International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 6

Number 4

December 2012

1 INTRODUCTION

The Baltic Sea, even though not large in the global

scale, is an important shipping lane. In winter,

especially in the region of the Gulf of Bothnia,

navigation is seriously obstructed by ice. Operation

of icebreakers to support maritime trade is usually

included in the infrastructure offered by the port

states. Only in Russia private icebreaker services are

operating, as, for example, the Gazprom icebreakers.

On the area of the Gulf of Bothnia – the severe

Baltic Navigation in Ice in the Twenty First

Century

M. Sztobryn

Institute of Meteorology and Water Management - National Research Institute,

Maritime Branch, Gdynia, Poland

ABSTRACT: The Baltic Sea, even though not large in the global scale, is an important shipping lane. In

winter, especially in the region of the Gulf of Bothnia, navigation is seriously obstructed by ice. The aim of

this work was investigation of changes in the intensity of the obstruction by ice, caused by climate change in

coming 90 years o

f the twenty first century. It was one of the first attempts of technical application of the

global climate scenarios effects. It should be stressed, that the presented work results (as application of the

climate scenarios), couldn’t be treated as forecast, as it is only the changes tendency assessment.

The climate changes were examined as the changes in air temperature (adaptation of global emission models

to regional scale) and atmospheric pressure gradient (model ECHAM5), according to three global scenarios

B, A2 and A1B. The number of cases, in which Swedish and Finnish icebreakers assisted the ships, was

assumed as the indicator of navigation obstruction by ice “K”.

The severity of sea ice conditions was presented by the indicator “S”, calculated as the mean value of regional

indices. The “S” is the function of the number of days with sea ice, observed at the stations in the particular

regions and probability of the sea ice appearances.

Relations between sea ice severity index “S” and regional climate

parameters (monthly and annual air

temperature and atmospheric pressure gradient ) were calculated for the calibration period of 1956-2004.

Three models were build: model 1a. “thermal” (the dependence on the mean monthly temperature of July and

December); model 1b. thermal “B” (the function of average annual air temperature and of the mean monthly

December temperature) and model 2 “thermal zonal” (the dependence on the mean temperature of July and

December and zonal component of air pressure gradient). The level of approximation was similar for the

analyzed models (over 0,6). Calculation of the future (in XXI century) changes of indicator “K” was done

according to three scenarios B, A1B and A2. The number of icebreakers’ assistance events should be lower

than the one in the twentieth century. The lowest intensity of this decrease is estimated by model 2 and

scenario A1Br1, the highest one – for model M1b and scenario B r1.

Otherwise, the minimum value, calculated for the scenarios, is higher than in a period of 1956-2004. It means,

that probably, the period with obstruction for navigation in ice could be longer, but not as severe as in the

period of 1956-2004. The obstruction intensity could increase during the 21 century according to scenario

B1r1, the same for empirical model 2. The similar tendency has been shown by scenario A1Br1 and by model

1a. Other models and scenarios estimated the decreasing trend up to 2100

517

region of the Baltic Sea, in general, Swedish and

Finnish icebreakers are in charge.

The aim of this work was investigation of

changes in the intensity of navigation obstruction by

ice, occurring in effect of climate changes in coming

90 years of the twenty first century.

The observed in XX century changes of climate,

especially a rise of the mean air temperature induced

more intensive interest in climate modeling and

forecasting future changes thereof within a long

lasting time period, e.g. by the end of XXI century.

From among many scientific research institutions

engaged in this subject, the most spectacular

achievements (Nobel prize) were gained by IPCC

(Intergovernmental Panel on Climate Change). The

works of IPCC have been published in a form of

reports. According to the reports, it seems to be the

most likely, that a cause of the observed climate

changes in XX century was the anthropogenic

growth of greenhouse gas concentration. The Special

Report on Emissions Scenarios (SRES) has

presented probable scenarios of greenhouse gas

emissions in XXI century. Thus, the predicted gas

emission and a rise of air temperature, calculated on

the grounds of the scenarios, were used as the input

data to the global climate models – the global

circulation models - (among the others: ECHAM

model for Europe and the North Atlantic).

In the research there have been used scenarios

B1, A2 and A1B of greenhouse gases emission,

worked out by IPCC for XXI century. The

differences between the scenarios result from

varying assumptions of the world evolution in XXI

century, due to globalization, economic

development, predicted changes of population and

also abiding by the sustainable development

principles.

Thus, in B1 scenario, an very integrated global

development with simultaneous implementation of

the sustainable development was assumed. It would

cause a rapid economic growth focused on services

and informatics sciences. By 2050 a growth of

population and next its drop are expected.

According to A2 scenario the much less

integrated development was assumed (the

independently developing regions /nations will be

more significant). It will result in fast and

continuous growth of people, free economic

development focused on regional benefits, less and

more diversified development, differentiated abiding

by sustainable development (to the total denial) and

minor changes in technology.

Furthermore, in A1B scenario there was assumed

a rapid economic growth with the balanced

exploitation of all accessible power supply sources

(fossil and non fossil ones) with introducing the

new, more productive technologies as well. It has to

be emphasized that the European Union accepted

A1B scenario as a basis for shaping the energy

policy.

Adaptation of the emission scenarios (A2, B1 and

A1B) results, also the results of the global climate

model ECHAM to the Baltic and Poland conditions

was carried out within the framework of KLIMAT

Project (www.klimat.imgw.pl) by two IMGW-PIB

Departments : Zakład Modelowania Klimatycznego

i Prognoz Sezonowych oraz Centrum Monitoringu

Klimatu Polski

The basis included re-analyzing and downscaling

performed for the selected reference period (1955-

2004) between the global driving factor (the same

climate elements and the same spatial grid as in

SRES and ECHAM models) and the regional one.

The obtained relations between the global and the

regional driving factor were used to create the

regional scenarios for the Baltic Sea and Poland in

XXI century; the global driving factor was

represented by SRES and ECHAM models’ results.

Adaptation of the global models to the Baltic and

Poland conditions and working out the regional

driving factor field are results of the works carried

out by the IMGW PIB and were described, among

the others, in the work by Jakusik (Jakusik at al.

2010a, Jakusik at al. 2010b), also in the reports of

Klimat project presented on the web page

klimat.imgw.pl.

Two elements of the regional driving factor field

were adopted for purposes of the research: monthly

air temperature and the component indicators

(meridian and zonal) of the atmospheric circulation

(monthly, annual). In this case as well, for the

calibration period there were constructed statistic

and empiric models, displaying relationship between

the selected parameters (Baltic sea-ice severity

index and implicit a number of icebreakers

assistance events) and the predictors, which are

characteristic for the regional driving factor field i.e.

the monthly and annual air temperature and the

component indicators (meridian and zonal) of the

atmospheric circulation (monthly, annual). The

obtained statistic and empiric relations and results of

the scenarios for the regional driving factor have

been used for assessment of changes in the XXI

century.

The algorithmic description of methodical actions

performed during work realization is shown in Fig.

(1)

2 DATA

In the calculations 3 types of data were applied:

results of the regional driving factor field, sea-ice

518

severity index and a number of the icebreakers

assistance events occurred in the Baltic Sea.

Regional driving factors

The following elements were subject to

analyzing: monthly air temperatures ( global

greenhouse gases emission scenarios downscaling –

SRES models) and the atmospheric circulation

(result of ECHAM model downscaling) - component

indicators (meridian and zonal) of the atmospheric

circulation (monthly, annual).

Figure 1. The algorithmic description of methodical actions

performed during work realization.

In the following part of the work, dedicated to

modeling of sea-ice conditions for various scenarios,

an applied scenario summary was presented together

with the model name. Thus for example, a name of

model M1 A2 designates M1 model, where the

original source of driving factors had been A2

scenario.

Sea Ice Severity index

Sea ice severity index „S” originally was

constructed for Polish Coastal zone (Southern

Baltic) in 2006 (Sztobryn M. 2006) and next, in

2009, was applied to determine conditions of the

whole Baltic sea ice condition (Sztobryn and others

2008) . The index is based on the probability of ice

occurrence and the number of days with sea ice

observed in the analyzed period in the selected

basins. The ice severity index is given by the

relation

∑













×=

05.0

where

S - severity index,

N - number of days with ice observed during the

winter season at the particular station,

p - probability of ice occurrence at the particular

station, calculated for the analyzed period and

i - number of stations taken into account.

To calculate the Baltic Sea Ice Severity there

were applied the data referring to numbers of days

with ice and its occurrence probability taken from 34

stations from 1955 to 2004 (Western Baltic Sea,

with: Unterwarnow, Warnemünde, Kiel LH;

Southern Baltic Sea, with: Zalew Szczeciński,

Świnoujście, Kołobrzeg, Gdańsk, Zalew Wiślany;

Gulf of Finland, with: Hanko, Russarö, Helsinki.

Harmaja, Helsinki LH, Loviisa, Oerengrund,

Hogland; Sea of Aland and the Archipelago, with:

Maarianhamina, Koppaklintar, Lågskär, Turku,

Bogskär (Kihti), Utö Sea of Bothnia, with: Rauma,

Kylmäpihlaja, Raumanmatala , Norra Kvarken,

with: Haasa, Ensten, Norrskär, Bay of Bothnia, with:

Ajos, Mutkanmatala, Kemi One, Ykspihlaja,

Repskär, Tankar).Variability of sea ice severity

index and possibility of application thereof in

prediction and modeling of influence of sea ice

conditions on navigation, including also icebreakers

activity were presented in 2009 at TRANSNAV

conference (Sztobryn and others 2009).

Icebreakers activity – navigation in sea ice

In this work it was assumed that the indicator of

difficulties “K” in the navigation related to the

occurrence of ice phenomena is the number of

assistance events with Swedish and Finnish

icebreakers, during the ice season. The data

including a number of icebreakers assistance events

in specific seasons come from the annual SMHI

Works: Report of sea ice condition and icebreakers’

activity.

519

3 RELATION BETWEEN THE ICEBREAKERS

ASSISTANCE EVENTS NUMBER (K) AND

SEA ICE SEVERITY INDEX (S)

Obstructions in navigation on the Baltic Sea while

ice cover occurrence were represented with the

indicator of difficulties in navigation, which was

equal to a number of Swedish and Finnish

icebreakers assistance events within a specific ice

season. There has been analyzed and next

constructed the mathematic model.

The relationship between ice condition

(represented by “S” index) and navigation condition

in ice (by number of icebreakers’ assists) was

formulated by the exponential regression method.

The number of icebreakers assists, recorded in

1956-2004 period, was given by the formula 1 :

𝐾 = 𝑎 ∗ exp (𝑏 ∗ 𝑆) (1)

where:

K – calculated number of assistance events,

S – indicator of the Baltic sea ice severity (mean),

a,b, - numerical coefficients

Comparison of the course of the calculated

number of icebreakers assistance events and the real

number thereof in 1956-2004 is shown on Fig. (2).

The black, thin line stands for the real number of

icebreakers assistance events, whereas the grey thick

one represents the number of events calculated using

the formula.

Figure 2. Comparison of the calculated and the real numbers of

icebreakers assistance events in 1956-2004 period.

Differences between the ”K” indicator real value

(a number of assistance events) and the one

calculated using formula 1 ,- are specially visible for

calculations performed for 1995/96 season. However

it is characteristic for the mentioned season that then

the ice conditions were very differentiated in various

water areas. The south and west Baltic ice severity

was similar to the Gulf of Bothnia; in this area

neither Finnish nor Swedish icebreakers operated.

Moreover, in March, meteorological conditions in

the Gulf of Bothnia proved to be highly changeable

(drift ice in several hours displaced between the

coastal zones of Sweden and Finland).

The correlation coefficient between the real

number of icebreakers assistance events and the one

calculated on the basis of the formula equals to 0,88.

4 ”S” AND ”K” INDICATORS DEPENDENCE

ON CLIMATE CONDITIONS

Functional dependencies between ”S” ice index and

the regional driving factor (the data from re-analysis

for the same climate conditions – the parameters and

spatial location – as the greenhouse gases emission

scenarios results and ECHAM model for the XXI

century) were analyzed and processed for the

calibration period of 1956-2004.

Primary, the influence of climate was represented

by 129 parameters, affecting formation of sea ice in

the Baltic Sea; these parameters were taken into

account for the calibration period of 1956-2004.

They were mainly the annual and monthly average

air temperatures and their combinations, also the

component indicators (meridian and zonal) of the

atmospheric circulation (monthly, annual).

The next step of this work was to formulate the

relationship (as the function) between ice severity

indicator “S” and climate parameters. Construction

of the optimal predictors set has been made on the

basis of the carried out analysis results: correlation

(statistically significant at 95%; level: the highest

value of the correlation coefficient between the

predictors and response variable and the lowest

between predictors), genetic algorithm (the longest

“survival” of the predictors) and model sensitivity

tests (better/worse work of the model with/without

analyzed predictor). The researches proved that the

sea-ice severity indicator ”S” is the most sensitive to

the mean monthly air temperatures in July and

December and the mean annual air temperature as

well as the zonal component of the atmospheric

circulation. Therefore these parameters were used in

constructing the models.

Relations between the parameters were calculated

for the calibration period of 1955-2004 by

multiplying linear regression for 3 types of the

models.

− “thermal” (the dependence on the mean monthly

temperature of July and December further called

M1a model ;

− thermal “B” (function of the average annual air

temperature and of the mean monthly December

temperature)- further called M1b model

500

1000

1500

2000

2500

3000

3500

4000

4500

1956/57

1959/60

1962/63

1965/66

1968/69

1971/72

1974/75

1977/78

1980/81

1983/84

1986/87

1989/90

1992/93

1995/96

1998/99

2001/02

2004/05

"K"

number of assists

years

real " K"

calculated "K"

520

− c “thermal-zonal” (dependence on the mean air

temperature of July and December and the zonal

component of the atmospheric circulation) - fur-

ther called M2 model

Comparison of the data from the period of

observation (1956-2004) with the calculation results

proved a statistic consistency, characterized by the

following correlation coefficients: 0,6 (model 1a);

0,64 – model 1b and model 2.

To enable potential changes of navigation

conditions in ice (represented with ”K” indicator) in

the XXI century there have been used combinations

of formula 1 and processed dependencies between

“S” and the climate parameters (items a, b and c)

5 RESULTS

To predict changes of the sea ice severity ”S”

indicator value in the XXI century the processed and

presented in the section 3 models have been used;

also to settle a number of assistance events “K” –

formula1.

The climate parameters, which the models were

based on, have been taken from calculations of the

IMGW PIB Climate Deparments and they are an

effect of adaptation of three global climate scenarios

B, A1B and A2 run1 and also ECHAM–5 to the

regional conditions.

It means that each of the models (M1a, M1b and

M2) was calculated for 3 scenarios (B, A1B and A2

run1 and also ECHAM–5). To differentiate

particular models and scenarios the following

concept of designation was adopted:

− M1a Br1 – stands for the model based on the de-

pendence “S” on the mean monthly temperature

of July and December) reckoned using the data

taken from the scenario B run 1,

− M1a A1Br1 – the model based on the dependence

“S” on the mean monthly temperature of July and

December) calculated using the data taken from

the scenario A1B run 1,

− M1a A2r1 – the model based on the dependence

“S” on the mean monthly temperature of July and

December) calculated using the data taken from

the scenario A2 run 1,

− M1b Br1 is a model based on dependence be-

tween S and the average annual air temperature

and of mean monthly December temperature cal-

culated using the data from the scenario B run 1,

− M1b A1Br1 is a model based on dependence be-

tween S and the average annual air temperature

and of mean monthly December temperature cal-

culated using the data from the scenario A1B

run 1,

− M1b A2r1 is a model based on dependence be-

tween S and the average annual air temperature

and of mean monthly December temperature cal-

culated using the data from the scenario A2 run 1,

− M2 Br1 designates a model based on dependence

between S on the mean air temperature of July

and December and the zonal component of at-

mospheric circulation, calculated using the data

from the scenario B run 1,

− M2 A1B designates a model based on depend-

ence between S on the mean air temperature of

July and December and the zonal component of

atmospheric circulation, calculated using the data

from the scenario A1B run 1,

− M2 A2 designates a model based on dependence

between S on the mean air temperature of July

and December and the zonal component of at-

mospheric circulation, calculated using the data

from the scenario A2 run 1,

The results of calculation carried out for these

types of models (3 models for 3 different scenarios)

for a period of next 20 years (2011-2030) are

presented in Table 1 with comparison thereof with

the values obtained for 20 years of 1971-1990

period.

Values of the following parameters: standard

error, standard deviation and range, are significantly

lower in 2011-2030 period than in the reference

period. However, the mean, median and minimum

values are in 2011-2030 period higher than in the

reference period. The Kurtosis value for M2 A1Br1

model is very close to the value calculated for the

reference period, as is the skewness, calculated on

the basis of M1b A2r1 and M2B1 models. A

maximum value calculated on the basis of M1b

A2r1 model is close to the maximal value of the

reference period (4,28 and 4,79).

It means that concentrations of „S” values around

the mean value are similar for the both periods under

consideration (model M2 A1Br1), as well as

asymmetry of calculations of M1b A2 r1 and M2 B1

models are. The values of the mean, median and

minimum are higher while the maximum value is

comparable, what indicates that in coming years

there are expected (according to the assumed scenar

ios) more winters with a small number of days with

ice (but still with ice) than in the reference period.

The course of “S” indicator in period 2011-2030,

calculated on the basis of scenario A2r1, together

with comparison with the last climate reference

period is shown on Figure 3.

Process variation of “S” indicator in the twenty

first century was estimated on the basis of three

models, each for three scenarios. Results of these

calculations for scenario A1B are presented in

Figure (4).

521

Table 1. Statistical parameters of „S” indicator for 2011-2030 and the reference period 1971-1990.

Period

1971-1990

2011

-2030

Input Scenario

real

M1a

Br1

M1b

B1r1

M1a

A1Br1

M1b

A1Br1

M1a

A2r1

M1b

A2r1

Mean

1,92

2,34

2,53

2,14

2,45

2,50

2,17

2,53

2,91

2,51

Median

1,52

2,26

2,48

2,13

2,54

2,59

2,14

2,54

2,83

2,62

Standard deviation

1,45

0,54

0,76

0,55

0,50

0,58

0,45

0,47

0,64

0,54

Kurtosis

-0,46

0,32

-0,76

1,18

-0,24

1,21

-0,43

0,68

0,14

-1,09

Skew/skewness

0,82

0,29

0,10

0,79

-0,45

-1,04

0,17

0,61

0,72

-0,27

Minimum

0,11

1,20

1,34

1,19

1,41

1,05

1,28

1,65

1,86

1,60

Maximum

4,79

3,53

3,78

3,60

3,35

3,43

3,01

3,60

4,28

3,35

In the course of the ice severity indicator,

calculated for the twenty first century using three

scenarios, one can see the similar characteristics,

recognized for period of 2011-2030.

The estimated number of icebreakers’ assistance

events was calculated using formula 1 for each of 3

models of “S” for 3 climate scenarios. Numbers of

the estimated occurrence of obstruction for

navigation in ice are shown in figures 5-7. Model 1a

is represented by thick, black line, model 1b by

drops and model 2 by thin dark line.

According to all three scenarios, a number of

icebreakers assistance cases should be lower than in

the twentieth century. The lowest intensity of such

decrease is estimated using model 2 and scenario

A1Br1, the highest one – using model M1b and

scenario B r1.

Otherwise, the minimum value, calculated for the

scenarios, is higher than in 1956-2004 period. It

means that probably the period, when obstructions

for navigation in ice occur, could be longer, but not

so severe as it happened in the period of 1956-2004.

According to scenario B1r1 as well as the empirical

model 2, intensity of such obstruction could increase

during the 21 century. The same tendency is shown

in scenario A1Br1 and using model 1a. The other

models and scenarios have proved a decreasing trend

up to 2100.

Figure 3. Comparison of the sea ice severity indicator

calculated for the reference period and estimated for 2011-2030

period according to A2r1 scenario.

Figure 5. The changes of the estimated number of icebreakers’

assistances in the twenty first century, calculated using B1

scenario and three empirical models

(M1a B1r1 M1b

B1r1, M2 B1r1)

Figure 6. Changes of the estimated number of icebreakers’

assistance events in the twenty first century, calculated using

A1B scenario and three empirical models (M1aA1Br1,

M1bA1Br1, M2A1Br1)

522

Figure 7. Changes of the estimated number of icebreakers’

assistance events in the twenty first century, calculated using

A2 scenario and three empirical models (M1a A2r1 M1bA2r1,

M2A2r1)

6 REMARKS

The aim of this work was investigation of changes in

intensity of obstruction by ice occurrence

(represented by the number of Swedish and Finnish

icebreakers’ assistance events, forced in effect of the

climate change in coming 90 years of the twenty

first century. The estimation was made under a

hidden assumption that technical parameters of

icebreakers and merchant ships would not change in

the twenty first century, the same as intensity of

Baltic navigation in ice.

It has been one of the first probes of technical

application of the global climate scenarios changes.

It should be stressed, that the presented work results

(as application of climate scenarios), cannot be

treated as a forecast, as they are only used to

estimate the changes tendency.

ACKNOWLEDGEMENTS

This work has been achieved through Project

KLIMAT Number POIG.01.03.01-14-011/08-00

(“The impact of climate change on the environment,

economy and society”) supported by the Programme

Innovative Economy under the National Strategic

Reference Framework, which is co-financed with

EU resources.

The project is realized by the Institute of

Meteorology and Water Management (Poland). The

authors wish to thank prof. dr. Miroslaw Mietus

(Institute of Meteorology and Water Management –

National Research Institute) for valuable suggestions

and participation in very useful discussion, also

workers of IMGW PIB E.Jakusik MSc and

R.Wójcik MSc for their support.

REFERENCES

Jakusik E. , Marosz M., Pilarski M., Mietus M, 2010 :Wpływ

pola barycznego na wysokość falowania wiatrowego w

południowej części Morza Bałtyckiego. Materiały Geo-

Sympozjum Młodych badaczy Silesia, 1-14,

Jakusik E, Wójcik R., Biernacik D., Miętus M., 2010: Wpływ

zmian pola barycznego nad Europą i Północnym

Atlantykiem na zmiany średniego poziomu Morza

Bałtyckiego w strefie polskiego wybrzeża , w Woda w

badaniach geograficznych, UJK Kielce, 59-74

Kalnay et al., 1996, The NCEP/NCAR 40-year Reanalysis

Project, Bull. Amer. Meteor. Soc., 77, 437-471

Mietus M., 1999, The role of atmospheric circulation over

Europe and North Atlantic in forming climatic and

oceanographic conditions in the Polish coastal zone,

Materialy Badawcze IMGW, seria Meteorologia, 29, 157

pp. (in Polish),

Mietus M., (main editor), 2009: Report of KLIMAT project,

www.klimat.imgw.pl,

Mietus M., ., (main editor) 2010 Report of KLIMAT project,

www.klimat.imgw.pl,

Schmelzer N., Seina A., Lundqvist J.E., Sztobryn M., 2008,

Ice, in: State and Evolution of the Baltic Sea, 1952-2005,

Wiley & Sons, 199-240,

Storch von H., Zwiers F., 2001, Statistical analysis in climate

research, Cambridge Univ. Press, 499 pp.

Sztobryn M.,2006, Sea ice condition – Polish coastal waters

1955-2005, IMGW, Report of Project DS.-H7, (in Polish)

Sztobryn M., Schmelzer N., Vainio J., Eriksson P.B., 2009.

„Sea Ice Index”. Report Series in Geophysics, University of

Helsinki, No 61, 82-91, Helsinki 2009

Sztobryn M., R.Wójcik, 2010, Impact of climate change on the

Baltic sea ice conditions, 20th IAHR International

Symposium on Ice, Finland,

Werner P. C., Storch von H., 1993, Interannual variability of

Central Europe mean temperature in January-February and

its relation to large-scale circulation, Climate Research 3,

195-207

523