Kanishkar et al. (2025) Enhancing temperature data analysis through Threshold-Optimized Ensemble Detection (TOED) approach of climate anomalies

Identification

• Journal: Advances in Space Research
• Year: 2025
• Date: 2025-11-07
• Authors: K. Kanishkar, J. Rufina Sherin, L. Gowri
• DOI: 10.1016/j.asr.2025.11.012

Research Groups

• School of Electronics and Communication Engineering, SRMIST, Tiruchirappalli, India
• School of Computing, SASTRA Deemed University, Thanjavur, Tamil Nadu, India

Short Summary

This study introduces a Threshold-Optimized Ensemble Detection (TOED) approach for identifying climate temperature anomalies, demonstrating its superior effectiveness (92.1% AUC-ROC) compared to individual anomaly detection methods.

Objective

• To evaluate and enhance the performance of temperature anomaly detection by developing a Threshold-Optimized Ensemble Detection (TOED) method that combines optimal isolation forest, Gaussian mixture model, and local outlier factor techniques.

Study Configuration

• Spatial Scale: Unspecified, focusing on the detection of "small-scale anomalies."
• Temporal Scale: Unspecified, focusing on the detection of anomalies in temperature data relevant to "long-term shifts in climate" and "extreme weather events."

Methodology and Data

• Models used: Optimal Isolation Forest, Gaussian Mixture Model, Local Outlier Factor, and a novel Threshold-Optimized Ensemble Detection (TOED) method combining these.
• Data sources: Unspecified temperature datasets.

Main Results

• The introduced Threshold-Optimized Ensemble Detection (TOED) method achieved an AUC-ROC score of 92.1% for identifying small-scale temperature anomalies.
• TOED demonstrated greater effectiveness and robustness in anomaly detection compared to individual methods (optimal isolation forest, Gaussian mixture model, and local outlier factor).

Contributions

• Introduction of a novel Threshold-Optimized Ensemble Detection (TOED) approach that combines multiple anomaly detection techniques with weighted averaging and threshold adjustment.
• Demonstrated superior performance of the TOED method in identifying small-scale temperature anomalies, offering a more reliable and robust tool for climate monitoring and research.

Funding

• Not specified in the provided text.

Citation

@article{Kanishkar2025Enhancing,
  author = {Kanishkar, K. and Sherin, J. Rufina and Gowri, L.},
  title = {Enhancing temperature data analysis through Threshold-Optimized Ensemble Detection (TOED) approach of climate anomalies},
  journal = {Advances in Space Research},
  year = {2025},
  doi = {10.1016/j.asr.2025.11.012},
  url = {https://doi.org/10.1016/j.asr.2025.11.012}
}