Determining the appropriate evaluation metrics for model

The gap

Evidence profile

Sourced from the stated research gap and future-work section and stated challenges and conclusions of the source papers, classified as general, drawn from work published between 2024 and 2026, spanning 3 journals. Those papers have been cited 71 times in total.

Supporting evidence — 6 representative gaps

• Non-stationary Multivariate Bias-corrected CMIP6 Climate Projections of Daily Precipitation and Temperature over India (2026) · Scientific Data · doi

  The lack of explicit representation of underlying physical processes in machine-learning-based bias-correction methods. The distortion of climate change signals and extremes in these methods. The need for a non-stationary multivariate bias-correction technique that preserves climate change signals.

  generalstated research gap
  Keywords: lack explicit representation underlying physical processes machine-learning-based bias-correction
• Non-stationary Multivariate Bias-corrected CMIP6 Climate Projections of Daily Precipitation and Temperature over India (2026) · Scientific Data · doi

  Further research is needed to improve the performance of bias correction techniques - The development of new machine-learning methods that can capture complex nonlinear relationships and spatial patterns in climate data - The application of the CDF-t technique to other types of climate data

  generalfuture-work section
  Keywords: further research needed improve performance bias correction techniques
• Non-stationary Multivariate Bias-corrected CMIP6 Climate Projections of Daily Precipitation and Temperature over India (2026) · Scientific Data · doi

  The reliance on long observational records for machine-learning-based bias-correction methods. The reduced performance of these methods for highly skewed and non-continuous variables such as precipitation. The need to preserve climate change signals while correcting systematic biases in climate model simulations.

  generalstated challenges
  Keywords: reliance long observational records machine-learning-based bias-correction methods reduced
• Multi-Model Ensemble Enhances the Spatiotemporal Comprehensive Performance of Regional Climate in China (2025) · Remote Sensing · cited 11× · doi

  Determining the appropriate evaluation metrics for model selection and weight allocation is essential when constructing MMEs. Investigating the applicability of the QM bias-correction method in inflating/deflating various conditions of climate change signals is necessary.

  generalfuture-work section
  Keywords: determining appropriate evaluation metrics model selection weight allocation
• Multi-Model Ensemble Enhances the Spatiotemporal Comprehensive Performance of Regional Climate in China (2025) · Remote Sensing · cited 11× · doi

  The study faces the challenge of selecting evaluation metrics, bias-correction methods, and comprehensive evaluation methods. The study has to deal with the limitation of the QM bias-correction method, which assumes that the function of error-correction values remains constant and time-invariant. The study has to address the issue of inflating/deflating various conditions of climate change signals.

  generalstated challenges
  Keywords: study faces challenge selecting evaluation metrics bias-correction methods
• A comprehensive comparison of bias correction methods in climate model simulations: Application on ERA5-Land across different temporal resolutions (2024) · Heliyon · cited 49× · doi

  Due to the scarce literature on the adjusting of biases at the hourly time scale, it is evident that a more in-depth analysis of high-resolution temporal climate data is required, and further new methodologies capable of correcting hourly and sub-hourly biases should be developed.

  generalconclusionsevidence 4/5
  Keywords: hourly biases scarce literature adjusting time scale evident depth high resolution temporal climate required further

Questions about this gap