Cen et al. (2025) Improving Remote Sensing Ecological Assessment in Arid Regions: Dual-Index Framework for Capturing Heterogeneous Environmental Dynamics in the Tarim Basin

Identification

Journal: Remote Sensing
Year: 2025
Date: 2025-10-22
Authors: Yuxin Cen, Li He, Zhengwei He, Fang Luo, Yang Zhao, Jie Gan, Wenqian Bai, Xin Chen
DOI: 10.3390/rs17213511

Research Groups

State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu 610059, China
College of Geography and Planning, Chengdu University of Technology, Chengdu 610059, China
College of Geophysical, Chengdu University of Technology, Chengdu 610059, China

Short Summary

This study introduces ARSEI and CoRSEI to improve ecological assessment in arid regions, demonstrating ARSEI's enhanced sensitivity to desert dynamics and CoRSEI's ability to capture heterogeneous environmental changes and long-term trends in the Tarim Basin from 2000 to 2023. The findings highlight the importance of differentiated ecological modeling for targeted ecosystem management in hyper-arid environments.

Objective

Develop an optimized Remote Sensing Ecological Index (RSEI) for desert areas by integrating surface albedo to form the Arid-region Remote Sensing Ecological Index (ARSEI).
Construct a zonally adaptive Composite Remote Sensing Ecological Index (CoRSEI) to assess ecological quality in both desert and non-desert regions of the Tarim River Basin and reveal spatial ecological patterns.
Use multivariate statistical analyses and machine learning interpretability methods to investigate region-specific environmental drivers and their influence mechanisms.

Study Configuration

Spatial Scale: Tarim River Basin, Xinjiang Uyghur Autonomous Region, China (75–93°E, 35–43°N). Data resolutions primarily 500 meters and 1 kilometer, resampled to 500 meters.
Temporal Scale: 24 years (2000–2023).

Methodology and Data

Models used:
- Remote Sensing Ecological Index (RSEI)
- Arid-region Remote Sensing Ecological Index (ARSEI)
- Composite Remote Sensing Ecological Index (CoRSEI)
- Principal Component Analysis (PCA)
- Two-Dimensional Kernel Density Estimation (2D-KDE)
- Theil–Sen median trend estimator
- Mann–Kendall test
- Hurst exponent (rescaled range (R/S) analysis)
- Coupling Degree and Synchronization analysis
- Spearman rank correlation analysis
- Partial Least Squares Regression (PLSR) and Variable Importance in Projection (VIP)
- Random Forest model for Partial Dependence Plots (PDP)
Data sources:
- Satellite:
  - MODIS MOD09A1 (Version 6.1): 500 m surface reflectance (Bands B1–B7) for NDVI, Wetness, NDBSI, Albedo.
  - MODIS MOD11A2 (Version 6.1): 1 km daytime Land Surface Temperature (LST).
  - MODIS/Terra Thermal Anomalies/Fire Daily L3 Global 1 km SIN Grid V006 (MOD14A1): Fire data.
  - MODIS Terra MOD17A3HGF version 6: Net Primary Productivity (NPP).
- Reanalysis/Gridded:
  - TerraClimate: Monthly precipitation and potential evapotranspiration (PET) for SPEI6 (~4 km).
  - Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) daily dataset (version 2.0): Daily precipitation (~5.6 km) for Extreme Rainfall.
  - ERA5-Land hourly reanalysis dataset: Evapotranspiration (ET, ~9 km) and monthly surface soil moisture (SM, ~11 km).
- Land Cover:
  - China Land Cover Dataset (CLCD): Land cover classification (30 m).

Main Results

Index Performance: ARSEI (average PCA1 = 78.62%) showed enhanced sensitivity to albedo (loading = 0.1199) and NDVI (Spearman ρ = 0.49) in arid environments, and was more strongly correlated with precipitation (ρ = 0.62) than RSEI. CoRSEI exhibited high internal consistency and spatial adaptability, with mean values ranging from 0.45 to 0.56.
Spatiotemporal Dynamics: The annual mean CoRSEI ranged from 0.45 to 0.56, peaking in 2001. The proportion of desert area gradually declined from 41.96% in 2000 to 40.55% in 2023, indicating a slow contraction.
Ecological Trends: Most of the study area (68.96%) remained stable from 2000 to 2023, with 1.97% experiencing significant degradation and 1.98% significant improvement. Hurst exponent projections indicated that 70.13% of the area showed uncertain future trends.
Environmental Drivers:
- Desert regions: Evapotranspiration (coupling degree C = 0.815), precipitation (C = 0.736), and soil moisture (C = 0.713) were the dominant positive drivers for ARSEI. Wetness (WET) showed negative effects, possibly due to salinization.
- Non-desert regions: Soil moisture (C = 0.737) and evapotranspiration (C = 0.725) were strongly associated with RSEI. NDVI showed a negative correlation, potentially due to "greening illusion" from artificial greening. Fire density and land surface temperature also exerted significant influences.

Contributions

Developed ARSEI, an optimized Remote Sensing Ecological Index for desert areas by replacing NDBSI with surface albedo, enhancing sensitivity to vegetation and albedo in hyper-arid environments.
Introduced CoRSEI, a zonally adaptive composite index that integrates ARSEI for desert regions and RSEI for non-desert regions, enabling comprehensive ecological quality assessment across heterogeneous landscapes.
Provided a spatially differentiated and driver-sensitive methodology for ecological monitoring in arid regions, addressing the limitations of traditional indices in diverse environments.
Identified region-specific environmental drivers and their influence mechanisms in desert and non-desert areas using multivariate statistical analyses and machine learning interpretability methods.
Offers robust scientific and technical support for targeted ecosystem management, conservation, and land degradation neutrality efforts in ecologically vulnerable arid regions.

Funding

Third Xinjiang Comprehensive Scientific Expedition Project (Grant No. 2023xjkk0103)
National Natural Science Foundation of China (Grant No. 42301456)
State Key Laboratory of Geohazard Prevention and Geoenvironment Protection Independent Research Project (Grant No. SKLGP2022Z017)

Citation

@article{Cen2025Improving,
  author = {Cen, Yuxin and He, Li and He, Zhengwei and Luo, Fang and Zhao, Yang and Gan, Jie and Bai, Wenqian and Chen, Xin},
  title = {Improving Remote Sensing Ecological Assessment in Arid Regions: Dual-Index Framework for Capturing Heterogeneous Environmental Dynamics in the Tarim Basin},
  journal = {Remote Sensing},
  year = {2025},
  doi = {10.3390/rs17213511},
  url = {https://doi.org/10.3390/rs17213511}
}

Original Source: https://doi.org/10.3390/rs17213511