Lim et al. (2026) Generation of Rainfall Maps from GK2A Satellite Images Using Deep Learning

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

Identification

• Journal: Remote Sensing
• Year: 2026
• Date: 2026-01-06
• Authors: Yerim Lim, Yeji Choi, Eunbin Kim, Yong-Jae Moon, Hyun-Jin Jeong
• DOI: 10.3390/rs18020188

Research Groups

Not explicitly mentioned in the provided text.

Short Summary

This study develops a deep learning framework using GEO-KOMPSAT-2A satellite imagery for accurate rainfall estimation, demonstrating that an ensemble model significantly enhances real-time, high-resolution rainfall monitoring capabilities.

Objective

• To develop and evaluate a deep learning-based framework for accurate rainfall estimation using multispectral GEO-KOMPSAT-2A (GK2A) satellite imagery, primarily focusing on daytime observations to leverage visible channel information.

Study Configuration

• Spatial Scale: 2 km and 4 km
• Temporal Scale: One year (May 2023 to April 2024) and the August 2023 monsoon season

Methodology and Data

• Models used: Model-HSP (based on Pix2PixCC architecture), Model-CMX (radar-based), Model-ENS (ensemble model integrating Model-HSP and Model-CMX outputs).
• Data sources: Multispectral GEO-KOMPSAT-2A (GK2A) satellite imagery (daytime observations), Hybrid Surface Precipitation (HSP) data from weather radar (for training Model-HSP).

Main Results

• The ensemble model (Model-ENS) demonstrated superior capability in accurately capturing rainfall distribution, intensity, and temporal evolution across diverse weather conditions.
• Deep learning significantly enhances GEO satellite rainfall estimation, enabling real-time, high-resolution monitoring.
• Performance was quantitatively evaluated using Root Mean Square Error (RMSE) and Correlation Coefficient (CC) metrics over the specified periods and spatial resolutions.

Contributions

• Presents a novel deep learning-based framework for high-resolution, real-time rainfall estimation using GEO satellite imagery.
• Enables accurate rainfall monitoring in regions with sparse or limited weather radar coverage.
• Offers strong potential for advancing global and regional hydrometeorological and climate research applications.

Funding

Not explicitly mentioned in the provided text.

Citation

@article{Lim2026Generation,
  author = {Lim, Yerim and Choi, Yeji and Kim, Eunbin and Moon, Yong-Jae and Jeong, Hyun-Jin},
  title = {Generation of Rainfall Maps from GK2A Satellite Images Using Deep Learning},
  journal = {Remote Sensing},
  year = {2026},
  doi = {10.3390/rs18020188},
  url = {https://doi.org/10.3390/rs18020188}
}