contents.gifindex.gifprev0.gifnext1.gif

AARI Sea Ice Chart statistics

General description

Technical description - Total ice concentration and ice thickness

Technical description - Old ice concentration

Technical description - Fast ice

General description

These files were prepared by INSROP project I.3.1 as part of the 1995 project work. Some results are shown in INSROP Working paper no. 45 (1996).

The files include results from a statistical analysis of the AARI sea ice chart database in SIGRID format. File names starting with SY include statistics from analysing data from a single year, while file names starting with MY include statistics from analysing all available data. The remaining part of the file names show month (abbreviated to 3 characters) and year (two last digits and only for SY-files) and a letter (a, b, or c) indicating type of data, where:

a - Total ice concentration and ice thickness

b - Old ice concentration

c - Fast ice

Files of type b and c include only data points where old ice or fast ice was observed.

Data acquisition

The AARI sea ice charts are based on airborne visual and SLAR data, and satellite data. Except for in 1972, the charts were split into western and eastern charts. These 'twin' charts may be a few days apart. The charts should in principle be issued with a 10-day interval (3 per month), but in some 10-day periods only the western or the eastern chart exists, and there are also a number of 10-day periods where no charts are available. The sea ice charts cover the period 1967-90.

The sea ice charts have been digitised into SIGRID format at the Arctic and Antarctic Research Institute (AARI) in St.Petersburg, Russia. The complete data set was acquired at SINTEF from the World Data Center-A (WDC-A) for Glaciology [Snow and Ice] maintained at the National Snow and Ice Data Center (NSIDC) through World Wide Web (http://nsidc.colorado.edu/NSIDC/

wdc-a.html) as three compressed UNIX tar files (aari-72.tar.z, aari-east.tar.z, aari-west.tar.z).

Data processing and analysis

As this data set is the most comprehensive and detailed sea ice data set available for Project I.3.1 in 1995, it was necessary to use the data set to derive a number of sea ice parameters, such as total ice concentration, ice thickness, concentration of old ice (second-year and multi-year ice) and presence of fast ice. The purpose of the data processing and analysis was to establish a statistical sea ice data set for NSR-related evaluations. The statistical data should be provided as monthly statistics, based on all available ice charts within each month, for each single year and for all years (multi-year statistics). The resulting data sets should also be prepared for use in INSROP GIS.

At the data set location (NSIDC), also software utilities to extract sea ice information from a SIGRID-formatted data set were available. The C-program (strip_geog.c) was the basis for developing the data processing and analysis software at SINTEF. The first step was to modify this program to export data in a format suitable for import into ARC/INFO, as ArcView can use ARC/INFO-formatted data directly and also because the initial plan was to run the analysis in the ARC/INFO GRID module. The format chosen was the ARC/INFO ASCIIGRID format. To display selected ice charts in INSROP GIS (ArcView), software utilities to convert a SIGRID ice chart to an ARC/INFO point cover were developed. Hence, at this point there were two options for displaying the original sea ice charts in INSROP GIS.

The next step was to derive the ice parameters to be statistically analysed from the original attributes of the data points in each sea ice chart. The SIGRID data comprise unique codes for each attribute, but as some codes may represent ranges rather than discrete values, this calls for special treatment to derive discrete values for use in the statistical analyses. The statistical parameters to be provided include minimum, mean, median and maximum values, and probability for a certain criteria (e.g. ice concentration > 70%) to be fulfilled.

The third step was to create single-year monthly statistics. This was originally planned to be handled by ARC/INFO (aml-code), but because we encountered problems with handling NODATA values in the statistical analysis and experienced large requirements for processing time and temporary storage, a change in strategy was made. Therefore a FORTRAN program (aaristats_sy.f) was developed to prepare monthly statistical files for each parameter for each year and the strip-geog.c program was modified further to be a subroutine of the FORTRAN program. By specifying ice parameter to be derived, threshold value (for ice parameters where this is required) and input file name, the modified C-program returns the derived ice parameter in a fixed grid. For grid cells where the source ice chart has no data value, a NODATA value (101) is used. The FORTRAN program handles the statistical analysis and stores the results on files in the ARC/INFO ASCIIGRID format. In this process NODATA values are excluded and the mean monthly value is derived by taking the sum of the real data values and dividing by the number of real data values (excl. NODATA values).

The fourth step was to create the multi-year statistics. The aaristats_sy.f FORTRAN program was used as basis to develop the aari_sy2my.f FORTRAN program. This program reads the single-year monthly files, runs the statistical analyses, and stores the multi-year statistics on monthly files in the ARC/INFO ASCIIGRID format for each parameter. In this process NODATA values are excluded and each single-year monthly parameter grid is given equal weight, that is, the number of data sets within each month is not considered. This avoids getting biased results due to some years having far more original data sets than others, but it also means that if a month in a year had only one (or two) ice chart value(s) at a given location, this value will be used as representative for the entire month.

The last step is to prepare the statistical grid files for use by INSROP GIS. This was achieved by developing a FORTRAN program that, for each month, reads the parameter grids for each time period (single-year months and multi-year months), and stores the data in the INSROP GIS Point ASCII import format. To save storage space and increase performance, the data are split into several files for each time period (YY = year [67-90], MM = month [01-12]):

File 1: Total ice concentration and ice thickness (myMMa00.pos, syYYMMa0.pos)

File 2: Old ice concentration (myMMb00.pos, syYYMMb0.pos)

File 3: Fast ice (myMMc00.pos, syYYMMc0.pos)

File 4: Data coverage (myMMd00.pos, syYYMMd0.pos)

Files of Type 1 and 4 include all points with at least one real data value, while files of Type 2 and 3 comprise only points where presence of old ice (Type 2) or fast ice (Type 3) is observed.

Implementation in INSROP GIS

Due to the size of the statistical data set, only the multi-year data subset and selected single-year data subsets (1983 and 1990) are implemented in INSROP GIS. The data were prepared in the INSROP GIS Point ASCII import file format and implemented using the "Theme - New Theme From ASCII File" menu option in the View window. The data set is called AARI Sea Ice Statistics.

Deriving ice concentration values

Ice concentration codes are used for total ice concentration and partial ice concentrations. To solve the problems of some SIGRID ice codes representing ranges, each ice concentration code is assigned a minimum, mean and maximum ice concentration value. The mean value is the average of the minimum and maximum values, and the median value is the median of the same average values. For codes representing discrete ice concentrations, these values are all equal, but for codes representing ranges, the range limits are used as the minimum and maximum values. Table 8.2 shows how the SIGRID codes are recoded (incl. special codes).

Table 8.2 Conversion of SIGRID codes to ice concentration values.

Code
Explanation
Minimum
Mean
Maximum
00
Ice free
0
0
0
01
Open water (< 1/10)
1
5
9
02
Bergy water (1)
2
2
2
04
Fast ice
100
100
100
10
1/10
10
10
10
.
.



13
1/10 - 3/10
10
20
30
.
.



71
7/10 - 10/10
70
85
100
.
.



90
9/10
90
90
90
91
more than 9/10, less than 10/10
91
95
99
92
10/10
100
100
100
99
Unknown (2)
101
101
101
102
Land (2)
102
102
102

(1) Artificial low concentration value to indicate presence of ice

(2) Values greater than 100 are ignored in the statistical analysis

When deriving the probability of ice concentration above a given threshold, the 'range codes' also require special treatment. If, for a given code, the minimum ice concentration value is above the threshold, the probability is 100 percent. Similarly, if the maximum ice concentration value is below the threshold, the probability is 0 percent. However, if the threshold concentration is within the ice concentration range of the given code, a uniform distribution of ice concentrations within the range is assumed, and the fraction of the ice concentration range above the threshold is taken as the probability; e.g. it is assumed that for areas where the ice concentration is coded as 4/10 - 6/10, there is a 50 per cent probability to encounter ice concentrations above 5/10.

Deriving ice thickness values

Ice thickness codes may represent ice thickness or stage of development. To solve the problem of some SIGRID ice codes representing ranges, each code value is assigned a minimum, mean and maximum ice thickness value. The average value is the average of the minimum and maximum values. For codes representing discrete ice thicknesses, these values are all equal, but for codes representing stage of development, associated ice thickness range limits are used as the minimum and maximum values. Table 8.3 shows how the SIGRID codes are recoded (incl. special codes).

Table 8.3 Conversion of SIGRID codes to ice thickness values

Code
Explanation
Minimum
Mean
Maximum
00
Ice free
103
103
103
01
Ice thickness in cm
1
1
1





50
Ice thickness in cm
50
50
50
51
Ice thickness in 5 cm intervals
55
55
55
.
.



60
Ice thickness in 5 cm intervals
100
100
100
61
Ice thickness in 10 cm intervals
110
110
110
.
.



70
Ice thickness in 10 cm intervals
200
200
200
71
Ice thickness in 50 cm intervals
250
250
250
.
.



74
Ice thickness in 50 cm intervals
400
400
400
75
Ice thickness in 100 cm intervals
500
500
500
.
.



79
Ice thickness in 100 cm intervals
900
900
900
80
No stage of development (1)
103
103
103
81
New ice
1
15
30
82
Nilas, ice rind less than 10 cm
1
5
9
83
Young ice
10
20
30
84
Gray ice
10
13
15
85
Gray-white ice
15
23
30
86
First year ice
30
115
200
87
Thin first year ice
30
50
70
88
Thin first year ice stage 1
30
40
50
89
Thin first year ice stage 2
50
60
70
91
Medium first year ice
70
95
120
93
Thick first year ice
120
160
200
95
Old ice
120
280
420
96
Second year ice (2)
120
185
250
97
Multi year ice (3)
240
330
420
98
Glacier ice (4)
104
104
104
99
Unknown (4)
101
101
101

1) Ice free (zero ice thickness) codes are ignored in the statistical analysis of ice thickness

2) Thicker than thick first year ice and less than 250 cm (Romanov, 1993)

3) According to Romanov (1993)

4) Values 101-104 are ignored in the statistical analysis of ice thickness

At any given ice chart location, there may be information on the thickest, second thickest and third thickest ice. The minimum ice thickness at a location is the minimum thickness registered, while the maximum ice thickness value is the maximum thickness of the thickest ice. When calculating the average ice thickness, the average thickness of all thickness registrations are weighted by the fraction of average associated partial ice concentration to the average total ice concentration. The mean and median ice thicknesses are derived from the average ice thicknesses.

When deriving the probability of ice thicknesses within a given ice thickness range, the same methodology as used to derive probability of ice concentrations above a given threshold is employed. However, as the probabilities are to be valid for a thickness range, not just above a thickness threshold, the fraction of the observed thickness range being within the specified analysis thickness range multiplied with the associated partial ice concentration, is used as the probability percentage.

Deriving old ice concentration values

Information on ice types is included in the Stage of development codes. As the term old ice includes both second year ice and Multi year ice, all partial ice concentrations associated with 'Stage of development' codes 95, 96 and 97, are counted as an 'old ice' concentration value (as specified in Table 8.1). In addition, the partial concentrations of ice with codes representing ice thickness above 200 cm (codes 71-79) are counted as old ice concentration values. All partial 'old ice' concentration values are summarised into one old ice concentration value for each point.

When deriving the probability of old ice above a given threshold, the same methodology as described for total ice concentration is employed, with the additional requirement that partial ice concentration ranges are involved rather than one total ice concentration range.

Deriving fast ice concentration values

Fast ice is shown as code value 08 in the Form of ice codes. For points with Form of ice code equal 08 (Fast ice), the ice concentration is set to 100 (ref. Table 8.1). For other code values (except the unknown code value: 99) the fast ice concentration is set to zero. For each location, the probability of fast ice within a time period is calculated as the percentage of fast ice data values out of the total number of data values with 'Form of ice' code different from unknown.

Technical description - Total ice concentration and ice thickness

Shapefile name: my_apr_a.shp (for example)

Path: <NSR_DATA>\icesnow\aari_sta

GeoDataset type: Shapefile with Point features.

Coordinate system: Latitude/longitude in decimal degrees

* My_apr_a.shp

11288 Points, 17 descriptive fields.

Fields: [<Name>] -- <Alias> (type of field)

[Id] -- "Point #" (Numeric, no decimals)

[Nt_val] -- "N" (Numeric, no decimals)

Number of values

[Ctmin] -- "Ctot, min" (Numeric, no decimals)

Minimum total concentration of sea ice, %

[Ctmean] -- "Ctot, mean" (Numeric, no decimals)

Mean total concentration of sea ice, %

[Ctmed] -- "Ctot, median" (Numeric, no decimals)

Median total concentration of sea ice, %

[Ctmax] -- "Ctot, max" (Numeric, no decimals)

Minimum total concentration of sea ice, %

[P_ct10_] -- "P(Ctot > 10%)" (Numeric, no decimals)

Probability of total ice concentration > 10%, %

[P_ct40_] -- "P(Ctot > 40%)" (Numeric, no decimals)

Probability of total ice concentration > 40%, %

[P_ct70_] -- "P(Ctot > 70%)" (Numeric, no decimals)

Probability of total ice concentration > 70%, %

[Thmin] -- "Ice thickness, min" (Numeric, no decimals)

Minimum ice thickness, cm

[Thmean] -- "Ice thickness, mean" (Numeric, no decimals)

Mean ice thickness, cm

[Thmed] -- "Ice thickness, median" (Numeric, no decimals)

Median ice thickness, cm

[Thmax] -- "Ice thickness, max" (Numeric, no decimals)

Maximum ice thickness, cm

[P_th70_120] -- "P( 70 < Th < 120cm)" (Numeric, no decimals)

Probability of ice thickness between 70 and 120 cm, %

[P_th120_20] -- "P(120 < Th < 200cm)" (Numeric, no decimals)

Probability of ice thickness between 120 and 200 cm, %

[P_th200_] -- "P( Th > 200cm)" (Numeric, no decimals)

Probability of ice thickness greater than 200cm, %

[Calc_fld] -- "Calc_fld" (Numeric, no decimals)

Field for storage of calculated values

Technical description - Old ice concentration

Shapefile name: my_apr_b.shp (for example)

Path: <NSR_DATA>\icesnow\aari_sta

GeoDataset type: Shapefile with Point features.

Coordinate system: Latitude/longitude in decimal degrees

* My_apr_b.shp

7936 Points, 10 descriptive fields.

Fields: [<Name>] -- <Alias> (type of field)

[Id] -- "Point #" (Numeric, no decimals)

[Noi_val] -- "N" (Numeric, no decimals)

Number of old ice data points

[Coimin] -- "Cold_ice, min" (Numeric, no decimals)

Minimum old ice concentration, %

[Coimean] -- "Cold_ice, mean" (Numeric, no decimals)

Mean old ice concentration, %

[Coimed] -- "Cold_ice, median" (Numeric, no decimals)

Median old ice concentration, %

[Coimax] -- "Cold_ice, max" (Numeric, no decimals)

Maximum old ice concentration, %

[P_coi10_] -- "P(Cold_ice > 10%)" (Numeric, no decimals)

Probability of old ice concentration > 10%, %

[P_coi40_] -- "P(Cold_ice > 40%)" (Numeric, no decimals)

Probability of old ice concentration > 40%, %

[P_coi70_] -- "P(Cold_ice > 70%)" (Numeric, no decimals)

Probability of old ice concentration > 70%, %

[Calc_fld] -- "Calc_fld" (Numeric, no decimals)

Field for storage of calculated values

Technical description - Fast ice

Shapefile name: my_apr_c.shp (for example)

Path: <NSR_DATA>\icesnow\aari_sta

GeoDataset type: Shapefile with Point features.

Coordinate system: Latitude/longitude in decimal degrees

* My_apr_c.shp

2819 Points, 4 descriptive fields.

Fields: [<Name>] -- <Alias> (type of field)

[Id] -- "Point #" (Numeric, no decimals)

[Nfi_val] -- "N" (Numeric, no decimals)

Number of fast ice data points

[P_cfi10_] -- "P(Cfast_ice > 10%)" (Numeric, no decimals)

Probability of fast ice, %

[Calc_fld] -- "Calc_fld" (Numeric, no decimals)

Field for storage of calculated values