Harvest to
decision.

Henrys field / Winter wheat

An inspectable analysis of recorded yield and application rates, with uncertainty and scope attached.

WHAT THIS ANALYSIS SAYS

Loading analysis…

Recorded rate and yield

Solid: spatially adjusted · dashed: rate only
Band: 95% spatial block bootstrap interval

Curves cover the middle 80% of matched cell rates. Application rates remain in source lb/ac units; the 2018 nitrogen-versus-product definition is unresolved.

Public data: OFPEDATA / Montana State University01 / RESULTS

▥ fieldwork.FIELD ANALYSIS / METHODS & EVIDENCE

What supports
the interpretation.

Data preparation

Model and uncertainty

Twenty-metre cells summarize matched observations using median yield and median recorded application rate. Cells require at least three observations and receive equal weight.

yield = intercept + rate + rate² + x + y + x² + xy + y²

Ordinary least squares estimates the spatial trend adjusted association. The rate-only comparison omits the coordinate terms. The contrast compares the 25th and 75th percentile cell rates, averaged over the observed spatial basis.

Intervals use 400 fixed-seed bootstrap resamples of 100 m spatial blocks. Blocks are analytical groups, not independent randomized treatment plots. The block size is a stated modeling choice; dependence beyond blocks can make intervals too narrow.

Interpretation boundaries

The released files do not establish randomized treatment assignments or independent trial-unit identifiers. The reported contrast is an association, not a causal treatment effect.
The 2018 source describes both urea-product and nitrogen-mass interpretations for recorded rate. No nitrogen conversion is performed.
Screening is a review policy, not certified yield-monitor cleaning. No calibration or sensor-delay correction is invented.
Soil, weather and satellite observations have not been added. Missing application coverage is not interpreted as zero application.
Two harvest years demonstrate batch processing. They do not establish multi-farm generalizability.

Reproduction and source

Original source archives, normalized CSVs, full screening flags, analysis code, a dependency lockfile and the model manifest accompany the case study. The source package carries Montana State University's MIT notice.

Source: github.com/paulhegedus/OFPEDATA
Revision: ae70c5d8ad4fe154fd4b746a7ec43f3031a6e0a9
Manifest: data/manifest.json · SHA-256 hashes included.

Harvest todecision.