Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.20085617 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901383587823616 |
|---|---|
| author | Elepathage, Thushani Suleka Madhubhashini |
| author_facet | Elepathage, Thushani Suleka Madhubhashini |
| contents | <p>This research leverages 23 years of NASA’s MERRA-2 reanalysis data (August 2002 to May 2025) to model the non-linear diagnostic relationships between latent heat, sensible heat, and kinetic energy fluxes in the Western Indian Ocean. By evaluating multiple machine learning algorithms, the study establishes that Random Forest regression significantly outperforms Gradient Boosting and Neural Network architectures, achieving an <span class="katex">R2</span> of 0.557. The findings reveal that cyclical seasonal patterns driven by monsoonal forcing are the primary predictors of flux variability, accounting for 64.5% of model importance, while also identifying a significant long-term declining trend in sensible heat flux. Ultimately, the work demonstrates that ensemble tree-based methods provide a robust framework for capturing complex air-sea interactions in climatically sensitive regions, offering critical insights for regional marine resource management.</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_20085617 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Non-Linear Modeling of Ocean-Atmosphere Heat and Energy Fluxes in the Western Indian Ocean: A Machine Learning Approach Elepathage, Thushani Suleka Madhubhashini <p>This research leverages 23 years of NASA’s MERRA-2 reanalysis data (August 2002 to May 2025) to model the non-linear diagnostic relationships between latent heat, sensible heat, and kinetic energy fluxes in the Western Indian Ocean. By evaluating multiple machine learning algorithms, the study establishes that Random Forest regression significantly outperforms Gradient Boosting and Neural Network architectures, achieving an <span class="katex">R2</span> of 0.557. The findings reveal that cyclical seasonal patterns driven by monsoonal forcing are the primary predictors of flux variability, accounting for 64.5% of model importance, while also identifying a significant long-term declining trend in sensible heat flux. Ultimately, the work demonstrates that ensemble tree-based methods provide a robust framework for capturing complex air-sea interactions in climatically sensitive regions, offering critical insights for regional marine resource management.</p> |
| title | Non-Linear Modeling of Ocean-Atmosphere Heat and Energy Fluxes in the Western Indian Ocean: A Machine Learning Approach |
| url | https://doi.org/10.5281/zenodo.20085617 |