Skip to study
SoilFuel — Subsidiary of Carbon OxyTech Inc., Canada
SoilFuel PlusField case study · BIE – E · 2026
View full study
Field case study / Methods & sources

Sources & methodology.

Sources

BIE-E-2026_harvest
John Deere WGS 84 point shapefile, DBF and metadata; exported 27 Oct 2026. Harvested 29 Sep 2026, 09:20–14:08 CST. Hard red spring wheat, AAC Brandon, S670 combine.

BIE - E Overview Application.pdf
John Deere Operations Center, generated 10 Oct 2026.

Export_20261010_Soilfuel_Plus_Application.zip
WGS 84 point shapefile, DBF attributes and Deere metadata. Product: Soilfuel Plus. Crop season: 2026; crop name blank.

BIE - E - Application Data.zip
Raw GS3 FDD/FDL archives retained locally. Repeated raw packages are not added to the spatial export totals; proprietary contents were not independently decoded.

Soilfuel product website · reviewed 13 Oct 2026. Product context only.

Calculation rules

Application turn sweeps: infer the 90 ft rigid-boom center from recorded section positions and their left-to-right offsets. Between adjacent samples, fit a cubic Hermite center path using the recorded start/end headings; rotate section edges with the shortest heading change. Tessellate the swept strips at no more than 2° heading change and 2 m center travel per step. This accounts for the different inner/outer section paths during a turn.

No sweep is joined across a time gap longer than 5 s, heading jump over 45°, boom-center disagreement over 0.5 m, near-stationary travel below 0.05 m, or position-versus-distance mismatch above the larger of 0.75 m and 35% of recorded travel. Such samples retain endpoint rectangles. These are transparent reconstruction thresholds, not manufacturer corrections. Recorded source areas, volumes and rates are unchanged; geometry is not a coverage union or a corrected application-rate estimate. Boom flex, GPS offsets and motion between logs are unmeasured.

Harvest footprints and application endpoint rectangles extend backward from each recorded position by its DISTANCE, oriented by Heading. Harvest footprint width uses each record’s SWATHWIDTH (5–40 ft), including section-control reductions. Sprayer sections retain their recorded widths (10–15 ft), totaling 90 ft across the complete boom. Dimensions are corrected for latitude in Web Mercator. This is an endpoint reconstruction assumption; no GPS offset correction is applied. The default application view now adds the turn-sweep reconstruction described above; the rectangle mode remains available for comparison. Both footprint geometry and statistical totals use original recorded swath widths.

Harvest area = distance (ft) × swath width (ft) ÷ 43,560. Yield and moisture are weighted by positive swath area. All 11,388 records have positive area and nonnegative yield and are included, including zero yields. Yield units follow Deere’s VRYIELDVOL (bu/ac); moisture correction and monitor calibration are not independently verified. No overlap removal, yield cleaning, or automatic economic-baseline substitution was applied.

Active records have applied rate >0 and positive distance × section width. Swath area (ac) = distance (ft) × width (ft) ÷ 43,560. Recorded volume (gal) = Σ(area × applied rate). Weighted mean = recorded volume ÷ active swath area.

Quantiles and histograms are area-weighted. No external boundary shapefile or overlap union is included in the source exports, so reported swath acres remain cumulative rather than unique ground coverage. For spatial yield partitioning, a fixed trial boundary is derived as the 2D convex hull around active 09 Oct application footprints (AppliedRate > 0, 143 vertices, 18.914 enclosed ground ac); harvest records are classified inside versus outside by swath footprint midpoint, and Before/After comparisons remain paused until a post-application harvest file is supplied. Map positions and satellite imagery use Web Mercator. Map elevations are converted from source feet to meters (× 0.3048); CSV exports retain source feet.

All timestamps use IsoTime converted to CST (UTC−6). Zero targets and apparent weather placeholders are not used to infer treatment quality or weather. Rates remain in source gal/ac; US versus imperial gallons is not independently verified.

The application is recorded in autumn 2026. Its intended crop and harvest season must be confirmed. No causal or statistical yield claim is supported yet.