Individual participating farmers, coordinated via Teknologisk Institut (DTI)
Farm-Supplied Operational Data (Yield Logs, Machine Data, Management Records)
Real, on-the-ground farm data — actual harvest yields, planting dates, fertiliser and pesticide use — voluntarily shared by farmers under a transparent, GDPR-compliant agreement with clear benefits returned to them.
Overview
Why it matters in GATE
Farmer-shared data is GATE's ground truth — real harvest yields, planting dates, and input-use records that no institutional dataset can replace, exchanged transparently for a benchmarking report back to the farmer.
Supports
- Training and validating the Land-CRAFT yield model.
- Farm-level benchmarking reports for participating farmers.
- Grounding all derived GATE insights in real, on-farm outcomes.
- Building trust through GDPR-compliant, benefit-sharing data agreements.
What data is available
Actual harvest yields (t/ha, kg/plot), moisture content, planting dates, harvest dates, crop variety, fertilizer application rates, and plant protection treatments.
Combine Yield Logs & Sensor Maps
Combine-harvester yield maps recording actual spatial yield variations, grain moisture, wet weight, and dry weight.
Farm Management & Application Records
Planting dates, crop variety, actual fertilizer application rates (N, P, K), pesticide logs, and irrigation volumes.
Access and technical profile
GATE status
Secure cloud ingestion environment managed by DTI under strict access control and anonymisation procedures.
Format
Coverage
| Spatial reference | Sub-field machine-sensor level (point tracks), Field-parcel aggregates |
|---|---|
| Temporal reference | Per harvest season / event-based machine logging |
| Update frequency | annually or depending on aggreement |
Linking & Use Cases in GATE
Linking keys
- Field parcel ID (linked to SGAV Marknummer)
- WGS84 / EPSG:25832 coordinates
- Harvest year
Typical GATE joins
- Join with SGAV Marker to ground-truth reported crop types with actual measured harvest volumes.
- Link combine yield tracks with AU 3D soil texture rasters and DMI climate grids to train Land-CRAFT machine learning algorithms.
Related projects
- Land-CRAFT
- GATE Pilot Farms
Use cases in GATE
Ground-truth foundation for the Land-CRAFT field-level crop yield dataset and farmer benchmarking reports.
- Crop yield dataset at field level (ML training and validation).
- Farm-level benchmarking reports for participating farmers.
Quality, limits and use conditions
Method
Raw telematics and combine-harvester yield logs are parsed, algorithmically filtered for sensor calibration drift and headland turns, harmonised to EPSG:25832, and linked to official field identifiers (SGAV Marknummer). Subsequently, data scientists and agricultural specialists perform manual inspection and domain-specific quality checks to identify, flag, and remove anomalies, out-of-scope entries, or unusable logs before inclusion in the pipeline.
Quality notes
Serves as essential, on-the-ground empirical ground truth for real-world farm outcomes rather than planned norms. Please note, data originates from individual self-reported farm contributions and varying on-combine sensor calibrations, data consistency varies across holdings and relies on domain-expert curation to ensure analytical reliability.
Limitations
- High data heterogeneity across machine manufacturer proprietary formats requiring format-specific translation parsers.
- Combine yield sensor accuracy depends on proper on-farm physical calibration by the machine operator.
- Strict data privacy and GDPR constraints prevent public open-access redistribution of raw un-anonymized farm records.
Licence and terms
Restricted access governed by individual GDPR-compliant benefit-sharing agreements with participating farmers.
Licence: Restricted / Private Farmer Data Sharing Agreement