Luo et al. (2026) Spatio-temporal monitoring of agricultural drought in China based on downscaled soil moisture data

Identification

• Journal: Ecological Indicators
• Year: 2026
• Date: 2026-10-21
• Authors: Xiaoyu Luo, Mengmeng Cao, Yaoping Cui, Xiangjin Meng, Yan Zhou, Yibo Yan
• DOI: 10.1016/j.ecolind.2026.115708

Research Groups

• College of Geographical Sciences, Faculty of Geographical Science and Engineering, Henan University, Zhengzhou, China
• Key Laboratory of Geospatial Technology for the Middle and Lower Yellow River Regions, Ministry of Education, Henan University, Kaifeng, China
• Key Laboratory of Lake and Watershed Science for Water Security, Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences, Nanjing, China
• College of Global change and Earth System Science, Beijing Normal University, Beijing, China

Short Summary

This study developed a framework integrating gap-filling and machine learning downscaling to generate a seamless 0.05° monthly soil moisture dataset for China from 2003 to 2023. Using this dataset, agricultural drought was found to have intensified significantly across China during the period, characterized by a northward migration of the drought center and spatially heterogeneous aridification patterns.

Objective

• To develop an integrative framework for generating a seamless, high-resolution (0.05°) monthly soil moisture (SM) dataset for China from 2003 to 2023 by synergistically integrating a spatiotemporally adaptive gap-filling algorithm with a machine learning-based downscaling approach.
• To utilize the generated SM dataset and the Standardized Soil Moisture Index (SSI) to characterize the spatio-temporal evolution, interannual variability, and long-term trends of agricultural drought across China.

Study Configuration

• Spatial Scale: China (area of 9.6 × 10⁶ km²), with the generated soil moisture dataset at 0.05° spatial resolution. The study area was divided into nine major river basins.
• Temporal Scale: 2003–2023, with monthly temporal resolution for the generated soil moisture dataset and drought analysis.

Methodology and Data

• Models used:
  • Gap-filling: Spatiotemporally adaptive algorithm combining enhanced Savitzky-Golay filter, Geographically Weighted Regression (GWR) model, and weighted least-squares fusion.
  • Cross-sensor calibration: Pixel-wise linear regression correction.
  • Downscaling: Machine learning-based Random Forest (RF) algorithm, incorporating the Temperature-Vegetation Dryness Index (TVDI) as a predictor.
  • Drought monitoring: Standardized Soil Moisture Index (SSI).
  • Trend analysis: Non-parametric Sen's slope estimator and Mann-Kendall (M–K) test.
• Data sources:
  • Satellite Soil Moisture: Level-3 SM retrievals from AMSR-E (2003–2011), SMOS-IC V105 (2011–2012), and AMSR2 (2012–2023), all initially at 0.25° resolution.
  • In-situ Observations: Soil moisture measurements (0–10 cm depth, hourly, aggregated to monthly) from China's national meteorological stations and agrometeorological/ecological observation networks (2003–2023) for validation.
  • Auxiliary Remote Sensing: MODIS monthly products (Land Surface Temperature (LST), Leaf Area Index (LAI), Normalized Difference Vegetation Index (NDVI)) at 0.05° resolution; MODIS MCD12Q1 Collection 6 Land Cover (500 m, yearly); SRTM 1 km Digital Elevation Model (DEM) and derived slope (resampled to 0.05°).
  • Reanalysis: ECMWF meteorological parameters (2 m temperature, total evaporation, total precipitation, surface net solar radiation) at 0.1° spatial resolution, monthly.
  • Benchmark Datasets: Five reconstructed SM products (SM-I, SM-II, SM III, SM IV, ERA5) for accuracy comparison.
  • Drought Reference Data: Observation-based drought-affected areas compiled from daily spatial agricultural drought-monitoring data, China Agricultural Meteorological Disaster Dataset, annual 30 m cropland maps, government-reported crop drought-affected areas, and China Flood and Drought Disaster Prevention Bulletin.

Main Results

• Improved Soil Moisture Dataset Accuracy: The downscaled SM dataset provides finer spatial resolution (0.05°) and improved accuracy. It achieved a 2.3–43.4% reduction in unbiased Root Mean Square Error (ubRMSE) and a 1.2–52.7% increase in correlation coefficients relative to benchmark datasets across different land-cover types. The product maintained stable performance under both extremely wet (R² = 0.81, ubRMSE = 0.047 m³/m³) and extremely dry (R² = 0.77, ubRMSE = 0.053 m³/m³) conditions.
• Accurate Drought Identification: Drought characterization based on the downscaled SM dataset and SSI accurately identified the extent of agricultural drought, showing significant spatiotemporal consistency with observation-based agricultural disaster areas (R² = 0.88 nationwide, R² = 0.89 for cropland scale).
• Drought Intensification and Spatial Patterns: Agricultural drought intensified significantly across China during 2003–2023, with a net decrease of 0.26 in the annual mean SSI. Spatially, there was a northward migration of the drought center and heterogeneous aridification patterns, with decreasing severity from northwest to southeast and increasing from northeast to southwest. High-frequency drought zones (32–52% occurrence, reaching up to 75% in some areas) were predominantly clustered in ecologically vulnerable regions, particularly the agro-pastoral ecotone of northern China.
• Intra-annual Drought Dynamics: A bimodal seasonal pattern of drought emerged, with peaks in spring (March–May) and late autumn to early winter (October–December). Drought patterns expanded from southwest to northeast from January to June, followed by bidirectional propagation from the Yellow River-Huaihe River Basin (YRB-HRB) to northwestern and southeastern regions from June to December.

Contributions

• Developed a novel integrative framework for generating a seamless, high-accuracy (0.05° resolution) monthly soil moisture dataset for China (2003–2023), addressing limitations of existing products in terms of accuracy, spatial resolution, and temporal continuity.
• Provided high-accuracy data support for agricultural drought monitoring, demonstrating superior performance in characterizing drought-affected areas compared to existing benchmark products.
• Offered comprehensive scientific insights into the spatio-temporal evolution, interannual variability, and long-term trends of agricultural drought across China, including its intensification, northward migration, and heterogeneous patterns.
• Contributed crucial scientific support for national drought risk assessment, informing adaptive agricultural management and aiding in the development of regionally differentiated drought mitigation strategies for China's food security.

Funding

• National Natural Science Foundation of China (42501143)
• Henan Province Science and Technology Research Project (262102321023)
• Central Plains Youth Top Talent Project
• University Science and Technology Innovation Team in Henan Province (No. 25IRTSTHN004)
• Postgraduate Education Reform and Quality Improvement Project of Henan Province (No. YJS2023JC22)
• Xinyang Academy of Ecological Research Open Foundation (No.2023XYMS01)
• National Civil Space Infrastructure 13th Five-Year Plan for Common Application Support Platform of Terrestrial Observation Satellites
• Natural Science Foundation of Henan (252300420827)

Citation

@article{Luo2026Spatiotemporal,
  author = {Luo, Xiaoyu and Cao, Mengmeng and Cui, Yaoping and Meng, Xiangjin and Zhou, Yan and Yan, Yibo},
  title = {Spatio-temporal monitoring of agricultural drought in China based on downscaled soil moisture data},
  journal = {Ecological Indicators},
  year = {2026},
  doi = {10.1016/j.ecolind.2026.115708},
  url = {https://doi.org/10.1016/j.ecolind.2026.115708}
}