A Conceptual Model for Effective Early Warning Information

Systems (EEWIS)

Mohamed Saad Eldin

, Sherif A. Mazen

, Ehab E. Hassanen

and Hegazy Zaher

Department of Information Systems, Faculty of Computers and Information, Cairo University, Giza, Egypt

Department of Information Systems, French University, Cairo, Egypt

Keywords: Early Warning, Crises, Forecasting, Model, Indicators.

Abstract: This paper addresses the need for effective early warning information systems (EEWIS) that are capable of

predicting future crises and that can help prevent them or reduce their negative effects. The main problem

facing any EWIS is the lack of effectiveness. The most effective early warning information systems are

characterized by accuracy, flexibility and the ability to detect risks. Effective early warning information

systems can empower communities to prepare for and confront risks and disasters. An effective EWIS

should be based on a reliable and consistent model, yet the models currently available are mostly

deterministic, simplified or inconsistent in application and assumption; thus making them unreliable and

impractical. The goals of this paper are to provide guidelines for professionals involved in implementing

effective early warning information systems, and to present a novel model for EEWIS that can be adapted to

the dynamic needs of the field of crisis management and preparedness.

1 INTRODUCTION

Although early warning information systems are

used to collect and share information at a time of

crisis, such systems are not always effective. While

Assilzadeha and Mansor (2008) believe that the

development and implementation of application

software for early warning- especially for disaster

data and information management- is crucial, Glantz

(2004) claims that there is no perfect Early Warning

System (EWS), except on paper, in governmental

plans, or in a PowerPoint presentation, and that most

of the current systems are not as effective as they

should be. Along the same lines, Sanada et al.,

(2006) also agree that the information systems used

to collect and share information in the time of a

disaster are not always effective. Furthermore, Harff

(1998) argues that at present, early warnings are

rarely "early," seldom accurate, and moreover lack

the capacity to distinguish among different kinds of

crises. In this context, after analyzing the current

problems in existing early warning information

systems used in crisis or disaster preparedness, we

have found that in many cases these systems tend to

be fairly narrow in scope and do not have an

adequate or clear model for collecting, classifying,

processing and producing accurate forecasting

information. The accuracy of forecasting models is

essential for building EEWIS. The limitations of

existing early warning information systems suggest

the need for a more comprehensive conceptual

model. This is the goal we aim to achieve from this

paper. Furthermore, other motivations for this paper

are: The scarcity of specialized EWIS researches

used in crisis preparedness, most of the previous

studies did not address the key steps used in building

EEWIS and there is no agreement on the ideal

structure or functions of EEWIS.

The proposed model provides a guideline for any

organization or sector in the country that needs to

have an EEWIS for crisis preparedness. Moreover,

this model will support any information system in

producing more effective and accurate predictions of

the future. In addition, it will provide decision

makers with a reliable and manageable amount of

warning information for taking preventive actions.

2 EARLY WARNING

INFORMATION SYSTEMS

The idea of early warning emerged in the fifties of

the past century, and was used for the first time in

134

Saad Eldin M., A. Mazen S., E. Hassanen E. and Zaher H..

A Conceptual Model for Effective Early Warning Information Systems (EEWIS).

DOI: 10.5220/0004419801340142

In Proceedings of the 15th International Conference on Enterprise Information Systems (ICEIS-2013), pages 134-142

ISBN: 978-989-8565-60-0

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

military domains to predict risks and potential

attacks before they occur. Until the early eighties;

the concept of early warning had not evolved

noticeably due to a number of reasons; such as the

difficulty of creating its applications and its high

cost. However, the concept has been rediscovered

again after a series of crises and disasters had taken

place in the world and after witnessing their major

impact on lives and property (Eldin, 2011). The

expression ‘Early Warning’ is used in many fields to

mean the provision of information on an emerging

dangerous circumstance where that information can

enable action in advance to reduce the risks involved

(Basher, 2006). A universally accepted definition of

an EWS does not yet exist and most probably never

will (Sivakuma, 2009). There are many definitions

of an EWS that are used to guide the actions of

individuals, groups, and governments. The formal

UN definition is as follows: “The provision of

timely and effective information, through identifying

institutions, that allows individuals exposed to a

hazard to take action to avoid or reduce their risk

and prepare for effective response” (ISDR, 2003).

An EWS can also be defined as “a social process for

generating maximally accurate information about

possible future harm and for ensuring that this

information reaches the people threatened by this

harm, as well as others disposed to protect them

from the harm” (Glantz, 2004); (Davies and Gurr,

1998). An 'Early Warning Information System

(EWIS)' (see Figure 1.) can be understood as a set of

institutional and technical solutions designed and

implemented in a coherent way to make available, to

a wide range of users and more particularly to

decision makers, information useful to carry out

vulnerability analyses, to evaluate and manage the

risk of a hazard that can become a disaster, and to

manage disasters from prevention to recovery and

rehabilitation (Scott, 2003); (ISCRAM, 2008); (IAD,

2002). The objective of EWIS is to generate accurate

information to empower individuals and

communities threatened by hazards to act in

sufficient time and in an appropriate manner so as to

reduce the possibility of personal injury, loss of life

and damage to property or the environment. We can

use the term EWIS for any information system that

collects, shares, analyzes data, produces future

predictions about potential crises and gives

recommendations or warnings for those involved.

Early Warning Information Systems are still not

widely practiced around the world. .The applications

of EWIS are costly, limited and not widely available

especially in some international organizations. The

best known EWIS is the HEWS- Humanitarian Early

Warning System- used by the Department of

Humanitarian Affairs in the United Nations, and the

GIEWS -The Global Information and Early Warning

System- used by the Food and Agriculture

Organization of the United Nations (Verstegen,

1999).

Figure 1: EWIS architecture.

3 EFFECTIVE EARLY WARNING

INFORMATION SYSTEMS

Many developing countries, in particular the least

developed among them, have limited capacities for

effective early warning systems, and in some cases

they are virtually non-existent (Villagran 2006). The

challenges facing any early warning system are

ineffectiveness and failure to achieve its goals. Any

information system can be called effective if it

supports the organization in reaching its objectives

(Malik, 2001). Effective early warning systems not

only save lives but also help protect livelihoods and

national development gains (United Nations, 2006).

Early warning systems are widely recognized as

worthwhile and necessary investments. However in

many cases, early warning systems do not exist, are

ineffective, or break down at critical points – risking

devastation, death, and destitution (ISDR, 2003).

Two international conferences on early warning, in

1998 and 2003 produced a set of internationally

agreed upon guiding principles for effective early

warning systems. The 1998 Potsdam Conference on

Early Warning Systems and the 2003 Second

International Conference on Early Warning in

Germany addressed technical considerations,

strategic issues and institutional requirements in the

early warning field, Moreover, the conferences made

specific recommendations for strengthening early

warning systems; including increasing the ability of

these systems to be more accurate and flexible

(United Nations, 2006); (EWC, 1998); (EWC-ll,

2003); (EWC-lll, 2006). Furthermore, the core

message of the session “People-Centred Early

AConceptualModelforEffectiveEarlyWarningInformationSystems(EEWIS)

135

Warning Systems” at the World Conference on

Disaster Reduction that was held in January 2005, in

Kobe Japan, was that effective early warning

systems must be embedded in an understandable

manner and relevant to the communities which they

serve. Therefore, these systems must be developed

in such a way that ensures that they are functioning

when needed and that the warnings are timely,

comprehensible and ultimately acted upon by the

diverse array of individuals at risk in any

emergency. The Global Survey of Early Warning

Systems (2006) and the UN-ISDR/Platform for the

Promotion of Early Warning (2006) concluded that a

complete and effective early warning system model

should comprise four inter-related elements: risk

knowledge, monitoring and warning service,

dissemination and communication, and response

capability. A weakness or failure in any one part

could result in failure of the whole system (United

Nations, 2006). Along the same lines, the World

Meteorological Organization (2011) suggested that

effective early warning systems are comprised of

four operational components

1. Hazards are detected, monitored, forecasted, and

hazard warnings are developed;

2. Risks are analyzed and this information is

incorporated in warning messages;

3. Warnings are issued (by a designated

authoritative source) and disseminated in a

timely fashion to authorities and the public at

risk;

4. Community-based emergency plans are activated

in response to warnings to reduce impact on lives

and livelihoods.

Failure in one component or lack of coordination

across them leads to failure of the whole system

(WMO, 2011). Martin (2008) also points out in his

paper that the effectiveness of any EWIS largely

depends on the transformation of the event

recognition into the report of warning to the

population or people at risk (Martin et al., 2008);

(Wikipedia, 2012).

From our review of previous studies about

effective early warning information systems; we

have concluded that the effectiveness of EWIS will

depend on the following characteristics:

1. Integrated: All early warning information system

phases should be integrated into one generic

model

2. Detectability: Effective early warning

information systems should have the ability to

confirm the prediction that impacts are going to

occur.

3. Predictability: Effective early warning

information systems should be highly predictive

and capable of forecasting the crisis/hazard

before they occur.

4. Accuracy: An Effective EWIS should produce

accurate results.

5. Certainty: An effective EWIS should have a very

high level of confidence in that the predictions

and detections will be accurate and not result in

false alarms.

6. Flexible: An effective early warning

information system should be flexible and

expand its activities to include different varieties

of risks and hazards.

4 EWIS MODEL

Various writers have identified what they consider to

be the components of a successful EWIS model; for

example, a paper by Verstegen (1999) suggests that

the EWIS model should have five components:

selection of indicators; communication of warnings;

reception of warnings; early warning education;

generation and maintenance of awareness. However,

this model does not specify the methods or steps of

data collection. Moreover, it does not explain how to

measure the precursors, evaluate the event or specify

the forecasting models. Along the same lines,

Lundin (2008) suggests that an EWIS is responsible

for issuing forecasts, warnings, and responses. Yet,

the model he proposes does not clarify how the data

is collected and analyzed; explain how to prepare

future forecasts or how to select the most suitable

model for forecasting. Obviously, there is no

agreement on the ideal structure or function of an

early warning system (Shrestha, 2009). This means

that the structure and functions of EWIS may vary

from one organization to another and from one field

to another. Therefore, after reviewing most of the

previous studies about the major components of an

EEWIS model we suggest that any EEWIS (see

Figure 2.) should take into considerations four

essential sub-models. The first model (Societal

Detection of Events sub-model) includes functions

that capture and analyze the event/crisis information;

the second model (Determining Early Warning

Indicators sub-model) determines the set of

mathematical indicators that should be measured

frequently; the third model (Future Forecasting sub-

model) provides future forecasts depending on the

data calculated from the previous model, and finally

the fourth model (Issuance of Warnings sub-model)

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

136

Figure 2: EEWIS model.

is concerned with ending warnings (alerts) to users.

Each of the four models is explained below in detail.

EEWIS (Proposed System) is based on the

conceptual model which consists of set of four sub-

models, the major inputs of the proposed EEWIS

depend on the heterogeneous information which is

gathered from different sources include (News,

Statistics, Reports, Databases, Radio & T.V, Data,

etc). The major outputs from the proposed system

are set of warnings (alerts).

5 CASE STUDY

In order to demonstrate some of the key concepts

introduced in the model described above, we have

implemented a proof-of-concept prototype purely in

software. The key objectives for the proof of concept

are to demonstrate the early warning theoretical

concept, and to show how the theoretical concept

can be implemented practically against data from the

law enforcement sector. The next few pages will

explain the case study in more detail.

5.1 Societal Detection of Event

a) The event data is collected from different sources

and is regularly inputted into the EWIS.

b) The number of occurrences is calculated for each

event in a specific time frame.

c) The EWIS selected the most frequent event;

which is "widespread drug abuse among the

AConceptualModelforEffectiveEarlyWarningInformationSystems(EEWIS)

137

youth" (see Table 1.), because this event has the

highest number of occurrences. This event was

detected from multiple sources

through the

period of time from January 2006 to December

2010 (see Table 2.).

d) Data is collected from different data sources and

entered into the EWIS.

e) After analyzing the data, the system found that

this event is increasing on an annual basis (see

Table 2).

Table 1: Event occurrences (2010).

Event /Phenomena No. of Occurrences

Drug abuse (youth) 64

Spinsterhood 24

The collapse of buildings 22

Hooliganism 21

Child molestation 20

Luxury consumer 19

Unknown Parentage 18

Train accidents 18

Drug trafficking 14

Trafficking in Persons 12

Table 2: Event (drug abuse) occurrences (2006-2010).

Year 2006 2007 2008 2009 2010

No of occurrences 36 40 43 52 64

5.2 Determine the Event Indicators

a) Determining the indicators that best describe the

event: The EWIS uses from 3 (minimum) to 10

(maximum) indicators for each event/phenomena

to work efficiently, the number of indicators

varies from one event/phenomena to another.

The process of determining the indicators is

implemented by a group of experts specialized in

designing law enforcement indicators and these

indicators are:

 Indicator A: Total number of drug users in

the country.

 Indicator B: Total number of drug cases.

 Indicator C: The percentage of people

arrested in drug cases to the total population.

 Indicator D: The percentage of local people

arrested to the total of arrests.

b) Defining the indicators and determining the

variables involved in calculating them:

 Total number of addicted youth.

 Total number of drug cases.

 Percentage of youth arrested in drug cases to

the total youth population = (total number of

youth arrested/ total youth population) * 100.

 Percentage of local people arrested to the

total of arrests of other nationalities= (total

number of local youth population arrested in

drug abuse cases/ total youth arrested from all

nationalities in drug abuse cases) * 100.

Table 3: Indicators time series data (2003-2010).

Year Indicator A Indicator B Indicator C Indicator D

2003 700 650 4% 40%

2004 800 800 5% 45%

2005 850 825 5.5% 48%

2006 1050 1000 7% 49%

2007 1110 1050 7.2% 55%

2008 1206 1200 7.7% 60%

2009 1708 1400 9% 62%

2010 2000 1650 10.6% 69%

Table 4: Mathematical measures for indicator (A).

Model Fitted Equation MAPE Correlation

Linear y = 386.9 + 175.8 x 11 R

= 0.891

Quadratic

y = 783.3 – 62 x +

26.4 x

5.2 R

= 0.971

Cubic

y = 547.3 +183.5 x –

37.9 x

+ 4.8 x

3.5 R

= 0.981

5.3 Designing a Forecasting Model for

each Indicator

a) Creating a Time Series Data for Each Indicator:

Table (3) shows the time series data for all

indicators and Figure (3) shows the same table is

implemented in the EWIS.

b) Specifying the Mathematical Model: The EWIS

uses three forecasting models to be applied

(Linear, Building mathematical equations for

different forecasting models:

 Linear equation (y) = mx +b (where m and b

designate constants, m is a slope of the line)

 Quadratic equation (y) = ax

+ bx+ c (where

a, b, c are constants with a≠0 )

 Cubic equation (y) = ax

+ bx

+ cx + d

(where a, b, c, d are constants with a≠0).

 MAPE ( Mean Absolute Percentage Error) =

 

ActualValueFittedValue

ActualValue



100

(Where n = total number of actual values)

 R

: The coefficient of determination.

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

138

Table 5: Forecasting Models for Indicator (A) for the

years (2011-2014).

Year X Actual Data Linear Quadratic Cubic

2003 1 700 563 748 698

2004 2 800 739 765 801

2005 3 850 914 835 885

2006 4 1050 1090 958 979

2007 5 1110 1266 1134 1112

2008 6 1206 1442 1362 1312

2009 7 1708 1618 1644 1608

2010 8 2000 1793 1978 2029

2011 9 1969 2366 2602

2012 10 2145 2806 3357

2013 11 2321 3299 4322

2014 12 2497 3844 5526

c) Creating Forecasts for Each Indicator (Forecast

Generation):

Indicator (A), (B), (C) and (D)

Table (4) shows the correlation of variables for

each model, the EWIS will automatically choose

the cubic model because its correlation value is

the largest from among other models (=0.981),

and the value of (MAPE) is the smallest (=3.5),

so the forecasting data for the cubic model will

be more accurate. Table (5) shows the

comparison between the forecasting data for

indicator (A). The system will carry out the same

processes as indicators (B, C and D); the EWIS

will automatically choose the cubic model

because its correlation value is the largest among

other models. In addition, the value of (MAPE)

is the smallest, so the forecasting data for the

cubic model will be more accurate.

d) Charting the forecasting data for each indicator

(see Figure 3.)

e) Choosing the best model: After analyzing and

validating the data, the EWIS will choose the

best forecasting model (see Figure 4.) depending

on the following criteria:

1. Lowest MAPE.

2. Highest correlation between variables.

Figure 3: EWIS: linear forecasting trend.

Figure 4: Choosing the best model.

5.4 Sending Warning Messages

a. Calculating the Probability of a Crisis:

1. Calculating the probability of each indicator

by setting a range for each one, the probabilities

will have values between 0.1 and 1 as shown in

(Table 6.).

2. Determining the level of danger (see Table

7.) the level of danger equation will be :

[(Probability of indicator A + Probability of

indicator B + Probability of indicator C +

Probability of indicator D) / total Number of

indicators] * 100.

3. Converting the result into the color coded

scale as shown in (Table 8.).

b. Calculating the Length of Time Remaining to the

Emergence of the Crisis: The system found that

in the year 2014 the detected event will reach a

dangerous level.

c. Charting Data to Determine the Level of Danger:

the EWIS charted the data for each indicator to

determine the level of danger (there are 5 levels

of danger: dangerous, high, medium, low, and

nil), the probability of a crisis and the time frame

remaining to its emergence.

d. Matching the Level of Danger to the Color

Coded Scale: The EWIS matched the level of

danger (see Table 7. and Table 8.) to the color

coded scale. (Red, orange, yellow, green and

blue).

e. Generating Alerts: the system will automatically

generate alerts based on the forecasting data

calculated through the system. Once generated,

alerts can be distributed through different

channels to many parties.

AConceptualModelforEffectiveEarlyWarningInformationSystems(EEWIS)

139

Table 6: Probability table.

Indicator A Indicator B Indicator C Indicator D

Range Probability Range Probability Range Probability Range Probability

<500 0.1 <400 0.1 < 1 0.1 <10% 0.1

50-1000 0.2 40-800 0.2 1- 3 0.2 10-19 0.2

1001-1500 0.3 80-1200 0.3 >3-6 0.3 20-29 0.3

1501-2000 0.4 1201-1600 0.4 >6-9 0.4 30-39 0.4

2001-2500 0.5 1601-2000 0.5 >9-12 0.5 40-49 0.5

2501-3000 0.6 2001-2400 0.6 >12-15 0.6 50-59 0.6

3001-3500 0.7 2401-2800 0.7 >15-18 0.7 60-69 0.7

3501-4000 0.8 2801-3200 0.8 >18-21 0.8 70-79 0.8

4001-4500 0.9 3201-3600 0.9 >21-24 0.9 80-89 0.9

> 4500 1 > 3600 1 >24 1 90-100 1

Table 7: Final results.

Year

Indicator A

Probability

Indicator B

Probability

Indicator C

Probability

Indicator D

Probability

Level of

Danger (%)

Description

2003 700 0.2 650 0.2 4% 0.3 40% 0.5 30 Low

2004 800 0.2 800 0.2 5% 0.3 45% 0.5 30 Low

2005 850 0.2 825 0.3 5.5% 0.3 48% 0.5 32.5 Low

2006 1050 0.3 1000 0.3 7% 0.4 49% 0.5 37.5 Low

2007 1110 0.3 1050 0.3 7.2% 0.4 55% 0.6 40 Low

2008 1206 0.3 1200 0.3 7.7% 0.4 60% 0.7 42.5 Low

2009 1708 0.4 1400 0.4 9% 0.4 62% 0.7 47.5 Low

2010 2000 0.4 1650 0.5 10.6% 0.5 69% 0.7 52.5 Low

2011 2602 0.6 1999.7 0.5 12.9% 0.6 75.2% 0.8 62.5 Medium

2012 3357 0.7 2454.4 0.7 15.9% 0.7 83% 0.8 72.5 High

2013 4322 0.9 3034.6 0.8 19.8% 0.8 92.9% 1 87.5 High

2014 5526 1 3658.9 1 24.8% 1 104.4% 1 100 Dangerous

Table 8: Description of the color coded scale.

Color Coded Scale Description

Dangerous (Red Color) Threat already occurring or its eventual occurrence is “almost certain”, Probability >= 90%

High (Orange Color) The occurrence of the threat is “probable” to “highly likely", Probability >= 70% but < 90%

Medium (Yellow Color) There is “likely” chance that the threat will occur, Probability >=55% but < 70%

Low (Green Color)

Occurrence is possible but “improbable”; “little chance” to “about even” chance of occurrence,

Probability >= 20% but < 55%

Nil (Blue Color)

Probability of event occurrence is negligible or “highly unlikely”, Probability < 20%

6 TESTING THE EFFICIENCY

OF THE EWIS

According to the effectiveness criteria in section 3 of

this paper we have obtained the following results:

1. Integration: The proposed EWIS sub-models are

integrated into unified model with specific inputs

and outputs.

2. Detectability: The proposed EWIS detected

number of events occurrences as described in

table (1,2).

3. Predictability: The proposed EWIS uses three

prediction models (Linear, Quadratic and Cubic)

to forecast future data trends, in addition the

system selects the best forecasting model

according to the accuracy and correlation

between variables.

4. Accuracy: The proposed EWIS produces

accurate results (see Table 9.), accuracy can be

measured using the following equation:

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

140

Table 9: EWIS results.

Year

Indicator A

Predicted

Value

Accuracy

(%)

Indicator B

Predicted

Value

Accuracy

(%)

Indicator C

Predicted

Value

Accuracy

(%)

Indicator D

Predicted

Value

Accuracy

(%)

2003 700 698

99.7 650 656.1 99.1

4 3.9

97.4 40 40.4 99

2004 800 801

99.9 800 774.2 96.8

5 5.1

98.0 45 44.1 98

2005 850 885

95.9 825 869.5 94.6

5.5 5.9

93.2 48 47.6 99.2

2006 1050 979

93.2 1000 960.5 96.1

7 6.5

92.3 49 50.9 96.1

2007 1110 1112

99.8 1050 1065.6 98.5

7.2 7.2

100 55 54.5 99.1

2008 1206 1312

91.2 1200 1203.6 99.7

7.7 8

96.3 60 58.4 97.3

2009 1708 1608

94.2 1400 1392.9 99.5

9 9

100 62 63 98.4

2010 2000 2029

98.6 1650 1652.1 99.9

10.6 10.7

99.1 69 68.5 99.3

2011 2600 2602

99.9 1987 1999.7 99.4

11.2 12.9

86.8 73 75.2 97

2012 3350 3357

99.8 2502 2454.4 98.1

14.5 15.9

91.2 77 83 92.2

f. 2

013

4322

3034.6

19.8

97.4

92.9

2014 5526

3658.9

24.8

98.0

104.4

Average 97.2% 98.4% 95.4% 97.56%



100



PredictedValueEWISActualValue

ActualValue

X100

5. Certainty: The proposed EWIS has a very high

level of confidence in its predictions which in

average =97.14% and correlation of data is above

98%.

6. Flexible: The proposed EWIS can be used in

different organization with different events or

hazards. The system featured in the case study

used an event from the law enforcement sector,

which had the highest number of occurrences

during the period of the study. However, it can

use events from other sectors and apply the same

processes that were mentioned above.

7 CONCLUSIONS

In this paper, we have proposed a conceptual model

to build an effective EWIS. We have also presented

the detailed structure of this model and how it will

be implemented. The model was tested through a

case study to prove the concept. A list of issues

related to the EWIS was also presented. Based on

our literature review, we have concluded that

hitherto there had only been a few models

worldwide for building an efficient EWIS. Hence, an

information system that is built using our new model

will be effective and beneficial due to a number of

reasons. Firstly, it will provide information on the

past, present and future and on relevant events inside

and outside any organization. Secondly, it will be an

integrated system for gathering relevant data,

converting it to warning information and supplying

the same to concerned executives and decision

makers. Thirdly, it will reduce the time needed to

build a sophisticated EWIS from scratch. Fourthly, it

will select the best model for forecasting; which

leads up to accurate results. And finally, it will

strengthen the ability of the organization to prevent

disasters and crises before they occur.

REFERENCES

Assilzadeha H., Mansor S.B.,2008. Natural Disaster Data

and Information Management System, International

Society for Photogrammetry and Remote Sensing,

Commission VII, Working Group 5, Methods for

Change Detection and Process Modelling, 2008-2012

Basher R., 2006. Global Early Warning Systems for

Natural Hazards: Systematic and People-centred, in

The Platform for the Promotion of Early Warning

(PPEW), Bonn, Germany.

Davies J. and Gurr T., 1998. Preventive Measures

Building Risk Assessment and Crisis Early Warning

System, Rowman & Littlefield Publishers, United

States of America.

Glantz M., 2004. Early Warning Systems: Dos and

Don’ts, in The Early Warning Systems Workshop,

Shanghai-China.

Harff B., 1998. Early Warning of Humanitarian Crises:

Sequential Models and the Role of Accelerators in

AConceptualModelforEffectiveEarlyWarningInformationSystems(EEWIS)

141

Preventive Measures: Building Risk Assessment and

Crisis Early Warning Systems, John L. Davies and

Ted Robert Gurr, eds., Lanham: Rowman & Littlefield

Publishers.

Intelligent Systems for Crisis and Disaster Management –

ISCRAM, 2008, 5th International Conference on

Information Systems for Crisis Response and

Management Washington, DC, USA, May 4-7, 2008.

Intergovernmental Authority on Development, 2002.

Disaster Risk Management Program, Strengthening of

EWISs and Vulnerability Analysis, June, 2002.

International Conference on Early Warning Systems for

the Reduction of Natural Disaster (EWC'98), Potsdam,

Germany, September 1998.

ISDR (UN International Strategy for Disaster Reduction,

2003. Terminology: Basic Terms of Disaster Risk

Reduction, at http://www.unisdr.org/, accessed 15

January 2012.

Lundin H., 2008. Crisis and Conflict Prevention with an

Internet based Early Warning System, Royal Institute

of Technology (KTH), Sweden.

Malik K., 2001. Information Systems Effectiveness: An

Integrated Approach, Paper presented at the Change

Management and the New Industrial Revolution,

International Engineering Management Conference

IEMC '01,NY, USA, October 9, 2001

Martin B., Wolfgang R., Eva O., Stephan M., Et al, 2008.

Development of Suitable Information Systems for

Early Warning Systems, FZI research Center,

Germany.

Saad Eldin M., 2011. Early Warning Systems and their

Positive Effects in Supporting Capacity Development.

Dubai Police General Headquarters, Decision Making

Support Centre, Dubai.

Sanada A., Kusakabe T. And Tamura K., 2006.

Establishing a Disaster Information System to

Practically Support Disaster Countermeasures:

Conducting and Reviewing Demonstration

Experiment, Public Works Research Institute, 38th

Joint Meeting Panel on Wind and Seismic Effects.

Scott J., 2003. Input for Second International Conference

on Early Warning (EWC II) Planning, Bonn,

Germany, January 23, 2003.

Second International Conference on Early Warning

(EWC-II) - 'Integrating Natural Disaster Early

Warning into Public Policy', Bonn, Germany, October

2003.

Shrestha B. K., 2009. Planning Food Security in Nepal,

ADB-FAO workshop on Agricultural Statistics and

Agriculture Census, Bangkok , Thailand, November 2-

5, 2009

Sivakuma M., 2009. Early Warning Systems for Drought:

Past and Present, World Meteorological Organization,

Climate and Water Department, Geneva2, Switzerland

The 3rd International Conference on Early Warning

(EWC-III), Bonn, Germany. 27-29 March 2006. See

www.unisdr-earlywarning.org and www.ewc3.org

United Nations, 2006. An Assessment of Capacities, Gaps

and Opportunities Toward Building a Comprehensive

Global Early Warning System for all Natural Hazards.

Global Survey of Early Warning Systems. Bonn:

United Nations.

Verstegen S., 1999. Conflict Prognostication: Toward a

Tentative Framework for Conflict Assessment,

Clingendael Institute, Netherland.

Villagran, J. and Bogardi J., 2006. Early Warning Systems

in the Context of Disaster Risk Management, United

Nations University: Institute for Environment and

Human Security (UNU-EHS).

Wikipedia. (n.d.). Definition of Event. Retrieved

September 15, 2012, from Wikipedia the free

encyclopedia: http://en.wikipedia.org/wiki/Event.

World Meteorological Organization (WMO)., 2011.

October). Good practices for Multi-hazard Early

Warning Systems (EWS). Retrieved from MeteoWorld:

http://www.wmo.int/pages/publications/meteoworld/ar

chive/oct11/mhew_en.html

ICEIS2013-15thInternationalConferenceonEnterpriseInformationSystems

142