Assessing the impacts of climate change on precipitation through a hybrid method of machine learning and discrete wavelet transform techniques, case study: Cork, Ireland
In one sentence. Combines machine learning and wavelet simulations with reweighting to bias-correct CMIP6 NEX-GDDP precipitation for Cork, projecting a decline to about 116–119 mm/month by 2076–2100.
Abstract
Cork City, Ireland. Reconstruction of precipitation timeseries is gaining increasing attention for monitoring and prediction studies. To address the challenges posed by non-smooth distributions of precipitation data, this research proposes an innovative reweighting method combining simulations from machine-learning and wavelet methods. By amalgamating multiple models, it aims to generate enhanced results, based on the concept that increasing the diversity of simulations and reweighting them improve the accuracy. Particularly in regions like Cork City, which have experienced heavy precipitation events, accurate prediction is vital for informed decision-making. Here, monthly precipitation simulations from 35 National Aeronautics and Space Administration (NASA) Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) models of Phase 6 of the Climate Model Intercomparison Project (CMIP6) were utilized. The simulations considered the Shared Socioeconomic Pathways (SSPs): SSP1-2.6 and SSP5-8.5, for the hindcast and forecast periods of 1990–2014 and 2076–2100, respectively. Using the proposed technique, the results demonstrate the superior performance of the approach in reconstructing precipitation timeseries from NEX-GDDP-CMIP6 simulations, particularly in dealing with high precipitation intensities, while maintaining accuracy for low intensities. Additionally, the projections indicate that precipitation is expected to decrease from 122.88 during 1990–2014–115.67 (SSP1–2.6) and 118.71 mm/month (SSP5-8.5) in 2076–2100. The findings support the effectiveness of the hybrid method in improving data accuracy and reducing potential uncertainties in simulations.
Labels
- Topic
- Post-processing & calibration, Climate analysis & attribution
- Type
- New model or method
- Method
- Classical ML (trees, GP, linear, kernels)
- Time scale
- Climate projections / centennial
- Domain
- Local / station
- Data
- CMIP / climate model output
- Status
- Research
- Identifiers
- DOI:10.1016/j.ejrh.2023.101523 OpenAlex:W4386463742
- Added
- 2026-10-22
Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.