Sea Ice Climate Normals for Seasonal Ice Monitoring of Arctic and Sub-Regions

Abstract

The climate normal, that is, the latest three full-decade average, of Arctic sea ice parameters is useful for baselining the sea ice state. A baseline ice state on both regional and local scales is important for monitoring how the current regional and local states depart from their normal to understand the vulnerability of marine and sea ice-based ecosystems to the changing climate conditions. Combined with up-to-date observations and reliable projections, normals are essential to business strategic planning, climate adaptation and risk mitigation. In this paper, monthly and annual climate normals of sea ice parameters (concentration, area, and extent) of the whole Arctic Ocean and 15 regional divisions are derived for the period of 1981–2010 using monthly satellite sea ice concentration estimates from a climate data record (CDR) produced by NOAA and the National Snow and Ice Data Center (NSIDC). Basic descriptions and characteristics of the normals are provided. Empirical Orthogonal Function (EOF) analysis has been utilized to describe spatial modes of sea ice concentration variability and how the corresponding principal components change over time. To provide users with basic information on data product accuracy and uncertainty, the climate normal values of Arctic sea ice extents (SIE) are compared with that of other products, including a product from NSIDC and two products from the Copernicus Climate Change Service (C3S). The SIE differences between different products are in the range of 2.3–4.5% of the CDR SIE mean. Additionally, data uncertainty estimates are represented by using the range (the difference between the maximum and minimum), standard deviation, 10th and 90th percentiles, and the first, second, and third quartile distribution of all monthly values, a distinct feature of these sea ice normal products.

Dataset: The dataset DOI: https://doi.org/10.25921/TRXE-M983.

Dataset License: License under which the dataset is made available (CC-BY 4.0).

1. Summary

Arctic sea ice coverage has been undergoing rapid depletion since satellite-based measurements became available in the late 1970s, especially the summer ice coverage, e.g., [1,2,3]. Sea ice decline is most pronounced in the coastal areas such as the Laptev, East Siberian, Chukchi, and Beaufort seas [3,4]. Arctic summer circulation may contribute to regional sea ice anomalies [5]. Spatial sea ice variability may lead to a large spread in climate model sea ice area projections and therefore induces high uncertainty on regional scales [6]. Superimposed on this strong downward trend is the pronounced inter-annual variability and those two characteristics are essentially not correlated [7].

Long-term averages on both regional and local scales are important for monitoring how the current regional and local states depart from their normal conditions to understand the variability and therefore potential vulnerability of marine and sea ice-based ecosystems and habitats. Combined with up-to-date observations and reliable climate projections, normals are essential to business strategic planning, climate adaptation and risk mitigation.

To help establish baselines for the climate state, monthly and annual climate normals of sea ice parameters (concentration, area, and extent) of the whole Arctic Ocean and 15 regional divisions, are derived using monthly data files from the NOAA and National Snow and Ice Data Center (NSIDC) passive microwave sea ice concentration climate data record (CDR) dataset [8]. The method of using the arithmetic average over the last three complete decades (1981–2010) as defined by the World Meteorological Organization [9] is adapted. Basic descriptions and characteristics of the normals are provided. The spatial modes of sea ice concentration variability within the time period are examined utilizing Empirical Orthogonal Function (EOF) analysis. To provide users with basic quality and uncertainty information on the data product, the climate normal values of Arctic sea ice extents (SIE) from the CDR are compared with those of another product from NSIDC and two products from the European Union’s Copernicus Climate Change Service (C3S). The differences between them range from 0.285 to 0.55 (10⁶ km²), which is about 2.3–4.5% of the Arctic SIE mean for the period of 1981–2010. In addition, data uncertainty is represented by using the range (the difference between the maximum and minimum), standard deviation, and 10th and 90th percentiles and quartile distribution of all monthly values.

2. Data Description

2.1. Input Data Description

Monthly sea ice concentration fields from the NOAA/NSIDC CDR are utilized to derive the climate normals presented here. The CDR is a long-term, consistent, satellite-based passive microwave record of sea ice concentration. This sea ice concentration product leverages two well-established concentration algorithms, the NASA Team (NT) and Bootstrap (BT). Both algorithms were developed and produced by the NASA Goddard Space Flight Center (GSFC) [8]. Description and verification of the CDR dataset can be found in [10] and [11], respectively. The CDR data files used in this study are from the version v03r01 [8].

The CDR data files include two primary sea ice concentration (SIC) parameters: the CDR concentration and similar Goddard Merged concentration. We will hereafter refer to them as CDR SIC and GSFC_m SIC, respectively. Two additional GSFC-derived SIC fields are also included in each CDR data file: GSFC-derived NT and BT. As the current CDR SIC only spans the years from 1987 to 2017, which is short of the three full decades required for a WMO-defined climate normal, GSFC_m SIC will be used instead. GSFC_m SIC is derived using the same processing algorithm as that for CDR SIC, but using the manually quality-controlled GSFC-derived NT and BT sea ice concentrations as input data sources [8]. Manual quality-control means sea ice concentration values may be modified manually at the cell level by examining concentration distributions. The approach is subjective and not reproducible. Furthermore, there is currently no quality flag to indicate if the value has been modified manually. All the monthly SIC fields are on the NSIDC polar stereographic grid with nominal 25 km × 25 km grid cells [10].

The monthly Arctic GSFC_m SIE climate normals are compared with that derived from three other sea ice products. The first comparison product is the NSIDC Sea Ice Index (SII) monthly sea ice extent values [12]. The SII monthly sea ice coverage values are computed from the daily SIC values using the NT algorithm product archived at NSIDC [13], which is also the source data for the monthly GSFC-derived NT field in the CDR data files [10].

The two other climate normal comparison products are computed from the ERA-Interim and ERA5 dataset, provided through C3S. ERA-Interim and ERA5 are the global reanalyses produced by the European Centre for Medium-Range Weather Forecasts (ECMWF), e.g., [14,15]. Neither ERA-Interim nor ERA5 contains prognostic parts for sea ice and as such conditions for sea-ice cover (SIC) need to be prescribed from other products. The source sea ice data are from the Met Office Hadley Centre sea ice products, e.g., [16] and the EUMETSAT OSI-SAF satellite product [17]. The sea ice cover in ERA-Interim and ERA5 differs from the original sea ice products for several reasons, including re-gridding and removal of sea ice based on physical consistency checks with temperature [15]. For the calculation of the whole Arctic sea ice extent normal, all grid boxes with sea ice concentration above 15% and north of 20° N are used in the calculation, including the polar hole (ERA-Interim and ERA5) and lake ice (ERA-Interim).

2.2. Data Set Description

For each of the sea ice coverage parameters (concentration, area, and extent), the climate normal data file contains the following fields:

-

Climate normal of the parameter,

-

Minimum, maximum, and standard deviation,

-

10th and 90th percentiles,

-

First, second, and third quartile,

-

Number of valid data records,

-

Quality flag.

Sea ice concentration represents the area fraction of sea ice at each grid cell. The sea ice extent is the area within the 15% concentration contour while the sea ice area is the area-integrated concentration of all grid cells with the sea ice concentration values of 15% or higher over the given region. The North Pole hole region is excluded from the calculation of the sea ice area, which mostly impacts the calculation of the sea ice areas in the whole Arctic and Central Arctic regions.

For the sea ice concentration climate normals, the following additional fields are also included:

-

Percentage of ice presence (sea ice concentration ≥15%),

-

Regional mask.

Monthly and annual sea ice area and extent climate normals are in the ASCII format for each of 16 regions (that is, the Arctic and 15 regional divisions; see

Table 1. Regional names and identifiers (IDs).

Region	Region ID	Region	Region ID
Whole Arctic	Arctic	Barents Sea	BarentsSea
Japan Sea	JapanSea	Kara Sea	KaraSea
Okhotsk Sea	OkhotskSea	Laptev Sea	LaptevSea
Bering Sea	BeringSea	East Siberian Sea	EastSiberian
Hudson Bay	HudsonBay	Chukchi Sea	ChukchiSea
St. Lawrence	StLawrence	Beaufort Sea	BeaufortSea
Newfoundland Bay	NewfoundlandBay	Canadian Archipelago	CanadianArchipelago
Greenland Sea	GreenlandSea	Central Arctic Ocean	CentralArctic

for regional names and identifiers).

Monthly and annual sea ice concentration climate normals, on the other hand, are in the Network Common Data Form (NetCDF) format which is self-describing and machine-readable. The region divisions are denoted by the regional mask field (Figure 1).

3. Methods, Variability, and Uncertainty Estimates

3.1. Approach to Computing Normals

The method of using the arithmetic average over the last three complete decades as defined by the World Meteorological Organization [9] is adapted. The sea ice normals are derived from the monthly GSFC_m SIC values which are the averages of daily GSFC_m SIC values. For the period of 1981–2010, the only period of missing monthly GSFC_m SIC data is December 1987–January 1988, due to missing sensor data. Therefore, there is no time period with 3 consecutive missing SIC records.

To help users better understand spatial and temporal characteristics of the normals and input data, the spatial distribution and variability of annual SIC normal are described in Section 3.2. Spatial modes of variability of monthly SICs used for computing annual SIC normal are depicted utilizing the Empirical Orthogonal Function (EOF) analysis and described in Section 3.3. The temporal distribution of monthly Arctic SIEs for computing the monthly and annual Arctic SIE normals is captured in Section 3.4. The decadal trends of the annual Arctic SIE minimum and maximum are also described. The data uncertainty estimates and normal quality flags are presented in Section 3.5 and product accuracy in Section 3.6.

3.2. Spatial Distribution of Climate Normal of the Annual Sea Ice Concentration

Spatial distributions of the annual SIC climate normal and its standard deviation (STD) are shown in Figure 2a,b. High SIC values are fairly persistent in the central Arctic ocean, which resulted in overall low STD in the region. The differences of the valid data point in the North Pole hole region result from the different pole hole mask sizes (Figure 2c,

Table 2. The North Pole hole attributes and the record periods used during the period of January 1981–December 2010.

North Pole Hole Mask	North Pole Hole Area (106 km2)	North Pole Hole Radius (km)	Latitude (°N)	Total Number of Grid Cells	Record Period Used
SSMIS	0.029	94	89.18	44	January 2008–December 2010
SMM/I	0.31	311	87.2	468	August 1987–December 2007
SMMR	1.19	611	84.5	1799	January 1981–July 1987

).

3.3. Empirical Orthogonal Function (EOF) Analysis

The EOF analysis has been commonly used in climate studies to examine possible spatial modes of variability and how the corresponding principal component time series change over time [18]. EOF analysis has thus been utilized to show the spatial modes of September sea ice concentration variability. In the present investigation, EOF analysis is only performed for grid cells for which at least 10 sea ice concentration values are less than 100% and for which at least 10 values are greater than or equal to 15%. This is because EOF analysis would not be meaningful for points in which there is not sufficient variability across the September series. Thus, we exclude areas where the sea ice concentration is usually/always 100% as well as areas where the sea ice concentration values usually/always indicate ice-free conditions (less than 15% concentration). For September Arctic sea ice concentrations, the first three EOF modes account for 45% of the total variance. The first EOF mode of sea ice concentration shows a distinct spatial pattern with a solid downward trend of 9% per decade, which is significant at the 95% confidence level (Figure 3a,d) and suggests that Mode 1 largely represents a climate change signal. Modes 2 and 3 account for 13% and 11% of the total variance, respectively, and feature less-energetic cross-basin patterns (Figure 3b,c), while the associated time series do not feature a significant trend (Figure 3e,f), suggesting that these modes are likely attributable to natural variability.

3.4. Temporal Distributions and Trends of Arctic Sea Ice Extent

The distinct interannual variability of Arctic SIE and reduction in both annual maximum and minimum can be easily seen in the temporal distribution (Figure 4a). The seasonal cycle of 30-year average of monthly Arctic SIE is shown in Figure 4b with a mean of 12.19 (10⁶ km²) (Figure 4b).

For the Arctic region as a whole and for the period of 1981–2010, the annual SIE maximum and minimum experienced significant downward trends at −2.68% and −12.9%, respectively (Figure 4c,d). These values are similar to those computed by [3] for the period of 1979–2015, which are −2.41% and −13.5% per decade, respectively. The differences primarily reflect the removal of relatively high values for the first two years for the annual SIE maximum and the accelerated annual SIE depletion in the recent years, especially with the record low in 2012, for the annual SIE minimum.

3.5. Data Uncertainty Estimates and Quality Flags

For a given time frequency, that is, monthly or annual, several ways are used to represent data uncertainty estimates. The historical ranges of ice parameters are represented by their maximum and minimum values. The amount of dispersion of the values is captured by the standard derivation. The 10th and 90th percentiles as well as the first, second, and third quartile values are used to represent the frequency distribution of all values. The second quartile represents the median value of all input data, which may or may not be equal to the mean. An example is given for the annual SIE climate normal (Figure 5), which shows that the SIEs are skewed towards higher values. The spatial distributions of the quartiles of the annual sea ice concentration normal are shown in Figure 2d–f. These quartile distributions imply that, in the coastal areas of the Kara, Laptev, East Siberian, Chukchi, and Beaufort seas, the areas adjacent to the central Arctic ocean are more likely to have a higher SIC value than a lower one throughout the 30-year period, demonstrated by the fact that the values of the first quartile in those regions are close to or even slightly higher than that of the mean (comparing Figure 2d to Figure 2a).

Based on guidance by WMO [9], the quality of the sea ice normals has been categorized as one of the following characterizations: Low record, Provisional, Standard, and Complete. They are assigned numerical values of 1 to 4, corresponding to the valid data record number, i.e., N_ice, in years: N_ice < 10, 10 ≤ N_ice < 25 without 3 consecutive missing years, 25 ≤ N_ice < 30 without 3 consecutive missing years, and N_ice = 30, respectively (

Table 3. Flag names, conditions, and values for the sea ice concentration climate normal.

Flag Name	Pole Hole	Lakes	Coastal	Land	Missing Data
Value	−0.05	−0.04	−0.03	−0.02	−0.01
Flag Name	All Water	Low Record	Provisional	Standard	Complete
Condition	SIC = 0	Nice < 10	10 ≤ Nice < 25	25 ≤ Nice < 30	Nice = 30
Value	0	1	2	3	4

). The cell with SIC = 0 throughout the whole period of record is considered to be an “All Water” cell. Quality flag names and values in the sea ice concentration climate normals are captured in Table 3. The same flag values in the CDR data for the North Pole Hole, Lakes, Coastal, and Land are adopted, which are not applicable for the sea ice area and extent normals.

3.6. Data Product Accuracy

Data product accuracy is examined by comparing the Arctic monthly SIE climate normals with other data products. This approach is similar to that used by [19], who estimated an absolute extent uncertainty based on the spread of the extents computed from six different products. Figure 6 shows the time series of the Arctic monthly SIE climate normals (thick blue line with circles) and corresponding range of all the GSFC_m SIC values for each month (light grey shade). Superimposed are monthly SIE climate normal values from SII (green solid line), ERA-Interim (purple dashed line), and ERA5 (red dashed line). Both SII and ERA5 SIE values are systematically lower than those computed from the GSFC_m SIC values, with a difference of 0.55 and 0.5 (10⁶ km²), respectively. A difference of 0.285 (10⁶ km²) was found between GSFC_m and ERA-Interim SIEs, largely due to lower ERA-Interim SIE values during the summer months. The differences can be attributed to both different numbers and locations of grid cells and different concentration values at co-located grid cells. The slightly higher ERA-Interim SIE values during the winter months are mainly the consequence of including lake ice.

The SIE differences between GSFC_m and the other products may be attributed to two main factors. One is the difference in how SIE was computed. The GSFC_m and ERA-Interim SIEs were computed from the monthly SIC fields while the SII and ERA5 SIEs were the monthly average of daily SIE values. Another is the difference in the number of cells with SIC values above the 15% SIC threshold. To help quantify the impact of how SIE was calculated, we have computed the monthly SIE values from the monthly GSFC-derived NT SIC fields included in the CDR data files. Figure 7 shows that while the overall difference between these two products is on the order of 0.55 (10⁶ km²), the impact of using the daily versus monthly SIC fields for computing the SIE values is on the order of 0.38 (10⁶ km²). Using the ERA-Interim data, the impact of using the daily versus monthly SIC fields for computing SIE values was found to be on the order of 0.29 (10⁶ km²), while that of lake ice on the order of 0.17 (10⁶ km²) using the ERA5 data (not shown). Thus, for the Arctic as a whole with a mean SIE value of 12.19 (10⁶ km²), the differences between GSFC_m and other SIEs are in the range of 2.3–4.5%.

4. User Notes

Zero values in SIA and SIE climate normals do not necessarily mean that there is no ice in the region. They actually represent a state that the sea ice concentrations of all the grid cells within the region are less than 15%.