Estimating daily climatologies for climate indices derived from climate model data and observations

Area of Science:

• Climatology
• Climate modeling
• Statistical analysis

Background:

• Climate indices offer more impact-relevant climate information than simple means.
• Indices often rely on daily data and thresholds, which can be sensitive to computation methods.
• Estimating daily climate means, standard deviations, and percentile thresholds involves uncertainties due to sample size limitations in observational and model data.

Purpose of the Study:

• To explore sampling uncertainties in estimating climate indices.
• To compare different approaches for improving climate index estimation.
• To assess the effectiveness of a statistical fitting approach for daily climate characteristics.

Main Methods:

• Utilized a large number of past ensemble seasonal forecasts (hindcasts) for a perfect model approach.
• Applied a statistical fitting approach to improve estimates of daily climatologies, percentile-based thresholds, mean, and variability.
• Investigated uncertainties arising from sample size issues in reference periods and model data.

Main Results:

• The statistical fitting approach significantly enhances estimates of daily climatologies for percentile-based thresholds over land areas.
• Improvements were observed in the estimation of daily mean and variability.
• The method demonstrates substantial benefits for bias removal in long-range forecasts and climate index predictions.

Conclusions:

• A statistical fitting approach, based on a perfect model, yields more robust estimates of daily climate characteristics.
• This method is valuable for bias correction in climate models and improving climate index predictions.
• The approach shows significant potential for climate change studies and impact assessments.