Chinese Antarctic Center of Surveying and Mapping, Wuhan University, Wuhan 430079, China
2.
School of Geospatial Engineering and Science, Sun Yat-sen University, Guangzhou 510275, China
3.
Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai), Zhuhai 519082, China
Funds:
The National Key Research and Development Program of China under contract No. 2018YFC1406102; the Funds for the Distinguished Young Scientists of Hubei Province (China) under contract No. 2019CFA057; the National Natural Science Foundation of China under contract Nos 41941010 and 41776200.
Information on the Arctic sea ice climate indicators is crucial to business strategic planning and climate monitoring. Data on the evolvement of the Arctic sea ice and decadal trends of phenology factors during melt season are necessary for climate prediction under global warming. Previous studies on Arctic sea ice phenology did not involve melt ponds that dramatically lower the ice surface albedo and tremendously affect the process of sea ice surface melt. Temporal means and trends of the Arctic sea ice phenology from 1982 to 2017 were examined based on satellite-derived sea ice concentration and albedo measurements. Moreover, the timing of ice ponding and two periods corresponding to it were newly proposed as key stages in the melt season. Therefore, four timings, i.e., date of snow and ice surface melt onset (MO), date of pond onset (PO), date of sea ice opening (DOO), and date of sea ice retreat (DOR), and three durations, i.e., melt pond formation period (MPFP, i.e., MO–PO), melt pond extension period (MPEP, i.e., PO–DOR), and seasonal loss of ice period (SLIP, i.e., DOO–DOR), were used. PO ranged from late April in the peripheral seas to late June in the central Arctic Ocean in Bootstrap results, whereas the pan-Arctic was observed nearly 4 days later in NASA Team results. Significant negative trends were presented in the MPEP in the Hudson Bay, Baffin Bay, the Greenland Sea, and Kara and Barents Seas in both results, indicating that the Arctic sea ice undergoes a quick transition from ice to open water, thereby extending the melt season year to year. The high correlation coefficient between MO and PO, MPFP illustrated that MO predominates the process of pond formation.
Figure 1. Map of the Arctic and eight sectors. The land masses shown in dark grey are omitted from the regional analysis.
Figure 2. Flow chart of data processing.
Figure 3. Mean (a) and linear decadal trends (b) of MO for the period of 1982–2017. Trends are calculated only in locations where MO occurs in more than 80% of the years over the entire record. Black points indicate that the trends are significant at the 95% confidence level. The linear trends of the Arctic MO were examined. Decadal trends in MO from 1982 to 2017 were computed and shown in Fig. 3b. The Arctic presented striking negative trends in general. The largest trends were over –6 d/decade in the Kara and Barents Seas and the Arctic Ocean (Table 1), which indicated a shift toward an earlier MO over the last 36 years. The largest downward trend in the Barents and Kara Seas may be partly associated with the decreased cooling efficiency in the Barents Sea over the past decades (Skagseth et al., 2020; Shu et al., 2021). The intensified decrease in sea ice that occurred in Chukchi and Beaufort Seas from 2000 to 2012 was in accordance with the second-largest significant downward trend in the Arctic Ocean (Frey et al., 2015). Notably, a slightly positive trend, which was statistically non-significant in MO with 1.5 d/decade, was observed in the Bering Sea. The accelerated increase in sea ice cover that resulted from lower temperature and cold northerly wind in 2006–2012 may in part explain the later MO in the Bering Sea (Frey et al., 2015).
Figure 4. Mean and linear decadal trends of PO for the period of 1982–2017 were derived by BT SIC, i.e., bPO mean (a) and bPO trend (d), and NT SIC, i.e., nPO mean (b) and nPO trend (e). The differences (BT result minus NT result) between the two estimations (c, f) are shown. Trends are calculated only in locations where PO occurs in more than 80% of the years over the entire record. Black points indicate that the pixels are significant at the 95% confidence level.
Figure 5. Mean and linear decadal trends in DOO and DOR for the period of 1982–2017 derived from BT SIC, i.e., bDOO mean (a), bDOO trend (d), bDOR mean (g), bDOR trend (j), and NT SIC, i.e., nDOO mean (b), nDOO trend (e), nDOR mean (h), nDOR trend (k). The differences (BT result minus NT result) of the four estimations (c, f, i, and l) are also shown. Trends are calculated only in locations where DOO or DOR occurs in more than 80% of the years over the entire record. Black points indicate that the trends are significant at the 95% confidence level.
Figure 6. Mean and linear decadal trends of MPFP and MPEP for 1982–2017 derived from BT SIC, i.e., bDOO mean (a), bDOO trend (d), bDOR mean (g), and bDOR trend (j), and NT SIC, i.e., nDOO mean (b), nDOO trend (e), nDOR mean (h), and nDOR trend (k). The differences (BT result minus NT result) of the four estimations (c, f, i, l) are also shown. Trends are calculated only in locations where MPFP or MPEP occurs in more than 80% of the years over the entire record. Black points indicate that the trends are significant at the 95% confidence level.
Figure 7. Mean and linear decadal trends of SLIP for the period of 1982–2017 derived from BT SIC, i.e., bSLIP mean (a) and bSLIP trend (d), and NT SIC, i.e., nSLIP mean (b) and nSLIP trend (e). The differences (BT result minus NT result) between the two estimations (c, f) are also shown. Trends are calculated only in locations where MPFP or MPEP occurs in more than 80% of the years over the entire record. Black points indicate that the pixels are significant at the 95% confidence level.
Figure 8. Mean annual evolution of the sea ice surface melt in the Arctic from 1982 to 2017. Blue, orange, and gray shaded bars define the mean MPFP (bMPFP), MPEP (bMPEP), and SLIP (bSLIP) obtained from BT SIC, respectively. The beginning and ending points of these bars denote the annual mean MO (bMO), PO (bPO), DOO (bDOO), and DOR (bDOR) derived from BT SIC. The timings determined from NT SIC are presented in curves. The black, green, red, and violet curves noted with filled triangles show annual mean MO (nMO), PO (nPO), DOO (nDOD), and DOR (nDOR), respectively. The span of each two contiguous curves denotes nMPFP, nMPEP, and nSLIP.
DownLoad:
