Development of the Arctic Ocean statistical parameters describing variability of ice conditions for the 20th century

1
Development of the Arctic Ocean statistical
parameters describing variability of ice
conditions for the 20th century

• Vasily Smolyanitsky
• Arctic and Antarctic Research Institute
• St.Petersburg, Russian Federation
• JCOMM Expert Team on Sea Ice (ETSI)

2

• Summary of ice charts data collections within the
  WMO Global Digital Sea Ice Data Bank Project
• Basic statistics describing mean sea ice state
  for the second part of the 20th century
• Extended statistics, including PDFs
  approximation, describing special features of ice
  conditions
• Analysis of temporal variability based on ice
  index data and revealed periods of sea ice
  variability

3
Summary of GDSIDB content
The main five GDSIDB datasets are described in
the table. Included are also the period, covered
by dataset, existence of gaps, sea ice parameters
in dataset and number of charts in SIGRID format
(WMO, 1989). Total number is gt 10000. Samples are
on next slide
? Title of data set Period / periodicity Gaps ? Sea ice parameters Number of charts
1a Arctic and Antarctic Research Institute (AARI) revised ice charts for the Eurasian Arctic 1950-1992 / 10 days, 1997-.. Yes CT, SD, FI gt2000
1b Arctic and Antarctic Research Institute (AARI) ice charts for the Antarctic 1971-1990 / 10 days Yes CT, SD, FI 475
2a National Ice Center (NIC) ice charts for the Northern Hemisphere (northward 39 N) 1972-1994 / 7 days, 1995-.. No CT, SD, FI gt1500
2b National Ice Center (NIC) ice charts for the Southern Hemisphere (southward 50 S) 1973-1994 / 7 days, 1995-.. No CT gt1200
3 Canadian Ice Service (CIS) ice charts for Canadian Arctic (4 sub-regions) 1968-1998 / 7 days, 1999- Yes CT, SD, FI gt3500
4 Baltic Sea Ice Meeting ice charts (Finland - FIMR and Sweden - SMHI) 1960-1979 / 3-4 days No CT, SV, FI 1042
5 Ice charts for the Sea of Okhotsk from the Japan Meteorological Agency (JMA) 1970-2004 / 5 days No CT gt1100
CT total concentration, SD stage of
development and partial concentration, SV ice
thickness, FI fast ice information.
4
Samples of ice charts from national ice services,
historic modern
1. AARI, Russia
4. FIMR, Finland (for BSIM)
5. JMA, Japan
2. NIC, USA
3. CIS, Canada
5

• Arctic total concentration blended data set
  for 1950-1998 based on GDSIDB ice charts

• First version was completed in December 2002
  (thanks to Hadley Center support)
• Blending was done on a 15x15 grid (45-90N) with
  monthly interval averaging
• Additional monthly Northern Hemisphere 1 degree
  ice extent dataset (J.Walsh dataset) included
• Gaps were populated by climatic monthly median
  values
• Result contains -) 26 (GDSIDB), 25 (JW), 49
  (Climate) -(
• Output is user-friendly and includes 2 files
  factual data and flags of origin for each cell
• In 2005 extension of blended dataset was done by
  inserting values (based on AARI data) for
  Eurasian Arctic for 1933-1949 period

6
Spatial statistical analysis (1950-1998)1) sea
ice mean state (blended dataset)
April (median)
Winter (robust) mean
August (median)
Summer (robust) mean
7
2) special features of ice conditions - extended
set of statistics assessed on individual datasets
Statistics for total concentration in summer
period (AARI data)
8
3) skew, curtosis and PDFs (app.by Pirson
curves)
Skew (r3), curtosis (r4), winter
Skew (r3) and curtosis (r4), summer
PDF type f(r3,r4)
PDF app., winter
PDF app., summer
AARI and NIC total concentration data
9
3add) PDFs based on SSMR-SSMI data (app.by
Pirson curves), useful for near NP region
J-type
PDF app., February
PDFs app., May
PDFs app., August
Data SSMR-SSM/I Bootstrap total concentration
daily patterns for 1978-2002
10
Temporal analysis of ice index dataVariability
and linear trends in August for Eurasian
(1900-2003) and Canadian Arctic (1968-2004)
Greenland, Barents, Kara
Laptevs, Eastern-Siberian, Chukcha
Western CA (blue) Eastern CA (green) East Coast
(red) (April) Hudson Bay (light blue)
11
Validity of linear trend coefficients by regions
(95)
12
Wavelet analysis of sea ice extent variations for
Eurasian Arctic Seas (based on 1900-2003 period)
and Canadian Arctic Seas (based on 1968-2004
period) in August (red more ice, blue less
ice)
Greenland
Barents Kara
Laptev
Eastern-Siberian Chukcha
Western CA Eastern CA
Hudson Bay
13
Conclusions

• Ice charts collections and blended datasets
  within the GDSIDB project provide possibility to
  assess various climatic statistics including sea
  ice mean state for the second part of the 20th
  century
• Different statistics may and should be chosen for
  climatic analysis and to underline special
  features of ice conditions, robust and PDFs are
  preferable
• Abnormal types of PDF like U, I, J-form are
  predominant for sea ice total and partial
  concentrations in seasonal cycle, approximations
  of PDFs may be used for classification schemes or
  in modelling
• Linear trend analysis based on ice charts and
  index data gives a typical picture of a negative
  trend from 1900s to 2000, however transition to
  sub-periods or regions gives a much more complex
  picture with coefficients varying in magnitude
  and sign
• There is a strong indication that the sea ice
  variability in at least Eurasian and Canadian
  Arctic has a polycyclic structure with noticeable
  periods of approximately 10, 20 and 60 years and
  periods of longer duration revealed as linear
  trend