2012 summer

Arctic sea ice forecast

June 1, 2012                         

yuriko ishizashi, daisuke osa,

akihiro tsurugi, yuya nakano,

hidenori mabe, shoutaro morimoto,

makoto kayano, noriaki kimura, hajime yamaguchi

研究室ホーム   メンバー   研究内容   所在と連絡先   ブログ


① The Eastern Hemisphere of the Arctic (Russian side);

    Sea ice melting in the Chukchi Sea and the East Siberian Sea will

    delay compared to last year. The sea ice extent in the Kara Sea

    and the Laptev Sea is likely to retreat.

② The Western Hemisphere (Canadian side);

    Sea ice melting will delay a little compared to last year. 

③ The Northeast Passage and the Northwest Passage will both open.


     This forecast was tackled by fourth year students of School of Engeneering, Ishizashi, Osa, Tsurugi, Nakano, Mabe, and Morimoto.

     In recent years, the sea ice area of the Arctic Sea in summer is decreasing rapidly.  Also, sea ice distribution changes every year (Fig. 4).

     The past research has shown that there is a relationship between the motion of sea ice during winter and sea ice distribution in summer. At places where sea ice converges, sea ice becomes thick, and does not easily melt in summer. On the other hand, at places where sea ice goes away, sea ice becomes thin, and it melts easier in summer. This prediction is based on this view.

     First, we prepare a daily ice velocity dataset calculated from images of the satellite microwave sensor AMSR-E. The procedure for detecting ice motion is based on the maximum cross-correlation method. Then, about 7,000 particles are arrayed over the ice-covered area on December 1 of each year at an interval of 37.5 km. The daily displacement of each particle from the initial location is calculated from the AMSR-E-derived ice velocity. The motions of the particles till the end of April are obtained in this way (Fig. 5).

     This method is carried out under the following assumptions.

1.  Sea ice existing on December 1 is thick.

2.  If the thick ice goes away, open water will be covered with thin ice.

3.  Thick ice does not melt easily but thin ice melts easily.

* Last year's forecast was based on the data of microwave radiometer AMSR-E. Since AMSR-E operation stopped, the data of SSM/I is used this year.

  Since the resolution of SSM/I is coarser than that of AMSR-E, prediction accuracy is considered to be less.

Fig. 3 The anomaly of sea ice concentration(%)

(Blue means less than usual, red means more.)

     Fig. 3 shows the difference between the predicted value and the average value of sea ice concentration. Sea ice in the Eastern Hemisphere (the Laptev Sea side, the Kara Sea side) will be less than usual, and the sea ice of the Western Hemisphere (Canadian side) will be more.

Fig. 5 Distribution of particles at the end of April

a) August 10

b) September 10

     In the Western Hemisphere (the Alaskan and Canadian sides) of the Arctic, sea ice moved only a little during last winter. Sea ice movement was significant in the Eastern Hemisphere (Siberian side). Divergence of sea ice there is larger than usual, especially around the Laptev Sea and the east of Severnaya Zemlya.

     In the Laptev Sea and the Kara Sea, sea ice area will retreat earlier, since sea ice in these areas is supposed to be thin. In these areas, sea ice will disappear by the end of July.

(However, sea ice may remain till later than usual in the west of Severnaya Zemlya.)

     In the Western Hemisphere, especially in the Chukchi Sea, sea ice is supposed to be thick and it is expected that sea ice will remain till later than usual.

Fig. 2 Sea ice concentration in this summer(%)

     Fig. 2 is the prediction of sea ice concentration in August and September calculated based on the past data. In addition, by considering the recent trend of sea ice area and the features of sea ice motion during last winter, we made a figure of this year's sea ice area (Fig. 1). The calculation process is explained later. Extraordinary large value in Fig.2(a) is an error arisen in the process.

Fig. 4 Sea ice distribution on September 10 for the past 18 years

Fig. 6 Animation of particles movements in 2012

Calculation Process

     The relationship between the sea ice area in September and the density of particles at the end of April is analyzed by using the data for the past 18 years.

     Based on the relationship, the sea ice concentrations in August and September are calculated. The results are shown in Fig. 1, 2, and 3.

     Fig.7 shows the correlation coefficient between the number of particles at the end of April and the sea ice concentration in September based on the data for 18 years. Correlation coefficient is high in areas where the sea ice concentration greatly differs year by year.


Fig. 7 Correlation coefficient between the number of particles at the end of April (area of the thick ice which existed on December 1 or before), and sea ice area in September.

     For example, about the area shown with the blue circle in Fig. 7, Fig. 8 shows the regression line of the sea ice concentration in September to the number of particles in April. Based on this line, the sea ice area in September is calculated from the number of particles in April.

Fig. 8 An example of the regression line of the number of particles in April, and the sea ice concentration in September

a) August 10

b) September 10

Fig. 1 Predicted sea ice area on September 10, 2012.

White color level shows sea ice concentration in each sector, the white line is the ice edge by human judgement.

     If you have any questions, please contact Prof. Yamaguchi or Dr. Kimura.

     This work was partly supported by GRENE Arctic Climate Change Research Project.