González-Márquez et al. (2025) Estimation of dam water volume in northern Sinaloa, Mexico using Landsat imagery and artificial intelligence models

Identification

• Journal: Advances in Space Research
• Year: 2025
• Date: 2025-10-09
• Authors: Luis Carlos González-Márquez, Ivette Renée Hansen-Rodríguez, Ramiro Ahumada-Cervantes, Franklin Torres-Bejarano
• DOI: 10.1016/j.asr.2025.09.101

Research Groups

• Engineering and Technology Department, Universidad Autónoma de Occidente, Unidad Regional Guasave, Sinaloa, Mexico
• Doctorado en Sustentabilidad, Universidad Autónoma de Occidente, Unidad Regional Guasave, Sinaloa, Mexico
• Environmental Engineering Department, Universidad de Córdoba, Montería, Colombia

Short Summary

This study developed and evaluated a methodology to estimate dam water volumes in northern Sinaloa, Mexico, addressing data gaps where traditional area-volume curves were unavailable. By integrating Landsat imagery, spectral indices, and artificial intelligence models, the Deep Neural Network achieved the best performance, accurately estimating volumes with high coefficients of determination.

Objective

• To develop and evaluate a methodology for estimating dam water volumes by integrating flooded surface areas (derived from spectral indices computed in Google Earth Engine) with in-situ volume measurements using artificial intelligence models and polynomial regression, particularly when traditional area-volume curves are unavailable.

Study Configuration

• Spatial Scale: Bacurato and El Sabinal dams in northern Sinaloa, Mexico.
• Temporal Scale: Historical period addressing data gaps from 1981/1985 (dam operation start) onwards, utilizing Landsat imagery for monitoring.

Methodology and Data

• Models used: Random Forest, Extreme Gradient Boosting, Deep Neural Network for Regression, Polynomial Regression.
• Data sources: Landsat imagery, spectral indices (specifically AWEInsh), Google Earth Engine, in-situ volume measurements.

Main Results

• The AWEInsh water index provided the most accurate flooded area delineation, achieving an overall accuracy of 99.5 %.
• The Deep Neural Network for Regression demonstrated the best performance among the evaluated models.
• The Deep Neural Network attained coefficients of determination (R²) above 0.99 during validation and above 0.95 with independent data.
• The methodology successfully combines Landsat imagery, spectral indices, and artificial intelligence to accurately estimate dam water volumes, supporting informed water management.

Contributions

• Developed a novel methodology for estimating dam water volumes that circumvents the need for traditional area-volume curves, which are often unavailable, addressing critical data gaps.
• Successfully integrated remote sensing data (Landsat imagery and spectral indices via Google Earth Engine) with advanced artificial intelligence models and in-situ measurements.
• Demonstrated the superior performance of a Deep Neural Network for this specific application, providing a robust tool for water resource management.
• Offers a valuable approach for monitoring and managing water resources in regions facing similar challenges of incomplete historical records and monitoring gaps.

Funding

Not explicitly mentioned in the provided text.

Citation

@article{GonzálezMárquez2025Estimation,
  author = {González-Márquez, Luis Carlos and Hansen-Rodríguez, Ivette Renée and Ahumada-Cervantes, Ramiro and Torres-Bejarano, Franklin},
  title = {Estimation of dam water volume in northern Sinaloa, Mexico using Landsat imagery and artificial intelligence models},
  journal = {Advances in Space Research},
  year = {2025},
  doi = {10.1016/j.asr.2025.09.101},
  url = {https://doi.org/10.1016/j.asr.2025.09.101}
}