Document presents a new method called Wasserstein Stability Analysis (WSA) for detecting changes in the probability distribution of climate data, which overcomes the limitations of traditional trend analysis based on average values.

Key ideas and facts:

• Inadequacy of average value analysis: While averages are useful for capturing overall energy conditions and filtering out noise in the data, with the advent of the big data era, it is becoming important to uncover deeper patterns and correlations in complex, high-resolution data sets. Changes in probability distributions that may be important for understanding climate change remain hidden when analyzing averages.
• Wasserstein distance: WSA uses Wasserstein distance (W-distance), a metric that quantifies the distance and similarities between two probability distributions. This metric has proven useful in various fields of climatology, but its potential in analyzing the physical properties of big data has remained largely unexplored.
• Detection of shifts in extreme events: By applying WSA to air temperature data from 1998-2013, the authors identified significant shifts in extreme events. For example, the eastern equatorial Pacific experienced a decrease in hot extremes and an increase in cold extremes, suggesting a temperature shift toward a La Niña-like phenomenon, even though no significant trend in average temperature was observed.
• Impact of melting sea ice: WSA also revealed significant changes in the probability distribution of air temperature in the Arctic regions. In the Barents and Kara Seas, where sea ice has been significantly reduced, WSA showed a weakening of the physical constraints on maximum air temperature, leading to changes in the shape of the probability distribution. This change is not easily detected using traditional trend analysis.
• Significance for climate modeling: The authors emphasize that the ability of WSA to detect changes in the probability distribution associated with sea ice melt could be a valuable tool in the development and validation of climate models. Improved sea ice modeling was one of the most significant advances in the IPCC Sixth Assessment Report compared to previous reports.

Quotes from the article:

• "While averages represent important statistical properties of data sets, capture overall energy conditions, and filter out noise in low-quality data sets well, the advent of the big data era requires new methodologies."
• "In this study, we applied W-distance to the physical analysis of climate data from the early 21st century. By using W-distance, we introduce a new method – Wasserstein stability analysis (WSA) – aimed at revealing the variability of the probability density function (PDF) under climate change."
• "The results show that despite a non-significant trend, the equatorial eastern Pacific experienced a decrease in hot extremes and an increase in cold extremes, suggesting a La Niña-like temperature shift."
• "The findings suggest that sea ice strongly limits hot extremes of air temperature near the surface. This influence diminishes as sea ice melts."

WSA appears to be a promising tool for investigating climate change dynamics, providing valuable information for a better understanding of climate evolution. WSA reveals changes in probability distributions that may be overlooked by traditional trend analysis, thereby allowing deeper insight into complex climate processes, such as shifts in extreme events and the impact of sea ice melt.