Research on the Method of Extracting Water Body Information in Central Asia Based on Google Earth Engine

K. I. Chang, Wei Yue, Hongzhi Wang, Kaijun Tan, Xinyu Liu, Xiaoyi Cao, Wenqian ChenWater 2025 · DOI

TL;DR

This study leverages Google Earth Engine and Landsat data to monitor water body changes in Central Asia from 2000-2019. They found a significant 11.25% decline in total water area, driven by rising temperatures and decreasing precipitation, with critical implications for regional water security.

What Problem Does This Paper Address?

Central Asia is the world's largest arid region, characterized by an uneven distribution of water resources and a high susceptibility to climate change. Understanding the dynamics of these water bodies (lakes, rivers, glaciers) is crucial for regional development, water resource management, and addressing potential conflicts. Previous studies often focused on specific water bodies like lakes or used traditional, data-intensive processing methods. This paper aims to provide a scalable, efficient, and accurate method for large-area, long-term water body extraction in this arid environment, which is vital for assessing climate change impacts and informing adaptive strategies.

Methods

The researchers compared three water extraction methods:

  • Modified Normalized Difference Water Index (MNDWI): An enhancement of the NDWI, designed to better delineate water boundaries and reduce background interference, especially in urban areas.
  • OTSU Algorithm: An automatic thresholding technique that optimizes image segmentation by maximizing the variance between background and foreground classes.
  • Random Forest (RF): A machine learning algorithm known for its robustness, ability to handle high-dimensional data, and strong performance in classification tasks.

The study utilized the Google Earth Engine (GEE) platform for its cloud-based processing capabilities, enabling large-scale analysis of time-series satellite imagery. After comparing the performance of the three methods, the Random Forest algorithm was selected as the optimal tool due to its superior accuracy and detail resolution.

Datasets

  • Landsat 7 ETM+ surface reflectance data: Used for water body extraction, providing a 30m spatial resolution and 16-day temporal resolution from 2000-2019.
  • GLDAS (Global Land Data Assimilation System) temperature dataset: Used for analyzing temperature trends.
  • CHIRPS (Climate Hazards Group InfraRed Precipitation with Station data) precipitation dataset: Used for analyzing precipitation trends.

Key Insights and Ideas

  • Random Forest Dominance: The Random Forest algorithm consistently outperformed MNDWI and OTSU in accurately identifying water bodies, especially in complex terrains and for smaller water features. This highlights the power of machine learning for remote sensing applications.
  • Climate as a Primary Driver: The strong negative correlation between rising temperatures and water body area, and the positive correlation with decreasing precipitation, clearly links climate change to water scarcity in Central Asia. This underscores the region's vulnerability.
  • Regional Disparities: While the overall trend is a decline in water, the impact is not uniform. The Aral Sea and southern Kyrgyzstan's glaciers show dramatic reductions, indicating specific hotspots of concern.
  • MNDWI's "Inflated" Accuracy: The study points out that MNDWI's high accuracy metrics might be artificially inflated due to subjective manual interpretation of the original imagery, especially in areas with salt marshes or dry riverbeds. This is a crucial caveat for interpreting results from simpler index-based methods.

Key Results

  • Overall Water Decline: Central Asia experienced an 11.25% decrease in total water area between 2000 and 2019.
  • Aral Sea Shrinkage: The Aral Sea's water area decreased by a dramatic 72.13% from 2000 to 2019. The East Aral Sea, in particular, saw a 93.3% reduction from 2003 to 2019.
  • Glacier Retreat: Glaciers in southern Kyrgyzstan decreased by 39.23% between 2000 and 2019.
  • Correlation Coefficients:
    • Water body area vs. mean annual air temperature: -0.5583
    • Water body area vs. mean annual precipitation: 0.6872
  • Algorithm Accuracy:
    • Random Forest: 95.21% accuracy, 90.29% Kappa
    • MNDWI: 97.14% accuracy, 94.21% Kappa (noted as potentially inflated)
    • OTSU: 90.80% accuracy, 83.48% Kappa

Limitations and Future Work

  • Data Quality: While Landsat 7 data is generally stable, cloud cover, shadows, and non-water features like salt marshes can still introduce errors in water body extraction.
  • Temporal Scope: The study suggests expanding the analysis to include data prior to 1984 to provide a more comprehensive view of long-term trends.
  • Altitudinal Influences: The paper acknowledges that neglecting altitude-based zoning might lead to generalized descriptions and obscure unique hydrological patterns. Future work could incorporate altitudinal stratification.
  • Future Directions: The authors advocate for continued use of the RF-GEE framework for large-scale, long-term hydrological monitoring in arid regions. Future research should focus on expanding the temporal scope and integrating more data sources to enhance accuracy and comprehensiveness.

Summary updated

Abstract

This study evaluates water body changes in Central Asia (2000–2019) using Landsat 7 data on Google Earth Engine, comparing the Modified Normalized Difference Water Index (MNDWI), OTSU algorithm, and random forest (RF). The random forest algorithm demonstrated the best overall performance in water body extraction and was selected as the analysis tool. The results reveal a significant 11.25% decline in Central Asia’s total water area over two decades, with the Aral Sea shrinking by 72.13% (2000–2019) and southern Kyrgyzstan’s glaciers decreasing by 39.23%. Pearson correlations indicate strong links between water loss and rising temperatures (−0.5583) and declining precipitation (0.6872). Seasonal fluctuations and permanent degradation (e.g., dry riverbeds) highlight climate-driven vulnerabilities, exacerbated by anthropogenic impacts. These trends threaten regional water security and ecological stability, underscoring the urgent need for adaptive resource management. The RF-GEE framework proves effective for large-scale, long-term hydrological monitoring in arid regions, offering critical insights for climate resilience strategies.
BibTeX
@inproceedings{chang2025research,
  author = {K. I. Chang and Wei Yue and Hongzhi Wang and Kaijun Tan and Xinyu Liu and Xiaoyi Cao and Wenqian Chen},
  title = {{Research on the Method of Extracting Water Body Information in Central Asia Based on Google Earth Engine}},
  booktitle = {Water},
  year = {2025},
  doi = {10.3390/w17060804},
}
Figure