International Journal

on Marine Navigation

and Safety of Sea Transportation

Volume 3

Number 3

September 2009

321

1 INTRODUCTION

1.1 Background

The Baltic Sea is the world’s largest brackish body

of water. It is designated as a PSSA (Particularly

Sensitive Sea Area). More than 2000 large ships in-

cluding large oil tankers are at any given time in the

Baltic Sea (HELCOM 2005, Rytkönen et al. 2002).

Maritime transport adversely affects different risk

receptors in various forms and degrees of extents.

Increasingly large amounts of different types of dan-

gerous goods, including oil and oil products, gases

and a wide range of chemicals, transported and han-

dled in the BSR (Baltic Sea Region) (estimated be-

tween 300-1000 million tons per year) and accidents

involving these goods are concerning issues for the

countries in the region (TSE 2006). The most recent

major oil spills that have occurred in the region are

the cases of the m/v “Fu Shan Hai” (2003) (1200

tons of oil spilt) and the m/v “Baltic Carrier” (2001)

(2700 tons of oil spilt). The costs of oil spills report-

ed yearly and the worst-case scenarios in Öresund

are respectively estimated $223,500 and between

$150-300 million (Mullai & Paulsson 2002).

The DaGoRus project (Safe and Reliable

Transport of Dangerous Goods in the Russian-EU

Logistics Chain) is an European Union (EU)/Tacis

project dealing with safe and reliable transport

chains of dangerous goods. The project consists of a

number of partners (including Lund University - LU,

Sweden) and Working Packages (WP). It can be

considered as continuation of the DaGoB project

(Safe and Reliable Transport Chains of Dangerous

Goods in the Baltic Sea Region) (INTERREG IIIB).

The main objective of the project is to provide a risk

analysis of dangerous goods transport in the BSR.

The project is in many respects unique.

1.2 Literature review

An extensive literature review showed that a holistic

view of the maritime risks in the BSR is limited, and

they deserve a better understanding. Projects co-

financed by the EU, including the BSR INTERREG

Neighbourhood programme, have covered a wide

range of issues concerning sustainable development

in the region. Baltic Master (2005-2007) and

OILECO (Integrating ecological values in the deci-

sion making process on oil) (2005-2007) are exam-

ples of the recent EU projects. None of these pro-

jects has particularly dealt with the risks of maritime

transport of dangerous goods at a wider BSR con-

text, including the Russian part. In addition, a few

peer-reviewed papers have been confined to a lim-

ited number of risk issues, such as the m/v “Estonia”

case (Soomer et al. 2001), marine pollution in

coastal waters, oil spills detection and remote sur-

veillance (Looström 1983) and monitoring by in-

service aircraft (Von Viebahn & Gade 2000) and

satellite (Kostianoy et al. 2005).

1.3 Research questions and objective

Marine incidents data are essentially important for

the risk analysis. Time and financial resources are

A Study of Marine Incidents Databases in the

Baltic Sea Region

A. Mullai, E. Larsson & A. Norrman

Lund University, Lund, Sweden

ABSTRACT: A comprehensive risk analysis makes use of different datasets. Marine incidents data are essen-

tially important datasets. The purpose of this study is to analyse marine incident databases in the BSR (Baltic

Sea Region). The marine incident data in the region are inhabited by a wide rage of issues, such as limited da-

ta accessibility and availability, and the diversity in data quality, structure, amount, accuracy, degree of detail

and languages. Preparing for the data analysis is a very cumbersome, labour intensive, time consuming and

expensive process. Merging different datasets from different countries into a single dataset is a very difficult

process, if not impossible for a complete data merging. The paper provides experiences on how to overcome

some of these issues and proposes some suggestions for improvements in the future.

322

often limited for research projects, including the

DaGoRus project. While the signing of the

partnership agreement for the project was still pend-

ing, the relevant research questions for this particu-

lar study presented in this paper are: What is the cur-

rent state-of-the-art marine incidents data in the

region? Is it feasible to perform a comprehensive

risks analysis for the entire region? The purpose of

this study is to analyse marine incident databases in

the BSR and propose suggestions for improvements

in data accessibility, structure and quality.

1.4 Materials, methods and paper outline

After several months of communication with the re-

sponsible authorities of the BSR’s countries, the fol-

lowing marine incident databases were acquired: 1)

Danish Maritime Administration Database (DMA

DB) (1997-2006; in Danish); 2) Finish Maritime

Administration Database (FMA DB1 and DB2) – the

database contains two datasets (1990-1996 and

1997-2007; in Finnish); 3) Swedish Maritime Ad-

ministration Database (SMA DB) (1985-2007; in

Swedish); and 4) Helsinki Commission Database

(HELCOM DB) (1989-2006; in English). In section

3 of the paper, the main results and discussions in-

cluding problems encountered during data collection

are presented. The properties of the databases are

described and compared. For the purpose of bench-

marking with some of the best technology and prac-

tices in the field, in section 4, the USA’s and world’s

largest incidents databases are described. Conclu-

sions and suggestions for improvements are provid-

ed in section 5. Initially, in section 2, the concepts of

risks and risk analysis are briefly described.

2 RISKS AND RISK ANALYSIS

2.1 Maritime risks

The risk is defined as the likelihood of consequences

of undesirable events (Kaplan & Garrick 1981, HSE

1991). The terms "marine accident and incident" and

"marine casualty" denote undesirable events in con-

nection with ship operations (IMO 1996). The term

“marine incident” is used to denote undesirable ma-

rine events, i.e. marine accidents, incidents and near

missing events. The dangerous goods risks can be

defined as the likelihood of consequences of hazard-

ous release events (HSE 1991). Maritime transport

risks are statistically verifiable technological and

human activity risks. The maritime transport system

and risks consists of many elements that are classi-

fied and defined by various coding systems.

2.2 Risk analysis

Contemporary risk management recognises the fact

that the risk analysis, which is a rigorous scientific

process facilitated by standardised frameworks and

techniques, is prerequisite for the decision making

process. The main purpose of every risk study is, to

the best abilities of researchers and data and re-

sources available, to provide decision makers with

valid and reliable information and tools that would

enable them to make informed decisions. The risk

analysis varies from simple screening to major anal-

ysis that requires many years of efforts, substantial

resources and a large team of experts using various

risk analysis techniques and datasets. The main stag-

es of the risk analysis are preparations for analysis,

the analysis process and conclusions and recom-

mendations. The first stage encompasses a wide

range of activities, including identification, selec-

tion, compilation and preparation of the relevant da-

tasets. Large amounts of diverse risk-related datasets

are required, but the most important datasets are ma-

rine incident data.

3 MAIN RESULTS AND DISCUSSIONS

3.1 Limited data accessibility and availability

The Baltic Sea is an unique area in terms of sensitiv-

ity and diversity of countries surrounding the area. It

is surrounded by nine different countries with differ-

ent backgrounds, languages and practices, which

may hamper data collection and merging and per-

formance of a robust risk analysis.

Data accessibility may be an issue for the region.

Our investigation suggests that marine incidents are

recoded into databases in all BSR’s countries. The

websites of the relevant authorities in several coun-

tries of the BSR were reviewed. None of them had

marine incident databases and other risk-related data

available in electronic format for the public use. Re-

quests for data acquisition were sent to all countries

(Denmark, Estonia, Finland, Germany, Latvia, Lith-

uania, Poland and Sweden), except Russia, and

HELCOM. Signing of the partnership agreement

with Russian partners was still (2007-2008) pending.

Further, data collection for the Russian part was the

responsible of another project group. Contact infor-

mation was obtained from the SMA and other

sources. Requests were sent to the maritime admin-

istrations, coast guards, bureaus of maritime casualty

investigation and maritime safety inspectorates. The

mail delivery system confirmed that request messag-

es were successfully delivered, received and dis-

played on the recipient's computer. We were able to

receive four (see Section 1.4) marine incident data-

bases. Two databases were primarily obtained as the

result of our personal contacts with the relevant au-

323

thorities. Two other databases were obtained after

considerable communication and assistance from our

personal contacts. Some respondents did not reply or

were not interested in cooperation. Requests were

sent several times to those who did not reply. In

some countries, the authorities may be unwilling or

uncomfortable in providing data, in particular to ex-

ternal parties. Some interrelated reasons were cited

inconvenient database format, limited human and fi-

nancial resources, and data confidentiality.

Inconvenient database format for preparing and

sending data in electronic format was cited by sever-

al respondents as one of the main issues. In one

country, the incident data recorded up until Decem-

ber 2007 were available only at a relatively old

computer. According to the respondent from that

country, the data were compressed in a way that was

practically impossible converting the data into a

modern program format, including Excel format. It

was very difficult and time consuming to convert all

data manually. One respondent from another country

replied that their organisation did not work with the

Excel program as database. They were still waiting

(2008) for the EMSA (European Maritime Safety

Administration) database for the statistical analysis

of ship incidents. Another one stated that their data-

base contains personal and other information that are

not necessary for the risk analysis. Further, convert-

ing their entire dataset into a convenient data format

was time consuming and impossible task for them.

All respondents stated that preparing and sending

data in electronic format were time consuming and

labour intensive processes. Due to workload and

other inquiries and in combination with limited hu-

man and financial resources, they were unable to

provide data at all or in due time. They were too

busy to assist us as their daily work was high on

their priority list. One respondent wrote that he will

not send the entire database. But, if we needed sim-

ple extractions they would be able to assist us. In

case of a large extraction requiring special adjust-

ments, they had to charge us for that.

Data confidentiality might have been one of the

main reasons why some authorities did not reply.

One respondent stated that in his country marine in-

cident data are confidential. The data are only avail-

able for the accident investigation in his country and

in his country language. A risk analysis for the BSR

as a whole based on exhaustive data may not be pos-

sible should all the countries share a similar policy.

Two respondents made reference to annual accident

reports in pdf-format published on their organisa-

tions’ website for the public use. The review of nu-

merous accident reports showed that they were com-

prehensive and well prepared. However, a number of

issues are observed. The data are mainly analysed

and presented in form of descriptive or summary sta-

tistics, such as frequency tables and charts. Applica-

tion of advanced inference statistics and specific risk

analysis methodology were lacking. Reports are pre-

pared by or for the responsible authorities. The

knowledge comes from different corners, from prac-

titioners and scientific community alike. However,

because of systematic and rigorous processes em-

ployed, it is widely accepted that the knowledge

generated by the scientific community has a higher

degree of confidence, validity and reliability than the

other forms. The scientific literature on the maritime

risks for the region is, however, very limited. Studies

concerning dangerous goods risks are largely con-

fined to oil spills in the territorial waters of individu-

al countries or in certain areas of the BSR. Further,

integrating information from different reports is an

impossible task.

3.2 Diverse and incompatible data

In this section, the data properties are explored (see

Tables 1-6 and Figs 1-3). The HELCOM DB con-

tains marine incidents reported by the BSR’s coun-

tries. This dataset may serve, to some degree, as a

sample for studying and drawing conclusions for the

maritime risks in the entire region. However, the da-

taset is a relatively small and biased sample. The re-

view of other databases showed that incidents are se-

lectively reported to the HELCOM. Thus, during the

period 1989-2006, a total number of 906 incidents

(50 incidents per year) has been reported, of which

123 (13.6%) and 82 (9.1%) are respectively pollu-

tion incidents and incidents with no information

about pollution. These numbers are smaller than pol-

lution incidents and marine incidents recorded in the

BSR’s databases (e.g. SMA DB - 5778 incidents re-

ported during 1985-2007). During the period 2004-

2006, the Swedish Coast Guard alone has observed

on average 308 spills per year. In addition, the

HELCOM DB contains 42 variables, where 17 vari-

ables (40%) represent ship properties and conse-

quences (Table 6 and Fig. 1). The consequences are

confined to the occurrence of pollution (yes/no), the

amount and type of pollutants. The variable labels

are not properly designed and partly or completely

missing in some variables. For example, the “ship

type details” variable contains some 126 items.

The risk estimation and presentation require ex-

haustive data. The results obtained from the risk es-

timation may serve as an empirical ground for estab-

lishing risk criteria in the region. The risk criteria

may serve as benchmarking standard for measuring

and comparing the maritime risks in the individual

countries and the region as a whole. In Sweden and

other countries in the region, these criteria are lack-

ing. Further, the reliability and validity of risk esti-

mation and presentation depends very much on the

data quality, diversity and amount. Therefore, it is

324

important to perform a comprehensive risk analysis

based on all datasets available. The best alternative

is to merge all datasets into a single dataset. A pre-

condition in the data merging process is that all vari-

ables must be compatible, i.e. they have to share

similar properties including variable type, number,

label, and value and measurement level. Variables

are not organized in any particular order in the data-

bases. Based on the SMA DB, variables are labelled

and organized in main categories (Table 1). A com-

plete data merging, which is merging all databases

including all cases and variables, is not possible.

Merging parts of datasets may be possible, but con-

siderable time, resources and expertise are required.

This process includes translation, codification and

de-codification and design and re-design of varia-

bles, conversion of data from Excel to SPSS data

format, data merging and filtering. One case history

is one A4 paper text (multiplied by ca 8000 inci-

dents) and many variables are string or text format

variables written in different languages. Text format

variables contain very important information. The

databases are mainly designed based on the DAMA

coding system, which was originally agreed (1990)

by Scandinavian countries for registration and anal-

ysis of marine incidents. The FMA DB and SMA

DB share more in common than the two other data-

bases. Deviations from the code and changes are al-

so observed within the same database (e.g. FMA

DB). The labels of many variables in the FMA DB

are coded according to the DAMA coding system.

These variables must be de-codified. The databases

are specially designed databases, which may be in-

convenient for converting data into advanced statis-

tical program formats. The data were sent to LU in

an Excel format. Data analyses and result presenta-

tions with this data format are limited. The present

Excel data format of the databases is not readily

convertible to the statistical program format. The da-

ta are organized on the “case” and “variable” bases.

It is unclear whether the case histories are compiled

on “event” or “ship” bases. In many cases, in a sin-

gle incident two or more ships have been involved.

The DMA DB is confined to a very limited number

of events such as collision, grounding and fire. In

one database, many cases were deleted. The SMA

DB and FMA DB contain incidents that have oc-

curred in the respective country territorial waters for

all nationalities and ships flying the respective coun-

try flag outside the territorial waters. Therefore, var-

iables should be designed for filtering or sampling

purposes. In terms of data properties and structure,

there are considerable discrepancies among the da-

tabases (Tables 2-6 and Figs 1-3). Only a very few

variables are compatible. There are significant gaps

in the number, category, type and measurement level

of variables. The string (Str.) and nominal (Nom.)

variables dominate (51-75%) all databases. The se-

cond largest numbers of variables are scale (Sc.) and

ordinal (Ord.) variables (10-38%). More analysis

methods are applicable to scale and ordinal variables

than nominal variables. The variables for measuring

risks of maritime transport of dangerous goods are

very limited, if not lacking for certain databases.

Table 1. The main categories and examples of variables in the

SMA DB (1985-2007).

__________________________________________________

Main category: examples of variables

__________________________________________________

Time: year, month, day, day of the week, time (hours)

Location: latitude and longitude, ports of departure and arrival,

country, geographical areas, traffic area, fairway etc.

Ship: call sign, name, type, class society, nationality, built, size

(dwt, brt, length), material etc.

Ship activity: ship activity (en-route, loading/discharging), ac-

tivity onboard etc.

Exposure: crew, passengers, visitors, total numbers etc.

Cargo: description, type, amount, dangerous goods etc.

Event: type, event grading, description

Cause: categories codified, description

Other: pilot presence onboard

Environment: light, visibility, precipitation, sea, wind etc.

Consequence: human (fatality, injury, disappearance – crew,

passengers, pilots, others, total), ship (damage – description,

location, extent), environment (oil and other pollutants – type,

amount)

__________________________________________________

Table 2. The categories, types and measurement levels of vari-

ables in the DMA DB (1997-2007).

_______________________________________________

Variable

Main category Type* Measurement level**

_______________ _________________

Str. Nrc. Date Other Nom. Ord. Sc. Other

_______________________________________________

Time 4 4

Location 1 2 3

Ship 6 2 6 2

Event 2 1 1

Cause 2 2

Other 1 1

Consequences 1 1 1 1

______________________________________________

Total 13 2 4 3 15 1 2 4

______________________________________________

* Variable type: Str. (String variables whose values are not

numeric and therefore are not used in calculations), Nrc. (Nu-

meric variables). ** Variable measurement level: Nom. (Nom-

inal variables whose values represent categories with no intrin-

sic ranking), Ord. (Ordinal variables whose values represent

categories with some intrinsic ranking), Sc. (Scale variables

whose values represent ordered categories with a metric)

(SPSS 16.0 for Windows 2007)

Table 3. The main categories, types and measurement levels of

variables in the FMA DB1 (1990-1996).

_______________________________________________

Variable

Main category Type* Measurement level**

_______________ _________________

Str. Nrc. Date Other Nom. Ord. Sc. Other

_______________________________________________

Time 5 5

Location 4 2 4 2

Ship 11 2 11 2

Ship activity 2 2

Exposure 2 2

Cargo 1 1 1 1

Event 3 3

Cause 4 4

Other 1 1

325

Environment 4 1 3

Consequences 9 6 2 3 6 6 2

Response 1 1

______________________________________________

Total 40 11 5 4 36 9 11 4

______________________________________________

*, ** See the foot note in Table 2

Table 4. The main categories, types and measurement levels of

variables in the FMA DB2 (1997-2007).

_______________________________________________

Variable

Main category Type* Measurement level**

_______________ _________________

Str. Nrc. Date Other Nom. Ord. Sc. Other

_______________________________________________

Time 5 5

Location 3 2 3 2

Ship 10 3 2 12 3

Ship activity 2 2

Exposure 2 2

Cargo 1 1

Event 3 3

Cause 4 4

Other 1 1

Environment 4 1 1 3 1

Consequence 11 19 4 11 19 4

______________________________________________

Total 39 25 7 6 43 3 25 6

______________________________________________

*, ** See the foot note in Table 2

Table 5. The main categories, types and measurement levels of

variables in the SMA DB (1996-2007).

_______________________________________________

Variable

Main category Type* Measurement level**

_______________ _________________

Str. Nrc. Date Other Nom. Ord. Sc. Other

_______________________________________________

Time 3 3

Location 10 2 10 2

Ship 13 3 13 3

Ship activity 3 3

Exposure 3 3

Cargo 2 2

Event 4 3 1

Cause 1 1

Other 3 3

Environment 5 1 3 2 1

Consequence 8 26 1 5 3 26 1

______________________________________________

Total 48 33 3 4 45 6 33 4

______________________________________________

*, ** See the foot note in Table 2

Table 6. The main categories, types and measurement levels of

variables in the HELCOM DB (1989-2006).

_______________________________________________

Variable

Main category Type* Measurement level**

_______________ _________________

Str. Nrc. Date Other Nom. Ord. Sc. Other

_______________________________________________

Time 4 4

Location 1 2 1 2

Ship 10 7 10 7

Cargo 1 1

Event 2 1 2 1

Cause 4 4

Other 3 3

Consequences 6 1 5 1 1

______________________________________________

Total 27 8 4 3 30 1 8 3

______________________________________________

*, ** See the foot note in Table 2

Figure 1. The number of variables in databases.

Figure 2. Comparison among the main categories of variables

in databases.

Figure 3. Comparison among periods and numbers of marine

incidents recorded in the databases.

4 INCIDENT DATABASES IN THE USA

The study of many incident databases (see Mullai

2004) show that, in terms of the public accessibility

and amount, diversity, accuracy, quality of danger-

ous goods risk-related data, the USA is one of the

most advanced countries in the world. Many types of

data are free of charge and available for public use

in the Internet. The USA Freedom of Information

Act (1974) requires all federal and national organisa-

tions to make data available in electronic form to the

public. Hazardous Material Information System

(HMIS) and National Response Center (NRC) data-

bases are two of the USA’s and the world’s largest

100

DMA DB FMA DB1 FMA DB2 SMA DB HELCOM DB

Database

Number of variables

Time

Location

Ship

Ship activity

Exposure

Cargo

Event

Cause

Other

Environment

Consequence

Response

DMA DB FMA DB1 FMA DB2 SMA DB HELCOM DB

Number of variables

100

150

200

250

300

350

400

450

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

Number of incidents

DMA DB FMA DB1

FMA DB2

SMA DB HELCOM DB

326

databases in the field. They are available to both sci-

entific and practitioner communities. The HMIS da-

tabase (ca 200,000 case histories organised in more

than 180 variables) records all dangerous goods in-

cidents occurring in all modes of transport. The

NRC database (over a half million case histories or-

ganised in more than 230 variables) records all inci-

dents involving all types of hazmat discharges into

the environment anywhere in the USA and its terri-

tories. The data are reported by individuals and a

wide range of organisations and agencies, and cover

a wide range of systems of the USA’s chemical sup-

ply chain. In contrast to the BSR’s databases, both

the USA’s databases offer many advantages, includ-

ing massive, diverse, high quality and very well or-

ganised data, no restriction and easy data access via

the Internet, and very convenient data format. Our

experience (see Mullai & Larsson 2008) shows that

data preparation and analysis are significantly less

time consuming, resource and labour intensive than

working with the BSR’s databases. The incidents

recorded to all BSR’s databases combined are only a

small fraction of the HMIS and NRC databases.

5 CONCLUSIONS AND RECOMMENDATIONS

With reference to the research questions, merging all

databases into a single dataset and performing a de-

tailed risk analysis for the BSR may not be possible

due to issues explored in this study. However, a risk

analysis based on partly merged datasets is feasible.

Some of the issues are partly attributed to different

practices, priorities and languages. The marine envi-

ronment and safety issues are gaining more attention

in the region. In order to tackle some of the issues,

we suggest the following solutions: (i) Enhance co-

operation among maritime authorities and other par-

ties in the region. Projects like the DaGoRus project

and conferences like the TransNav 09 can contribute

to cooperation. They can serve as forums where

problems and solutions are identified and discussed,

stakeholders meet and information is disseminated.

(ii) Improve and harmonise marine incident data-

bases in the BSR. Immediate changes cannot be ex-

pected in the near future as several databases are de-

signed based on the established coding system.

Significant changes may render many years (two

decades) of data records incompatible. Therefore,

the process should be well studied and performed in

a stepwise manner. (iii) Marine incident data should

be made publicly available in electronic format via

the Internet, at least for the research purposes. The

USA experience can serve as an inspirational exam-

ple. (iv) Upgrade data compilation systems. (v) Im-

proving the HELCOM DB is a good solution, which

include reporting all marine incidents (accidental

and deliberate events) occurring in the BSR, improv-

ing the quality of variables and ensuring a higher

degree of data completeness. The maritime risks, in-

cluding risks due to the large and increasing

amounts of dangerous goods, deserve a better under-

standing and management. These can be achieved

only by united efforts.

ACKNOWLEDGEMENT

We wish to thank the DaGoRus project office for fi-

nancing this study, and Danish, Finnish, HELCOM

and Swedish authorities for their assistance in data

collection.

REFERENCES

HELCOM (Helsinki Commission) 2005. Overview of the

ships’ traffic in the Baltic Sea

HSE (Health and Safety Executive) 1991. Major Hazard As-

pects of Transport of Dangerous Substances. Report of Ad-

visory Committee on Dangerous Substances, Health and

Safety Commission, HMSO, UK

IMO (International Maritime Organisation) 1996. Reports on

Marine Casualties. Harmonized Reporting Procedures, An-

nex 3, Draft MSC/MEPC Circular, IMO FSI4/18

Kaplan S. and Garrick B.J. 1981. On the quantitative defini-

tions of risks. Journal of Risk Analysis, Vol.1, No. 1, pp.

11-27

Kostianoy A.G., Litovchenko K.Ts., Lebedev SA Stanichny

S.V., Soloviev D.M., Pichuzhkina O.E. 2005. Operational

Satellite Monitoring of Oil Spill Pollution in the Southeast-

ern Baltic Sea. Oceans - Europe, IEEE, pp. 182-183

Looström B. 1983. Swedish Remote Sensing Systems for Oil

Spill Surveillance at Sea. Oil and Petrochemical Pollution,

p. 235

Mullai A. 2004. A Risk Analysis Framework for Maritime

Transport of Packaged Dangerous Goods (PDG), in Brind-

ley, C (2004) Supply Chain Risk, Ashgate Publishing Com-

pany, UK, Chapter 9, pp130-159

Mullai A. and Larsson E. 2008. Hazardous Materials Incidents

– Some Key Results of a Risk Analysis. WMU Journal of

Maritime Affairs, pp. 65-108

Mullai A. and Paulsson U. 2002 Oil Spills in Öresund – Haz-

ardous Events, Causes and Claims. Report on the SUN-

DRISK Project, Lund University Centre for Risk Analysis

and Management (LUCRAM), Sweden, pp. 1-160

Rytkönen J., Siitonen L., Riipi T., Sassi J., and Sukselainen J.

2002. Statistical Analyses of the Baltic Maritime Traffic.

Project financed by Finnish Environment Institute and Min-

istry of Traffic and Communications.

Soomer, H. Ranta, H. Penttila, A. 2001. Identification of vic-

tims from the M/S Estonia. International Journal of Legal

Medicine, Vol. 114, Issue 4-5, pp. 259-262, Springer

TSE (Turku School of Economics, Finland) 2006. DaGoB Pro-

ject (Safe and Reliable Transport Chains of Dangerous

Goods in Baltic Sea Region): http://www.dagob.info/2006

Von Viebahn C. and Gade M. 2000 A proposal for improved

oil spill monitoring of peripheral marine regions in the Bal-

tic Sea by in-service aircraft. Geoscience and Remote Sens-

ing Symposium Proceedings. IEEE international, Vol. 1, pp

342-344