by Frank Bosse
Short answer: NO
A recent article Kim et al. (2023)the following is K23, which gets some media attention, for example this Articles at CNN: “The Arctic may be free of summer sea ice by the 2030s, new study warns.”
Key conclusion of K23: “The results indicate that the first sea ice-free September will occur as early as the 2030s–2050s regardless of emissions scenarios.”
How did the authors come to this conclusion? They used the CMIP6 Multi-Model Mean (MMM)! In the methods section: “We use multi-model CMIP6 and DAMIP historical simulations performed under different combinations of climate constraints…”
DAMIP: “Comparison project between discovery and attribution models (DAMIP v1. 0)”
This is the same approach described in this blog post (for the Antarctic inversion case), with pitfalls like those described by Gavin Schmidt in 2021.
Gavin’s conclusion: “The default behavior in the community should be to move away from treating the raw model group as meaningful.” Well, this is not simply Gavin’s wisdom, but the IPCC AR6 WG1 didn’t use MMM, for the same reasons. Unfortunately, this approach did not reach certain sections of the community, not to mention the editorial staff of “Nature Communications.”
To show the impact of CMIP6 MMM selection over not too biased CMIP5 models, I compared the 2020-2050 trend slopes for both cases for September in the Arctic region:
Figure.1: Steep September temperature trend from 2000-2050 (K/year) for CMIP5 MMM (top) and CMIP6 MMM (bottom). Numbers are generated with KNMI . Climate Explorer.
Notice the trend is about 30% steeper in CMIP6!
Key figures for K2023:
Figure 2: Reconstruction of parts of Figure 4 of K23.
The authors “expanded” the CMIP6 MMM to the Observed Sea Ice Area (SIA) 1979-2019 because the MMM produced too much Arctic Ice during this time. Indeed, temperatures modeled for the period 1979-2019 are cooler than the observed data. The reason is not clear, it has probably been estimated that too much aerosol (cooling) is forced during this time in the CMIP forced data.
After “scaling” (colored lines in Figure 2) also “Sustainable Development Scenario” SSP1-2.6 leads to SIA disappearing (less than 1 million km²) around 2050, for SSP2-4.5 , this is almost the same (2045) and the whole August Sea of Ice will perish before 2060! Amazing, but very questionable due to the choice of MMM.
Here is my method for calculating the possible SIA in September under 1 Mio km². i regress NSIDC- Data (September) with brand new ERF (effective radiant force) -data from this paper (still in preprint) from 1979 to 2022.
Figure 3: Regression of SIA data versus ERF data. Note that the variance of the ERF determines only 55% (R²) of the variance of the September SIA.
As a result, the Sea Ice’s sensitivity to double CO2 (ERF= 3.9 W/m², according to IPCC AR6) gives -3.69 Mio km²/2*CO2. The range of 17…83% possible: -3.16…-4.13 Mio km²/2*CO2, calculated with the CI of the regression.
Approximately 45% of the SIAs are affected by internal variability, which is not described in any of the MMMs because all the variability of single models is zero when averaging multiple models running like the MMM. Therefore, the use of any MMM is as biased as the result of K23.
The trend slope calculated in Figure 3 results that the forced portion of the September SIA results in a residual amount of 1 Mio km² of 5.24 W/m². However, one must take into account the internal variability (iv), described in the residuals between observations and trends due to the ERF:
Figure 4: Trend residual in Figure 3. The highest impact on the decrease in the number of observations was in 2012 with – 1.18 Mio km² below the trend.
Therefore, I calculated the ERF for the remaining 2.18 km² SIA to find the year that could reach 1 Mio km² considering the most negative iv of the last 45 years. This gives an ERF of 3.92W/m².
It is now possible to estimate in time:
Figure 5: SSP ERF data over time. Source
As for the probable September ice-free years, one can find: never for SSP1-2.6, around 2060 in the SSP 2-4.5 scenario. The North Pole has no continuous ice in September due to the iv force alone (ERF=5.24 W/m²) which we will not see, neither in the SSP1-2.6 scenario nor in the SSP2 scenario. -4.5 W. Compare this result with the result at K23 (Figure 2). My method uses only observed data and thus avoids possible biases in climate models.
A very recent article (published 12lame pants of June 2023 in “Nature Climate Change”) also uses the same climate modeling approach as K23, but takes advantage of limited single-choice models rather than MMMs. It leads to results very similar to my (ERF-based only) approach: An Arctic Sea Ice Area with an area of less than 1 Mio km² would be observable no earlier than 10 years (when searching). see K23) which is 10 years after 2050, also around 2060 .
Conclusion: K23 takes advantage of the rejected CMIP 6 MMM and achieves strongly biased results. It’s overdue for the “Nature communication” editorial team to check the review process to make sure it doesn’t mislead the audience and the media.
Related
