August et al. (2026) Remote Sensing and Machine Learning Approaches for Hydrological Drought Detection: A PRISMA Review

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Water
• Year: 2026
• Date: 2026-01-31
• Authors: Odwa August, Malusi Sibiya, Masengo Ilunga, Mbuyu Sumbwanyambe
• DOI: 10.3390/w18030369

Research Groups

Not specified in the provided text.

Short Summary

This review systematically synthesizes the state-of-the-art in using convolutional neural networks (CNNs) and satellite-derived vegetation indices for hydrological drought detection, concluding that hybrid spatiotemporal models, particularly CNN-LSTM with multi-modal data fusion, are the most effective.

Objective

• To systematically synthesize the state-of-the-art in using convolutional neural networks (CNNs) and satellite-derived vegetation indices for hydrological drought detection.

Study Configuration

• Spatial Scale: Global (reviewing studies across various regions)
• Temporal Scale: Studies published between 1 January 2018 and August 2025.

Methodology and Data

• Models used: This review synthesizes studies that primarily utilize Convolutional Neural Networks (CNNs), often in hybrid configurations with Long Short-Term Memory (LSTM) models (e.g., CNN-LSTM).
• Data sources: Satellite-derived vegetation indices (VIs) and complementary data such as Land Surface Temperature (LST). The review itself is based on a systematic search of 138 published studies following PRISMA guidelines.

Main Results

• 58% of the 138 included studies focused on hybrid CNN-LSTM models, with a notable increase in publications observed after 2020.
• Hybrid spatiotemporal models are identified as the most effective for hydrological drought detection, demonstrating superior forecasting skill.
• These hybrid models can achieve 10–20% higher accuracy compared to standalone CNNs in some cases.
• The most robust models employ multi-modal data fusion, integrating vegetation indices (VIs) with complementary data like Land Surface Temperature (LST).

Contributions

• Provides a comprehensive, systematic synthesis of the current state-of-the-art in applying CNNs and satellite VIs for hydrological drought detection.
• Identifies the most effective model architectures (hybrid spatiotemporal models like CNN-LSTM) and data fusion strategies (multi-modal data, VIs with LST) for improved drought detection.
• Highlights critical future research directions, including enhancing model transferability and incorporating explainable AI (XAI) to strengthen the operational utility of drought early warning systems.

Funding

Not specified in the provided text.

Citation

@article{August2026Remote,
  author = {August, Odwa and Sibiya, Malusi and Ilunga, Masengo and Sumbwanyambe, Mbuyu},
  title = {Remote Sensing and Machine Learning Approaches for Hydrological Drought Detection: A PRISMA Review},
  journal = {Water},
  year = {2026},
  doi = {10.3390/w18030369},
  url = {https://doi.org/10.3390/w18030369}
}