Mishra et al. (2025) Improving the Prediction of Land Surface Temperature Using Hyperparameter-Tuned Machine Learning Algorithms

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

Identification

• Journal: Atmosphere
• Year: 2025
• Date: 2025-11-15
• Authors: Anurag Mishra, Anurag Ohri, Prabhat Singh, Nikhilesh Singh, Rajnish Kaur Calay
• DOI: 10.3390/atmos16111295

Research Groups

Not explicitly stated in the provided text.

Short Summary

This study developed a machine learning framework to predict Land Surface Temperature (LST) at a 10 m spatial resolution by leveraging Sentinel-2 spectral indices and Landsat 8-derived LST data, demonstrating improved accuracy for urban thermal dynamics monitoring.

Objective

• To predict Land Surface Temperature (LST) at a 10 m spatial resolution using Sentinel-2 spectral indices as independent variables and Landsat 8-derived LST data as the target variable within a hyperparameter-tuned machine learning framework.

Study Configuration

• Spatial Scale: 10 m (predicted LST resolution), 30 m (Landsat 8 LST resolution).
• Temporal Scale: Pre- and post-monsoon seasons (for model evaluation).

Methodology and Data

• Models used: Random Forest (RF), Gradient Boosting Machine (GBM), Support Vector Machine (SVM), and k-Nearest Neighbours (kNN). Hyperparameter tuning was performed using grid search combined with cross-validation.
• Data sources: Sentinel-2A (multispectral data, providing spectral indices: Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Normalized Difference Built-up Index (NDBI), and Bare Soil Index (BSI)), Landsat 8 OLI/TIRS (Land Surface Temperature (LST) data).

Main Results

• The machine learning models, specifically RF, GBM, SVM, and kNN, successfully modeled complex nonlinear relationships between Sentinel-2 spectral indices and LST.
• The grid search-based hyperparameter tuning, combined with cross-validation, enhanced the model’s prediction accuracy for both pre- and post-monsoon seasons.
• This approach, integrating Sentinel-2 data with tuned machine learning models, surpassed earlier methods that either employed untuned models or failed to integrate Sentinel-2 data.
• The study demonstrated an enhanced capability for urban heat island monitoring, climate adaptation planning, and sustainable environmental management models by capturing urban thermal dynamics at fine spatial and temporal scales.

Contributions

• Introduction of a novel method for predicting LST at a 10 m spatial resolution by combining Sentinel-2 multispectral data with Landsat 8 LST using hyperparameter-tuned machine learning algorithms.
• Significant improvement in LST prediction accuracy compared to previous methods that used untuned models or did not integrate Sentinel-2 data.
• Enables more detailed and accurate monitoring of urban thermal dynamics, supporting applications in urban planning and climate change adaptation.

Funding

Not explicitly stated in the provided text.

Citation

@article{Mishra2025Improving,
  author = {Mishra, Anurag and Ohri, Anurag and Singh, Prabhat and Singh, Nikhilesh and Calay, Rajnish Kaur},
  title = {Improving the Prediction of Land Surface Temperature Using Hyperparameter-Tuned Machine Learning Algorithms},
  journal = {Atmosphere},
  year = {2025},
  doi = {10.3390/atmos16111295},
  url = {https://doi.org/10.3390/atmos16111295}
}