CROPSIGHT-US: Crop Type Ground Truth from Street View and Satellite Imagery, 2013-2023

Contiguous United States crop type ground truth data for 124,419 cropland fields from 2013 to 2023. The dataset covers 17 major crop types across 294 Agricultural Statistics Districts and includes uncertainty metrics. It was created by Zhijie Zhou and is hosted on figshare.

Use Cases

• Training crop type classification models based on Sentinel-2 satellite imagery and street view data.
• Analyzing spatial and temporal patterns of crop cultivation across the contiguous United States.
• Assessing prediction uncertainty in agricultural mapping using provided entropy and variance metrics.
• Validating national-scale land use and agricultural survey data.

Strengths

• Contains 124,419 cropland field records with associated geospatial polygons.
• Covers 17 distinct crop types and spans 294 Agricultural Statistics Districts.
• Includes explicit uncertainty metrics (entropy, variance, confidence) for each prediction.
• Provides a decade of temporal coverage from 2013 to 2023.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic or temporal bias inherent to the street view and satellite imagery sources.

Provenance

Source

Zhijie Zhou, with support from the Illinois Computes project at the University of Illinois Urbana-Champaign.

Collection Method

Generated using object-based methods combining Google Street View imagery and Sentinel-2 satellite data, with uncertainty quantification via Monte Carlo simulations.

Time Range

2013 to 2023

Freshness

Last updated 2026-04-22 06:55:44; freshness should be verified.

Geography

Contiguous United States (CONUS)