Statistical Estimation of Crop Management Zones from Multi-Year Yield Data and the Oada Api Framework

Alexander Layton, Purdue University

Abstract

Precision agriculture equipment enables treating different areas of a field differently (i.e., site-specific management). The first part of this work presents an algorithm for inferring the management zones of fields based on multiple years’ yield data. It seeks regions that correspond to the same underlying yield distribution. Zones are assumed to be the same each year, but their distributions are allowed to change year-to-year to account for variability. Zones are estimated using stochastic expectation maximization and maximization of the posterior marginals. The underlying assumption is that the yields corresponding to a given zone will behave similarly, and are drawn from the same distribution. This requires only the yield data automatically collected during harvest. This method requires no crop-specific calibration. The second part of this work presents the Open Ag Data Alliance (OADA) Application Programming Interface (API) framework. It is a generic specification that can be used by third parties’ APIs to reduce the complexity of interoperating with multiple entities. This is especially useful in intercloud scenarios, for example, moving data between a farmer, a processor, and a distributor. Several existing standards that were leveraged are identified, the graph-based data representation is illustrated, and key API specifications and features are highlighted. Some of the contributions of OADA include user-centric Representational State Transfer (REST) so users can select API clients, resource meta-data stored externally to the resource, live data graphs via change feeds, intercloud data push, and format indifference. A reference implementation is presented and use cases are demonstrated.

Degree

Ph.D.

Advisors

Krogmeier, Purdue University.

Subject Area

Agriculture

COinS