Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Water Mapping and Change Detection Using Time Series Derived from the Continuous Monitoring of Land Disturbance Algorithm

Huong Pham, Samuel Cheng, Tao Hu, Chengbin Deng

TL;DR

The paper tackles the problem of monitoring surface water dynamics using satellite time series by evaluating the Continuous Monitoring of Land Disturbance (COLD) algorithm for water frequency estimation, water-body segmentation, and breakpoint-change detection. It leverages Landsat time series and COLD-derived coefficients to train regression and classification models, achieving a water-mapping accuracy around $0.90$ and a breakpoint-change detection accuracy of about $0.81$, with NMSE around $0.43$. The study demonstrates that COLD-derived data can robustly delineate water bodies and reveal water-trend changes across regions despite disturbances and temporal gaps, offering a scalable approach for environmental monitoring. The findings highlight the practical value of integrating COLD with traditional water indices to support water resource management and conservation in the face of land-disturbance events.

Abstract

Given the growing environmental challenges, accurate monitoring and prediction of changes in water bodies are essential for sustainable management and conservation. The Continuous Monitoring of Land Disturbance (COLD) algorithm provides a valuable tool for real-time analysis of land changes, such as deforestation, urban expansion, agricultural activities, and natural disasters. This capability enables timely interventions and more informed decision-making. This paper assesses the effectiveness of the algorithm to estimate water bodies and track pixel-level water trends over time. Our findings indicate that COLD-derived data can reliably estimate estimate water frequency during stable periods and delineate water bodies. Furthermore, it enables the evaluation of trends in water areas after disturbances, allowing for the determination of whether water frequency increases, decreases, or remains constant.

Water Mapping and Change Detection Using Time Series Derived from the Continuous Monitoring of Land Disturbance Algorithm

TL;DR

The paper tackles the problem of monitoring surface water dynamics using satellite time series by evaluating the Continuous Monitoring of Land Disturbance (COLD) algorithm for water frequency estimation, water-body segmentation, and breakpoint-change detection. It leverages Landsat time series and COLD-derived coefficients to train regression and classification models, achieving a water-mapping accuracy around and a breakpoint-change detection accuracy of about , with NMSE around . The study demonstrates that COLD-derived data can robustly delineate water bodies and reveal water-trend changes across regions despite disturbances and temporal gaps, offering a scalable approach for environmental monitoring. The findings highlight the practical value of integrating COLD with traditional water indices to support water resource management and conservation in the face of land-disturbance events.

Abstract

Given the growing environmental challenges, accurate monitoring and prediction of changes in water bodies are essential for sustainable management and conservation. The Continuous Monitoring of Land Disturbance (COLD) algorithm provides a valuable tool for real-time analysis of land changes, such as deforestation, urban expansion, agricultural activities, and natural disasters. This capability enables timely interventions and more informed decision-making. This paper assesses the effectiveness of the algorithm to estimate water bodies and track pixel-level water trends over time. Our findings indicate that COLD-derived data can reliably estimate estimate water frequency during stable periods and delineate water bodies. Furthermore, it enables the evaluation of trends in water areas after disturbances, allowing for the determination of whether water frequency increases, decreases, or remains constant.

Paper Structure

This paper contains 15 sections, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. Study site and datasets
  4. Landsat data
  5. COntinuous monitoring of Land Disturbance
  6. Methods
  7. Water bodies extraction
  8. Water and non-water bodies classifier using COLD-derived data
  9. Determining Water Changes at Breakpoints in COLD-derived data
  10. Results
  11. Water Frequency Prediction Using a Regression Model
  12. Assessing Water Frequency Prediction: Classification vs. Regression Models
  13. Determine water change using COLD-derived data
  14. Discussion
  15. Conclusion

Figures (2)

  • Figure 1: Methodology to predict the water mapping and water change from COLD data.
  • Figure 2: Visualization of water map based on the outcomes from the classification (middle) and regression model (right) in a region withs its Landsat image (left). Blue pixels indicate water bodies, while brown and dark brown pixels represent non-water bodies. Lighter-colored pixels represent the inference on pixels with stable periods beyond the timeframe of those used in the training process.