal and geospatial indexing strategies and aggregation were explored toward the development of functional data structures for soils, weather, solar, and machinery-collected yield data that enhance data context, scalability, and sharability.
The data structures were then employed in the creation of decision support tools including web-based applications and visualizations. One such tool leveraged a geospatial indexing technique called geohashing to visualize dense yield data and measure the outcomes of on-farm yield trials. Additionally, the proposed scalable, open-standard data structures were used to drive a soil water balance model that can provide insights into soil moisture conditions critical to farm planning, logistics, and irrigation. The model integrates SSURGO soil data,weather data from the Applied Climate Information System, and solar data from the National Solar Radiation Database in order to compute a soil water balance, returning values including runoff, evaporation, and soil moisture in an automated, continuous, and incremental manner.
The approach leveraged the Open Ag Data Alliance framework to demonstrate how the data structures can be delivered through sharable Representational State Transfer Application Programming Interfaces and to run the model in a service-oriented manner such that it can be operated continuously and incrementally, which is essential for driving real-time decision support tools. The implementations rely heavily on the Javascript Object Notation data schemas leveraged by Javascript/Typescript front-end web applications and back-end services delivered through Docker containers. The approach embraces modular coding concepts and several levels of open source utility packages were published for interacting with data sources and supporting the service-based operations.
By making use of the strategies laid out by this framework, industry and research canenhance data-based decision making through models and tools. Developers and researchers will be better equipped to take on the data wrangling tasks involved in retrieving and parsing unfamiliar datasets, moving them throughout information technology systems, and understanding those datasets down to a semantic level.
History
Degree Type
- Doctor of Philosophy
Department
- Agricultural and Biological Engineering
Campus location
- West Lafayette