Review on: Weather pattern dynamics over Western Europe under climate change: Predictability, Information Entropy and Production

Author: Stephane Vannitsem

Submitted to Nonlinear Processes in Geophysics, manuscript  egusphere-2022-778

General

This study considers the atmosphere as a nonequilibrium steady state system and applies methods published by Gaspard (2004) to determine predictability and irreversibility by the information entropy in observational and simulated data. The author uses block entropies for the forward and the time reversed block entropy in time series describing the North Atlantic/West European weather. The time evolution is described as a coarse-grained sequence of visited boxes. The predictability is assessed by the forward entropy and the entropy production by the irreversibility due to time reversal asymmetry.  

The data are Großwetterlagen in the Eastern North Atlantic/Western Europe sector, which had been  extracted in observations and scenario simulations. The daily time series are reduced by clustering of the patterns to 3, 6 and 8 time series. The observational time period is 1850-2019, and the simulated data is for 1900-2100. As the numerical effort for the joint probabilities is enormous, the 30 patterns had to be drastically reduced to 3, 6 and 8. Furthermore, the block lengths had to be reduced to two, to calculate the entropy S₂. Thus, the present study is at the border of computational feasibility.

The study is insightful and relevant, although somehow preliminary, mostly due to computation restrictions. The agreement with previous studies hints at a reproducible core of results. The author should try to respond to the concerns, and if possible, less costly analyses might be added.

Specific Comments

I have several concerns, mostly on the use of Großwetterlagen and the nonstationarities in the data (mentioned in line 199).

• Großwetterlagen: Großwetterlagen are certainly useful and have their merits in synoptics, but do they form a complete basis in state space? Do they depend on the domain similar to EOFs? Is a comparison with other sets meaningful? A short list of the selected Großwetterlagen patterns and the clusters would be useful.

• S₂ in the 3-pattern (Fig. 2): For the 3-pattern the forward and backward entropies are S₂^R=S₂, hence there is no information entropy production in this basis. What does that mean for the choice of patterns? Is it possible to determine the entropy production independent of the basis?

• Anthropogenic climate change since 1860: Global warming started early. Is it possible to find a similar behavior in the 21nd century, hence a common imprint of global warming?

• Natural low-frequency variability: Is the sea surface temperature relevant for the frequencies of the patterns in Fig. 2?

• Decrease of S₂ during 1850-1900: Is the strong decrease of S₂ in Fig. 2 a hint for an overlooked nonstationarity?

Minor/Typos

Line 10: is?

Lines 149-154: the paragraph could be clearer, is n=7?

Figure 3 caption:  length of words, n?

Figures 4,5: a legend would be useful.  

Review of “Weather pattern dynamics over Western Europe under climate change: Predictability, Information Entropy and Production” by S. Vannitsem (Nonlinear Processes in Geophysics)

In this manuscript, the author analyzed the entropy and its production in the information theory to study the impact of climate change on the weather patterns over Western Europe. In historical data for the period 1885-2000, the author found a decreasing trend of the block entropy except afterward of 1980, which suggests a less diverse set of pairs of events. In addition, an increasing trend of the entropy production for 6 and 8 weather patterns indicates a time-asymmetry related to the irreversibility of the process. The analysis of the UK Met Office CMIP5 model showed a wide range of the block entropy evolution depending on the realization. These findings suggest that the degree of irreversibility is increasing under climate change.

The manuscript is well written with a clear structure. I believe this study would contribute to understand how climate change affects the weather pattern over Western Europe. In addition, the application of block entropy to climate model runs would be useful index to evaluate climate models. However, I have concerns due to the lack of the discussion that connects the obtained results and weather events.

Major comments:

Discussions that connect the (information) entropy production to atmospheric dynamics would be helpful to deepen our understanding of climate change impact on the weather regime. The authors wrote that the production of information entropy is related to irreversibility of the system, and its trend would be associated with the heat production/dissipation in the thermodynamic entropy. I wonder if the production of information entropy is associated with the irreversible processes in the large-scale atmospheric dynamics, like irreversible mixing of momentum that occurred at the Rossby wave breaking. Modulation of Rossby wave breaking is known to be associated with the transition of weather regime (Michel and Riviere, 2011).

Minor comments:

Line 12: A quantitative assessment of the change of the entropy production (10% in the RCP2.6 and 30-40% in the RCP8.5) is missing in the result and conclusion.

Line 110-114: As the main finding using the 15 model runs shows the diverged block entropy evolution depending on the realization, it would be useful for readers to briefly describe the difference in the realizations between 15 runs. (The difference is only in parameterization? or also in boundary conditions?)

Line 124: “A clear trend … is visible, and chi^2 tests … are highly significant”. Please provide the evidence (figure/table) of this sentence.

Figures: labels (e.g. (a), (b),…) are too small to see. Please consider put larger labels in the upper left of the figure.

Figures 4-5: Legend would be useful.

Reference:

Michel, C., & Rivière, G. (2011). The Link between Rossby Wave Breakings and Weather Regime Transitions, Journal of the Atmospheric Sciences, 68(8), 1730-1748.