Liu et al. (2026) Multidimensional aspects of drought event evolution drive the spatiotemporal heterogeneity of vegetation photosynthetic responses

Identification

• Journal: International Journal of Applied Earth Observation and Geoinformation
• Year: 2026
• Date: 2026-03-27
• Authors: Xiangping Liu, Zhuowei HU, Mi Wang, Wenxing Hou, Tengxun Hu
• DOI: 10.1016/j.jag.2026.105253

Research Groups

• College of Resources Environment and Tourism, Capital Normal University, Beijing 100048, China

Short Summary

This study employs a three-dimensional clustering approach with daily soil moisture data (2000–2024) to identify and track 32 typical drought events across China, linking their evolutionary characteristics to vegetation photosynthetic response thresholds. It reveals that distinct drought types (High-Disturbance Migratory, Quasi-Stationary, Localized Outbreak) drive heterogeneous vegetation sensitivities, with varying environmental factors governing these responses.

Objective

• To accurately identify the full lifecycle of drought events from an event-based perspective.
• To quantify the multi-dimensional attributes of drought events and evaluate their differential impacts on vegetation photosynthetic responses.
• To disentangle the regulatory role of environmental conditions in shaping vegetation drought sensitivity.

Study Configuration

• Spatial Scale: China, with a spatial resolution of 0.05° × 0.05°. Identified drought events had a minimum impact area of 2 × 10^5 km^2. Root-zone soil moisture was considered for a depth of 0–100 cm.
• Temporal Scale: 2000–2024 (25 years). Data were processed at a daily temporal scale, with 4-day SIF observations linearly interpolated to daily. Identified drought events had a minimum duration of 60 days.

Methodology and Data

• Models used:
  • Extended Triple Collocation (ETC) method for evaluating soil moisture product performance.
  • Three-dimensional (3D) DBSCAN algorithm for identifying and clustering drought events in space and time.
  • K-means clustering for classifying drought events based on five diagnostic metrics (severity, propagation speed, propagation perturbation, tortuosity, and spatial autocorrelation).
  • XGBoost model for quantifying the relative contributions of environmental driving factors to vegetation sensitivity.
• Data sources:
  • Soil Moisture (SM): ERA5-Land (ECMWF), SMCI1.0 (National Tibetan Plateau Data Center), and GLEAM v4.2a. ERA5-Land was selected as the optimal product.
  • Climate data: ERA5-Land reanalysis dataset (ECMWF) for 2 m air temperature, 2 m dew-point temperature, relative humidity, soil temperature (ST), latent heat flux (LHF), net solar radiation at the surface (NSRS), daily maximum and minimum air temperature, precipitation (PRE), and total evapotranspiration (E). Vapor pressure deficit (VPD) and diurnal temperature range (TD) were derived.
  • Satellite data:
    • Solar-Induced Fluorescence (SIF) reconstructed from Orbiting Carbon Observatory-2 (OCO-2) observations.
    • Normalized Difference Vegetation Index (NDVI) from MODIS MCD43C4 product; kNDVI calculated.
    • Leaf Area Index (LAI) from MODIS MOD15A2H product.
    • Gross Primary Productivity (GPP) from MODIS MOD17A2HGF product.
    • Ecosystem-scale Water Use Efficiency (WUE) calculated as GPP/E.
  • Topographic information: Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM).
  • Vegetation rooting depth: Global rooting depth distribution dataset (Stocker et al., 2023).
  • Land Use and Land Cover (LULC): ESA CCI land cover dataset (ECMWF).
  • Soil type: Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences.

Main Results

• Summer is the predominant drought season in China, with events persisting significantly longer than in other seasons. Southern droughts primarily occur in winter and spring, while northern droughts concentrate in early summer and often extend into late autumn.
• Extreme drought events are mainly distributed across the North China-Northeast Plain, Southwest China, and the middle-lower Yangtze River Plain, exhibiting marked heterogeneity in process dynamics.
• Three distinct drought types were identified:
  • High-Disturbance Migratory (HDM): Characterized by high propagation speed (≥0.5 km/day), pronounced perturbation, and complex pathways, primarily in the North China Plain. These are the most frequent nationwide.
  • Quasi-Stationary (QS): Exhibits weak spatial propagation (≤0.4 km/day), moderate-to-high perturbation, significant spatial clustering, and moderate severity, distributed across Southwest and Northeast China, and the middle Yangtze.
  • Localized Outbreak (LO): Shows extremely low propagation speed (≤0.1 km/day), low-to-moderate perturbation, high clustering, and pronounced severity, mainly in Southwest China. These are less frequent but typically more intense and longer-lasting.
• Vegetation photosynthetic responses, as indicated by SIF, exhibited the earliest and strongest negative responses to drought, with optimal lag times to soil moisture deficits mostly within 1–30 days.
• Vegetation responses showed distinct patterns across drought types:
  • LO events: Induced the highest sensitivity to soil moisture, but displayed a declining interannual trend under extreme low-threshold conditions (soil moisture percentile q ≤10).
  • QS events: Maintained relatively stable responses with localized amplification under severe droughts.
  • HDM events: Showed the lowest overall sensitivity, but their mean vegetation–SM coherence gradually increased over time, while coherence under extreme conditions weakened.
• Mechanistic analyses using XGBoost (R² = 0.718) revealed type-specific driving factors:
  • LO droughts: Net surface radiation (NSRS) and land-use type were the strongest modulators of vegetation responses.
  • QS droughts: Topographic factors (DEM) and vegetation root depth were dominant, with precipitation (PRE) and vapor pressure deficit (VPD) also influential.
  • HDM droughts: Energy balance components (evapotranspiration (E) and latent heat flux (LHF)) played stronger roles, with topographic modulation being comparatively minor.

Contributions

• Introduces a novel three-dimensional clustering approach to identify and track the full lifecycle of drought events, shifting from static grid-based analyses to dynamic event-based characterizations.
• Develops a process-oriented classification of drought events into three distinct types (High-Disturbance Migratory, Quasi-Stationary, Localized Outbreak) based on multi-dimensional diagnostic metrics.
• Quantifies the differential impacts of these drought types on vegetation photosynthetic responses using Solar-Induced Fluorescence (SIF) and coincidence analysis, revealing non-linear and type-specific sensitivities.
• Disentangles the hierarchical influences of climatic, environmental, and plant physiological drivers on vegetation drought sensitivity for each drought type, providing mechanistic insights into drought-vegetation coupling.
• Offers a more robust and operational framework for drought risk assessment and water-resource management by considering the complete spatiotemporal evolution and diverse ecological impacts of drought events.

Funding

• National Key Research and Development Program of China [grant numbers 2023YFF1303703].

Citation

@article{Liu2026Multidimensional,
  author = {Liu, Xiangping and HU, Zhuowei and Wang, Mi and Hou, Wenxing and Hu, Tengxun},
  title = {Multidimensional aspects of drought event evolution drive the spatiotemporal heterogeneity of vegetation photosynthetic responses},
  journal = {International Journal of Applied Earth Observation and Geoinformation},
  year = {2026},
  doi = {10.1016/j.jag.2026.105253},
  url = {https://doi.org/10.1016/j.jag.2026.105253}
}