Chen et al. (2026) High spatiotemporal resolution monitoring of crop water stress across the contiguous United States using Harmonized Landsat and Sentinel-2 data

Identification

• Journal: Agricultural Water Management
• Year: 2026
• Date: 2026-01-06
• Authors: Na Chen, Yanlei Feng, Na Wang, Jevan Yu, Mohammad Reza Alizadeh, Yifeng Cui, Ning Ye, Wenzhe Jiao, Joshua B. Fisher, César Terrer
• DOI: 10.1016/j.agwat.2025.110094

Research Groups

• Department of Civil and Environmental Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA
• School of Agriculture, Food and Ecosystem Sciences (SAFES), Faculty of Science, and Faculty of Engineering and Information Technology (IE-FEIT), University of Melbourne, Melbourne, Australia
• School of Forestry, University of Canterbury, Christchurch, New Zealand
• Department of Ecology and Conservation Biology, Texas A&M University, College Station, TX, USA
• Schmid College of Science and Technology, Chapman University, Orange, CA, USA

Short Summary

This study explored the potential of Harmonized Landsat and Sentinel-2 (HLS) data for near-real-time crop water stress monitoring across the contiguous United States (CONUS). It demonstrated that HLS data can provide timely alerts of crop water stress with an overall accuracy of 74.0% and a mean detection lag of approximately 9 days.

Objective

• To investigate the feasibility of using Harmonized Landsat and Sentinel-2 (HLS) data for detecting the time and location of crop water stress across the contiguous United States (CONUS).
• To quantify the accuracy of crop water stress detected from HLS data.

Study Configuration

• Spatial Scale: Contiguous United States (CONUS) croplands (approximately 147.6 Mha), utilizing 923 HLS satellite tiles at 30 m spatial resolution.
• Temporal Scale: Historical period from 2016 to 2019, monitoring period for 2020 (a drought year). HLS data provides observations every 2–3 days. Crop water stress events were monitored from March to August 2020.

Methodology and Data

• Models used: Breaks For Additive Season and Trend Monitor (BFAST monitor) for breakpoint detection in Normalized Difference Moisture Index (NDMI) time series, and Random Forest classifier for post-processing and mapping crop water stress.
• Data sources:
  • Harmonized Landsat and Sentinel-2 (HLS) Version 2.0 surface reflectance data (L30 and S30 products) at 30 m spatial resolution.
  • Cropland Data Layers (CDL data) for 2019 and 2020 (USDA National Agricultural Statistics Service) for crop type information.
  • Land Surface Phenology (LSP) metrics (Onset Greenness Increase - OGI, and Onset Greenness Minimum - OGMn) derived from HLS data.
  • Standardized Precipitation Index (SPI) derived from the Gridded Surface Meteorological dataset (Abatzoglou, 2013) for validation.

Main Results

• HLS data enabled near-real-time alerts of crop water stress with an overall spatial accuracy of 74.0% and a kappa coefficient of 0.48.
• Approximately 12.3 Mha of water-stressed crops were mapped across the CONUS from March to August 2020, identifying around 3.8 million crop water stress events.
• Small-scale water stress events (≤0.5 ha) were the most frequent (41.8% of events) but covered only 4.4% of the total affected area.
• Major water stress events (≥5 ha) were the least frequent (10.0% of events) but dominated in terms of area, affecting 74.2% of the total mapped extent.
• The mean time lag (MTL) for detected crop water stress across the CONUS using HLS data was approximately 9 days.
• The most significant variables for detecting crop water stress were the "presence of break" and "magnitude" derived from BFAST Monitor, followed by phenological metrics and crop types.

Contributions

• This study is the first to explore the potential of the recently released Harmonized Landsat and Sentinel-2 Version 2.0 dataset for high spatiotemporal resolution monitoring of crop water stress across a national scale (CONUS).
• It demonstrates the feasibility of using HLS data for timely and accurate crop water stress monitoring, providing spatially and temporally explicit information crucial for precision agriculture.
• The findings highlight the importance of medium to high spatial resolution satellite imagery for revealing fine-scale details of crop water stress, which can be overlooked by coarser resolution data.
• The research supports evidence-based policy decisions in agricultural and water resource management, including the design of crop insurance programs and the promotion of sustainable farming practices.

Funding

• Wenzhe Jiao (Funding acquisition)

Citation

@article{Chen2026High,
  author = {Chen, Na and Feng, Yanlei and Wang, Na and Yu, Jevan and Alizadeh, Mohammad Reza and Cui, Yifeng and Ye, Ning and Jiao, Wenzhe and Fisher, Joshua B. and Terrer, César},
  title = {High spatiotemporal resolution monitoring of crop water stress across the contiguous United States using Harmonized Landsat and Sentinel-2 data},
  journal = {Agricultural Water Management},
  year = {2026},
  doi = {10.1016/j.agwat.2025.110094},
  url = {https://doi.org/10.1016/j.agwat.2025.110094}
}