Crop rotation practice plays a role in crop disease and persistent fungal growth over seasons. Several modeling groups in the organization were interested in using previous crop types as features in their model for predicting disease risk.
Geospatial data pipeline to summarize previous crops for a given collection of boundaries, CSV data stored in AWS S3 with the predominant and second most prevalent extracted from USDA Cropland Data Layer in year minus one and year minus two as well as the proportion of the field covered by the two crop types.
Descartes Labs (DL) Python APIs for retrieving imagery data for requested collection of farm fields; numpy for determining the predominant and second most prevalent crop in each year and calculating the cover proportion of each crop type in each field
Developed a geospatial data pipeline with DL APIs to provide previous crop data derived from the USDA Cropland Data Layer (CDL) for large collections of geometries as large as five hundred thousand unique fields. Collaborated with several modeling teams to tailor the layer specifically to their unique data needs.
