Shi et al. (2026) Spatiotemporal drought variability in Gansu Province based on reconstructed land surface temperature

Identification

• Journal: Natural Hazards
• Year: 2026
• Date: 2026-02-01
• Authors: Yaya Shi, Xiangxiang Hu, Shuailing Liu
• DOI: 10.1007/s11069-026-07981-6

Research Groups

• School of Resources and Environmental Engineering, Tianshui Normal University, Tianshui, China

Short Summary

This study developed a novel framework integrating reconstructed Land Surface Temperature (LST)-derived Temperature Vegetation Drought Index (TVDI) with the Standardized Precipitation Index (SPI) to analyze spatiotemporal drought variability in Gansu Province from 2003 to 2022, revealing a "south mild, northwest severe" drought pattern with overall intensification since 2008.

Objective

• Reconstruct MODIS LST data using spatiotemporal interpolation to mitigate cloud-induced missing values.
• Reveal the spatiotemporal variations of agricultural drought (via TVDI) and meteorological drought (via SPI) in Gansu from 2003 to 2022.
• Quantify the relationship between drought indices and key climatic factors to clarify drought drivers in arid and semi-arid regions.

Study Configuration

• Spatial Scale: Gansu Province, China, covering the transition zone between the Loess Plateau, Qinghai-Tibet Plateau, and Inner Mongolia Plateau, characterized by diverse topography (mountains, plateaus, deserts, Gobi) and significant elevation differences.
• Temporal Scale: 20 years (2003–2022).

Methodology and Data

• Models used:
  • LST reconstruction: Harmonic ANalysis of Time Series (HANTS) algorithm, Z-score method (outlier detection), neighborhood averaging (outlier imputation).
  • Agricultural drought: Temperature Vegetation Drought Index (TVDI).
  • Meteorological drought: Standardized Precipitation Index (SPI) calculated using the non-parametric Standardized Drought Analysis Toolbox (SDAT).
  • Trend analysis: Sen slope and Mann-Kendall (M-K) statistical test.
• Data sources:
  • MODIS LST products: MYD11A2 (8-day composite, 1 km spatial resolution) from LP DAAC.
  • MODIS NDVI products: MOD13A1 (16-day composite, 500 m spatial resolution) from LAADS DAAC.
  • Monthly precipitation: ChinaMet dataset (1 km spatial resolution) from National Cryosphere Desert Data Center.
  • Annual air temperature and precipitation: Meteorological stations in Gansu Province from the China Meteorological Data website.

Main Results

• LST reconstruction achieved high accuracy (Mean Error < 0.2 °C, Mean Absolute Error < 1.5 °C, Pearson Correlation Coefficient > 0.98) with a final missing data rate below 0.05%.
• Drought in Gansu Province exhibits a "south mild, northwest severe" spatial pattern, with southern regions experiencing no or slight drought, while northwestern regions (e.g., central Jiuquan, northern Wuwei) face severe drought.
• Temporally, Gansu's overall drought situation has worsened since 2008, with the TVDI increasing at a rate of 0.0075 per decade from 2003 to 2022.
• Seasonal drought is most severe in northern Gansu during spring (41% of area under severe/extreme drought) and autumn (38%).
• Inter-annual trends show significant drought improvement in southeastern Gansu (e.g., Longnan, Tianshui) and increasing drought in northern Gansu (e.g., Jiayuguan, Zhangye, Wuwei, Baiyin).
• Drought (TVDI) shows a moderate positive correlation with mean annual air temperature (r = 0.52, p < 0.05) and a weakly negative correlation with annual precipitation (r = -0.11, p < 0.1).
• The spatial correlation between TVDI and precipitation is moderately negative (r = -0.53, p < 0.01), indicating precipitation's primary control on agricultural drought, though underlying surface factors weaken this correlation.
• Strong correlations between agricultural drought (TVDI) and meteorological drought (SPI) are concentrated in the central Loess Plateau and eastern Hexi Corridor.
• Weak or no correlations between TVDI and SPI are found in southeastern Gansu (due to buffering by forests/grasslands) and western Hexi Corridor (due to irrigation decoupling).
• Divergent drought patterns (e.g., agricultural drought intensification despite meteorological drought mitigation) are observed in central Jiuquan, Lanzhou, Baiyin, Dingxi, and Gannan, driven by complex interactions of climate, soil, and human activities.

Contributions

• Developed a novel framework integrating reconstructed MODIS LST-derived TVDI with SPI for comprehensive drought monitoring.
• Addressed critical challenges of cloud-induced LST data gaps through a multi-step reconstruction method (HANTS, Z-score, neighborhood interpolation).
• Provided a regionally tailored framework for spatiotemporal drought analysis in arid and semi-arid regions, generalizable to similar ecologically fragile areas.
• Quantified the complex relationships between agricultural and meteorological drought and their climatic drivers, highlighting areas of strong coupling and divergence.

Funding

• National Natural Science Foundation of China (42361020)
• Gansu Province Science Foundation for Youths (23JRRE727)
• Innovation Fund Program of Tianshui Normal University (CXJ2022-02)

Citation

@article{Shi2026Spatiotemporal,
  author = {Shi, Yaya and Hu, Xiangxiang and Liu, Shuailing},
  title = {Spatiotemporal drought variability in Gansu Province based on reconstructed land surface temperature},
  journal = {Natural Hazards},
  year = {2026},
  doi = {10.1007/s11069-026-07981-6},
  url = {https://doi.org/10.1007/s11069-026-07981-6}
}