CMIP6 Precipitation Downscaling for Europe Using Bias Correction and Random Forest

A 2026 study by Smit Chetan Doshi evaluated bias-correction and machine learning downscaling for precipitation data from 11 high-resolution CMIP6 models across Europe. The dataset, 8.6 MB in size, compares methods like empirical quantile mapping and random forest-empirical quantile mapping to align model outputs with observed data. It is shared under a CC-BY-4.0 license on figshare.

Use Cases

• Evaluating bias-correction methods for climate model precipitation outputs based on the described empirical quantile mapping technique.
• Training machine learning models for regional climate downscaling based on the random forest-empirical quantile mapping approach mentioned.
• Analyzing seasonal precipitation error patterns across Europe based on the described lower residual errors in summer versus winter.
• Assessing the reduction of precipitation overestimations in Central and Eastern Europe as described in the study.

Strengths

• Compares outputs from 11 high-resolution CMIP6 climate models.
• Methodology specifically evaluated for low rainfall intensities (0–5 mm/day).
• Includes seasonal analysis comparing summer and winter performance.

Limitations

• Row count and column-level documentation are unknown, limiting suitability assessment.
• The primary data file is a DOCX document (8.6 MB), which may not be a standard format for structured climate data.
• Description metadata is limited; actual data quality and structure require manual inspection after download.

Provenance

Source

Smit Chetan Doshi via figshare.

Collection Method

Bias-correction and downscaling of precipitation data from 11 CMIP6 models using empirical quantile mapping and random forest techniques.

Freshness

Last updated 2026-04-14 04:19:05.

Geography

Europe