In-Season and Predictive Cropping
If you want to have a competitive edge, you need the most up-to-date information. That’s why we provide crop data for the current season, as well as reliable predictions for next year.
Use this data in strategy, marketing and sales to make high-level strategic decisions, as well as on-the-ground tactical decisions.
2011-2018 Crop Data
- Over seven years of historic data
- Crop rotation, crop mix, acres farmed, Gross Farm Income and more
2019 Crop Data
- In-season crop identification covering corn, soybean, wheat and cotton in 20 states
- Over 200 million acres and 90 percent coverage of U.S. output of these crops
- Satellite-based remote sensing using machine learning with ground truthing
- Available at field and grower level to link grower crops to an overall market view
2020 Crop Data (coming soon!)
- Leverages eight year field-level crop rotation, crop yields, commodity prices and weather data to project field-level planting decisions
- Employs multivariate regression analysis and statistical models to analyze and project planting decisions
- Covers corn, soybean, wheat and cotton in 20 states – over 200 million acres
We analyzed our in-season crop data against our ground truthing, revealing the following accuracy figures for the 2019 season.
Farm Market iD’s in-season and predictive data covers four crops in 20 states, approximately 70 percent of all planted acres in a given season.
Total acres: 78,831,074
Total acres: 87,436,853
Total acres: 26,792,700
Total acres: 9,702,277
Where the Data Comes From
We gather this data from three primary sources, in addition to various other sources. The primary sources are:
- Historic and In-Season Satellite Imagery from Landsat and Sentinel Sensors
- Historic and In-Season Weather Data Observations
- Historic planting information derived from the NASS Cropland Data Layers
Imagery collected in-season is the primary driver of crop identification. Historic and in-season weather information is used to predict crop growth stage and to fine-tune identification. Data drive from the NASS datasets are used to create spectral profiles unique to crops of interests.
Information from satellite imagery is used to distinguish crops early in the growing season. In addition, the crop prediction model is calibrated using ground reference data provided by trusted partners.
We then implement a statistical machine learning model to capture the predominant crop rotation patterns within each state and project those patterns into the future. This model takes into account regional variations in suitability for different crops, commodity prices at the time preliminary planting decisions are made, data from the USDA on planted areas, and weather data that can affect decisions closer to planting time.
With this historically-based prediction as a starting point, satellite imagery from medium resolution sensors is processed chronologically to update the machine learning classification system’s estimate of crop type.
Imagery from previous seasons, pertinent weather data such as growing degree days, soil temperature, and precipitation are used to create the baseline model against which new satellite imagery is compared.