INTRODUCTION

Global warming in recent decades has been accompanied by a rapid shrinking of the Arctic sea ice area (SIA). In September, during the seasonal minimum in SIA, the trend of SIA decline was about 11% per decade over the past 40 years [10]. Here, the reduction in the SIA accelerated at the beginning of the twenty-first century, which may be due to a number of positive radiation and dynamic feedbacks in the atmosphere‒sea ice‒ocean system [5]. The current changes have a significant impact on the environment, biodiversity, and economic activity, in particular, contributing to the development of maritime shipping and facilitating the extraction of minerals in the offshore zone of the seas of the Russian Arctic [1, 3].

The dates of closing and ending of ice cover and the duration of the open water period associated with these dates are important characteristics of the recent transformation of the Arctic sea ice. The data on the concentration of sea ice, continuously obtained using satellite microwave sensing from 1979 to the present, allowed us to estimate key dates (see methods) of the seasonal course of the ice cover and the changes in recent decades [14]. In general, in the last forty years there has been an increase in the length of the open water season (LOWS) by about ten days/decade (with a spread from 4 to 14 days/decade) [7].

The dates of closing and ending of ice cover and the duration of the open water period associated with these dates are important characteristics of the recent transformation of the Arctic sea ice. Such assessments are important to ensure operating activities and strategic planning in the oil and gas industry and the development of navigation along the Northern Sea Route [3]. One of the key regions is the Kara Sea, where a number of promising license oil and gas exploration areas are located. This determines the relevance of studies of the predictability of the sea ice conditions on different time scales [6].

The numerical model experiments on the global climate with a given anthropogenic impact on average for an ensemble of models reproduce the total reduction in the sea ice area (SIA) in the Arctic in the period of satellite measurements since 1979. At the same time, ensembles of models of different generations show a significant variation in the rate of changes in SIA that is especially pronounced on a regional scale [4]. In general, it should be noted that the probability of climate models to reproduce the observed changes in September during the last 40 years in each subsequent generation of models has improved [4, 9, 12].

The paper [8] provides estimates of the time limits and the data on variations in the global temperature upon reaching the certain LOWS values, as well as its evolution in the Arctic seas according to the latest generation of CMIP6 models under various scenarios of anthropogenic impact. In general, the LOWS in the Arctic may increase by about 70 and 110 days by the end of the twenty-first century under moderate (SSP245) and aggressive (SSP585) scenarios of anthropogenic impact on the climate, respectively [8].

This article presents an analysis of the spatial distribution of dates of ice cover closing and the duration of the open water period in the seas of the Russian Arctic according to remote sensing data obtained in the years 1981‒2020, an estimate of their reproduction compared to satellite data, and changes in the period of 2061–2080 according to the ensemble of CMIP6 models. In addition, changes in these characteristics are presented for the southwestern part of the Kara Sea, where a number of large natural gas fields are located.

MATERIALS AND METHODS

There are several key dates in the seasonal sea ice (including the beginning and ending, the duration of melt periods, formation, open water), the definition of which depends on the goals of the study and the data used (for example, see [2, 6, 14]). We have analyzed the DOC and LOWS, which were calculated as follows. The period of maximal duration of all the periods with a daily concentration of more than 85% in each year was taken into consideration. Its start date was considered the date of ice cover closing. The duration of the open water period was understood as the number of ice-free days, calculated taking into account days with a daily sea ice concentration of less than 15% during a calendar year. In some years, the upper limit of dates of ice cover closing in some satellite and model grid nodes exceeded 365 (366) days, since the beginning of ice formation may occur at the beginning of the next year and not at the end of this year.

We used daily data on sea ice cover with a spatial resolution of 25 × 25 km over the period of 1981–2020 from the National Snow and Ice Data Center (NSIDC) of the National Oceanic and Atmospheric Administration (NOAA) [11]. These data were obtained using measurements of the passive microwave brightness temperature by the Nimbus-7 satellite and a series of DMSP satellites using SMMR and SSM/I-SSMIS microwave radiometers.

In order to assess possible climate changes in the future, the data of Coupled Model Intercomparison Project CMIP6 [13] was used. The data of 14 models were analyzed, including the daily data on the ice concentration in the historical and scenario experiments with SSP245 and SSP585 (scenarios of moderate mitigation and non-mitigation anthropogenic impact on the climate, respectively). Ten models were used for  the analysis: CESM2-WACCM, CNRM-CM6-1, EC-Earth3-CC, GFDL-CM4, IPSL-CM6A-LR, MIROC6, MPI-ESM1-2-HR, MRI-ESM2-0, NESM3, and NorESM2-MM. The choice of models based on the estimation of the mean square error in reconstructing the seasonal sea ice drift in all seas of the Russian Arctic allowed us to exclude models showing a significant discrepancy with the observational data in the modern period from research.

To eliminate the problem of multiple transitions of the ice concentration during the periods of the beginning of the formation and destruction of the sea ice cover, a run mean filter of the initial data was used for the daily ice concentration data. A comparison of the data averaged using a different sliding window showed that optimal results are achieved when using a 20-day window, which we have used in this study.

DOC and LOWS were calculated for each year of the period 1981–2020 using satellite data and selected model data for all grid nodes of the satellite and model grid, as well as of the period 2061–2080 using the model data.

The ice characteristics for each model were averaged over an ensemble of ten models after calculations. The expected changes in the considered characteristics in the period 2061–2080 compared with 1981–2020 were investigated. It was found that, according to Student’s criterion, they are statistically significant at the level of 0.05 in all grid nodes of the ensemble of models in the study area.

RESULTS

The dates of ice cover closing and the length of the open water period are shown in Fig. 1. The most recent DOC in January and the first half of February in the period of satellite observations are observed in the Barents Sea (Fig. 1a). There is a meridional isochronous structure with rapidly increasing DOC in the easterly direction from the second half of October at the eastern margin to late–early December in the western and central regions in the Kara Sea (East of Novaya Zemlya). Such features are associated with the inflow of warm Atlantic waters inflow into the Barents Sea [15]. In general, the DOC in the Laptev, East Siberian, and Chukchi seas has a zonal distribution with a relatively weak zonal gradient. The ice cover in a wide coastal area in the Laptev and East Siberian seas is established during the second ten-day period of October, on average. The Chukchi Sea is characterized by a later DOC near the Bering Strait (the end of November–the beginning of December), which may be due to oceanic inflow and the influence of the Aleutian minimum.

Fig. 1.
figure 1

The average dates of ice cover closing (decades) in seas of the Russian Arctic over the period of 1981–2020 according to (a) satellite data and (b) data from an ensemble of climate models, as well as their changes (days) in the period in 2061–2080 compared with 1981–2020 according to the (c) SSP 245 and (d) SSP 585 scenarios.

Full size image

The models ensembe mean demonstrate the observed features of the spatial DOC structure (Fig. 1b). At the same time, it should be noted that the DOC is reproduced by an ensemble of models at a later date compared to the satellite data: in the Barents Sea, in late December–early January; in the Kara Sea, from late November to the first half of December in the eastern part to the third ten-day period of October in the western and central areas; in the Laptev and East Siberian seas, in the third ten-day period of October; in the Chukchi Sea, in the first half of December. Such a difference, in particular, can be linked to averaging over an ensemble of models, which makes the spatial structure smoother.

According to model estimates, the beginning of ice cover over the period of 2061–2080 will occur significantly later compared to the period of satellite measurements. As shown in Table 1, DOC in the Russian Arctic  seas will shift by 1.1‒1.5 and 1.6‒2.3 months according to the SSP245 and SSP585 scenarios, respectively. As evidenced from both scenarios, the Kara Sea will be affected by the greatest changes in DOC, on average (Table 1). At the same time, it is assumed that the greatest regional changes in DOC will occur in the northern part of the Barents and Kara seas, south of Franz Josef Land, as well as in the northwestern part of the Kara Sea, and they will amount to 2–2.5 months for the SSP245 scenario (Fig. 1c) and 2.5‒3 months for the SSP585 scenario (Fig. 1d). Thus, the ice cover in the Barents Sea in the second half of the twenty-first century will appear in late January and early February only in a narrow zone in the northern and southern parts of the sea. According to SSP245, the beginning of ice cover in the Kara and Chukchi seas will be shifted to December, and in the Laptev and East Siberian seas, to November. The SSP585 scenario suggests a shift of the DOC to a month later, on average, compared to SSP245 (Table 1, Figs. 1c, 1d).

Table 1. DOC (day number from the beginning of the year) and the LOWS (days), on average, for the seas of the Russian Arctic: according to satellite data and those from the ensemble of climate models over the period of 1981–2020, the scenarios SSP 245 and SSP 585 over the period of 2061–2080; as well as their changes (days) in 2061–2080, compared to 1981–2020 (all changes are statistically significant at the 0.05 level)
Full size table

The analysis of the LOWS from satellite data and model data showed the similarity of their spatial distribution in the period of 1981–2020 (Figs. 2a, 2b). The greatest variability of the LOWS is typical of the Barents Sea. The open water period in the central part of the sea is observed year-round, and its duration in the northern part of the sea is 100 days, on average. The LOWS in the Kara Sea ranges from 50 days in the east to 100 days in the west. The LOWS is the shortest in the Laptev and East Siberian seas: the open water period exceeds 50 days only in the coastal zone. The LOWS in the Bering Strait is 100–150 days.

Fig. 2.
figure 2

The average duration of the open water period (days) in seas of the Russian Arctic, on average, over the period of 1981–2020 according to (a) satellite data and (b) those from an ensemble of climate models, as well as their changes (days) in the period of 2061–2080 compared with the period of 1981–2020 according to the (c) SSP 245 and (d) SSP 585 scenarios.

Full size image

Prognostic estimates of the LOWS over the period of 2061–2080 obtained from an ensemble of climate models indicate its significant increase compared to the period of 1981–2020 throughout the seas of the Russian Arctic. According to the model results, the LOWS will increase most noticeably (by 3–3.4 and 4.5 months according to the SSP245 and SSP585 scenarios, respectively) in the Barents and Kara seas; in other eastern seas, by 2.2‒2.4 and 3.3‒3.6 months according to the SSP245 and SSP585 scenarios (Table 1). As seen in Figs. 2c and 2d, the most noticeable regional changes in the LOWS are expected in the northern Barents Sea (4–5 months and 6–7 months according to the SSP245 and SSP585 scenarios, respectively) and in the northwestern (4.5 months and 6 months according to the SSP245 and SSP585 scenarios, respectively) and southwestern (~4 months and 5–5.5 months according to the SSP245 and SSP585 scenarios, respectively) Kara Sea. Thus, the LOWS in a narrow coastal zone in the northern and southern parts of the Barents Sea in the second half of the twenty-first century may increase, depending on the scenario, up to ten or eleven months per year (Table 1, Figs. 2c, 2d). The LOWS in the Kara and Chukchi seas is expected to be six or seven months according to the SSP245 and SSP585 scenarios, respectively. The expected increase in the LOWS in the Laptev and East Siberian seas can be up to four or five months per year according to the SSP245 and SSP585 scenarios, respectively.

According to model forecasts following from both scenarios of anthropogenic impact (Figs. 2c, 2d), the most noticeable changes of the LOWS in the coastal part of seas of the Russian Arctic in the second half of the twenty-first century will occur in the southwestern part of the Kara Sea. Figures 3a and 3b show that the changes in both analyzed characteristics of the sea ice conditions begin to differ significantly from each other only from the second half of the twenty-first century. Thus, according to the SSP585 scenario, the monotonic increase in DOC and LOWS in the region will continue after ~2050, but will slow down as evidenced by the SSP245 scenario. The accordance of the model estimates (ensemble mean) of DOC and LOWS trends in the period 1981–2020 with estimates based on satellite observations should also be noted.

Fig. 3.
figure 3

Interannual changes in (a) the dates of ice cover closing and (b) the duration of the open water period (b) in the southwestern part of the Kara Sea (56°–74° E, 70°–74° N) according to satellite data (bold brown line) over the period of 1981–2020, and also, data from ten climate models CMIP6 (thin colored lines) and their ensemble (bold blue lines) in the period of 1981–2080 according to the scenarios SSP 245 (solid lines) and SSP 585 (dashed lines).

Full size image

CONCLUSIONS

In general, ten CMIP6 climate models selected for analysis reproduce the spatial structure of dates of ice cover closing and the duration of the open water period in the Russian sector of the Arctic in comparison with satellite sensing data over the period of 1981–2020 quite well.

Prognostic estimates of changes in these characteristics were obtained using moderate mitigation (SSP245) and non-mitigation (SSP585) scenarios of the anthropogenic impact on the climate over the period of 2061–2080. According to model estimates, the beginning of ice cover in the seas of the Russian Arctic will occur about one month and a decade later (SSP245 scenario) and two months later (SSP585 scenario), on average, with the most noticeable changes in the northern parts of the Barents and Kara seas, as well as in the northwestern part of the Laptev Sea (2–2.5 and 2.5‒3 months for the SSP245 and SSP585 scenarios, respectively). At the same time, the ice cover in the Kara and Chukchi seas will appear in December; in the Laptev Sea and the East Siberian Sea it will occur in November and December, according to the SSP245 and SSP585 scenarios, respectively. The LOWS will increase by more than three and four months in Barents and Kara seas according to the SSP245 and SSP245 scenarios, respectively. Similar changes in the eastern seas will be more than two and three months. The open water period in the Kara and Chukchi seas will increase to six months according to the SSP245 scenario and up to seven months according to SSP585. The increase in the LOWS in the Laptev and East Siberian seas is expected to be up to four and five months per year according to the SSP245 and SSP585 scenarios, respectively.

A comparison of the evolution trends of DOC and LOWS calculated based on satellite and model data for the southwestern part of the Kara Sea revealed a good correspondence of the averaged climatic trends of DOC and LOWS in the period of 1981–2020. It should be noted that under two scenarios of anthropogenic impact significant discrepancies between DOC and LOWS will occur only in the second half of the twenty-first century. At the same time, the expected changes until the 2050s are described quite well by linear extrapolation of the trend over the past 40 years.