Correction to “Evaluation of the simulation of the annual cycle of Arctic and Antarctic sea ice coverages by 11 major global climate models”

[1] In the paper “Evaluation of the simulation of the annual cycle of Arctic and Antarctic sea ice coverages by 11 major global climate models” by Claire L. Parkinson, Konstantin Y. Vinnikov, and Donald J. Cavalieri (Journal of Geophysical Research, 111, C07012, doi:10.1029/2005JC003408, 2006), the ice amounts from the French Institut Pierre Simon Laplace (IPSL) CM4 model were underrepresented because of mistakenly showing the amounts for ice thicknesses of at least 60 cm instead of ice thicknesses of at least 6 cm. This affected Figures 2j, 3j, 4 and 5, and the text discussion of each of these figures. The authors apologize for the error and thank Olivier Arzel, Thierry Fichefet, and Jean-Louis Dufresne for pointing it out and for confirming the correctness of the revised figures and text. Corrected Figures 2, 3, 4, and 5 appear below, along with corrected portions of the text, all resulting exclusively from the revised IPSL results. [2] In the abstract, the statement that “in the Southern Hemisphere all except one simulate the monthly averages to within ±6.3 × 106 km2 of the observed values” should be revised to delete “except one.” [3] Paragraph [30] should read as follows: “The French IPSL CM4 model does an excellent job in simulating the annual cycle of ice extents in the Northern Hemisphere, although with a peak 1 month early, in February, and with a subsequent slight under simulation of the extent during the ice decay period (Figure 3j). However, in the winter the excellent ice extent values are deceiving, as they reflect significant but offsetting errors in different regions (Figure 2j). Specifically, the winter ice distribution has too little ice in the Bering Sea and the Sea of Okhotsk but too much ice on the Atlantic side, in the Labrador Sea, Greenland Sea, and Barents Sea. Also affecting the winter ice extent values, ice is not calculated for the Canadian Archipelago or James Bay (southeast of Hudson Bay). The simulated summer ice distributions are considerably better, being excellent in the Greenland Sea, showing slightly too much ice in the northern Barents Sea and western Kara Sea, and showing slightly too little ice in the Beaufort Sea as well as the Canadian Archipelago (Figure 2j). In the Southern Hemisphere, the timing of minimum and maximum ice extents in February and September, respectively, is correct, although the model simulates too little ice for most of the year, with too much ice in the peak winter months of August and September (Figure 3j). Spatially, the March, late summer Southern Hemisphere ice distribution correctly has the most ice in the western Weddell Sea and in the Ross Sea although has too little ice in both those seas and has none of the coastal ice that the observations show around the rest of the continent. The late winter ice distribution has close to the correct amount of ice but has it positioned with far too much ice in the Bellingshausen, Amundsen, and western Weddell seas, to the west and east of the Antarctic Peninsula, and too little ice around the rest of the continent.” [4] The third sentence of paragraph [33] should begin as follows: “In the Southern Hemisphere, 10 of the 11 models….” [5] The last sentence of paragraph [36] should read as follows: “Similarly, the much better performance of the Norwegian and UK HadGEM1 models in the Atlantic versus the Pacific portion of the sub-Arctic (Figures 2c and 2h) could result from a greater interest in the Atlantic region, even if the attention was toward the atmospheric and oceanic circulations, affecting weather conditions in Norway and the UK, rather than specifically toward the sea ice.” [6] In the next to the last sentence in paragraph [38], “all models except one have simulated” should be “all models have simulated.” [7] Paragraph [39] should read as follows: “Despite the variety of problems with the individual simulations, when the results from the 11 models are averaged, the ensemble average of the model simulations does quite well in simulating the annual cycle of sea ice extents in each hemisphere, although with the interesting contrast that the ensemble Northern Hemisphere monthly averages are all greater than the observations while the ensemble Southern Hemisphere monthly averages are mostly less than the observations (Figure 5). Percentage-wise, the Northern Hemisphere averages exceed the observations by values ranging from 2.8% in December to 14.1% in September and the Southern Hemisphere averages differ from the observations by values ranging from 2.5% too high in October to 23.1% too low in April. In terms of monthly ice extents, the Northern Hemisphere ensemble averages are greater than the observed by amounts ranging from 0.35 × 106 km2 in November to 1.3 × 106 km2 in May and the Southern Hemisphere averages differ from the observed by amounts ranging from 0.45 × 106 km2 too high in October to 1.6 × 106 km2 too low in April.” [8] The first two sentences of paragraph [40] should read as follows: “Using the root mean square (RMS) of the 12 monthly departures from the observed ice extents as one measure of how well a model is performing overall, the 11-model composite (with an RMS of 0.8 × 106 km2) comes out superior to any of the individual models in the simulation of Southern Hemisphere ice extents, followed by the ECHAM5 and CSIRO Mk3 models (RMS = 1.6 × 106 km2). In the simulation of the Northern Hemisphere ice extents, the CGCM3 and IPSL CM4 come out on top (with an RMS of 0.6 × 106 km2), followed by the 11-model composite and the MIROC3 and ECHAM5 models (RMS = 0.8 × 106 km2).”

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.