Extreme gradient and boosting algorithm for improved bias-correction and downscaling of CMIP6 GCM data across indian river basin
Study region: The Godavari River basin, situated between the geographical coordinates of 73°21′ E to 83°09′ E and 16°07′ N to 22°50′ N, India Study focus: The present study employed an extreme gradient boosting algorithm to enhance bias correction and spatial downscaling of climate model data from t...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-06-01
|
| Series: | Journal of Hydrology: Regional Studies |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S221458182500268X |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Study region: The Godavari River basin, situated between the geographical coordinates of 73°21′ E to 83°09′ E and 16°07′ N to 22°50′ N, India Study focus: The present study employed an extreme gradient boosting algorithm to enhance bias correction and spatial downscaling of climate model data from the Coupled Model Intercomparison Project Phase 6. The methodology utilized diverse training datasets, including five plausible climate models and topographic variables such as elevation, slope, and aspect. The effectiveness of the extreme gradient boosting framework in reproducing climate data was compared with the conventional quantile delta mapping approach across the basin. Additionally, both methods were evaluated across different seasons, including monsoon, pre-monsoon, and post-monsoon. New hydrological insights for the region: The results demonstrated that the extreme gradient boosting model outperformed the quantile delta mapping approach and significantly reduced biases in downscaled climate variables. For instance, the proposed model achieved Nash-Sutcliffe efficiency values of 0.44, 0.96, and 0.97 for precipitation, maximum temperature, and minimum temperature, respectively, whereas the conventional quantile delta mapping method yielded a comparatively less values of −0.34, 0.56, and 0.75. Additionally, uncertainty estimates using the p-factor indicated that the extreme gradient boosting model exhibited lower uncertainty in reproducing the observed spatio-temporal patterns of climate variables. Overall, the proposed framework enhances the reliability of global climate model simulations, supporting robust regional-scale hydrological modeling and climate change impact assessments. |
|---|---|
| ISSN: | 2214-5818 |