A large-scale, multitask, multisensory dataset for climate-aware crop monitoring in the US from 2018–2022

The escalating challenges of climate change, extreme weather events, and increasing food demand impose a significant strain on global food production. To develop and apply sustainable agriculture practices, farmers and organizations require detailed, timely information about weather, crops, and yields. While efficient agricultural monitoring relies heavily on remote sensing, the existing literature suffers from a notable lack of comprehensive, large-scale crop monitoring datasets. This paper introduces CropClimateX, a novel database built by optimizing location sampling to substantially cover cultivated areas throughout the contiguous United States. The database comprises 15,500 small 12 × 12 km data cubes spanning 1,527 counties. Crucially, each data cube integrates a rich array of multi-source information, including multi-sensor imagery (Sentinel-1/2, Landsat-8, MODIS), weather and extreme events (Daymet, heat/cold waves, and drought monitor maps), and environmental features (soil and terrain characteristics). This comprehensive, integrated dataset is designed to support a wide range of agricultural monitoring tasks, providing a vital resource for advancing research in sustainable farming and crop modeling.