Data Understanding for Flash Flood Prediction in Urban Areas

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

J. Environ. Treat. Tech.

ISSN: 2309-1185

Journal web link: http://www.jett.dormaj.com

Data Understanding for Flash Flood Prediction

in Urban Areas

Nur Shuhada Abdul Malek , Syamil Zayid , Zaifulasraf Ahmad , Suraya Ya’acob , Nur

Azaliah Abu Bakar ⁵

Razak Faculty of Technology and Informatics, Universiti Teknologi Malaysia, 54100 Kuala Lumpur, Malaysia

Received: 19/11/2019

Accepted: 18/02/2020

Published: 20/05/2020

Abstract

Flash flood has become one of the major disastrous events, especially in urban areas in Malaysia. It has become more prominent

to city dwellers, causing massive loss of infrastructures, damage to people, and disruption in business and daily activities.

Population growth and rapid development of urban areas have worsened the situation even more. Since the era of Big Data, the

possibility to analyse complex data coming from heterogeneous sources, which can be used to predict flash flood, has given a

different perspective and hope for finding innovative ways to reduce the impact of flood, especially in urban areas. The purpose of

this study is to understand data needed to produce predictive visual analytics for flash flood forecasting using Cross-Industry

Standard Process for Data Mining (CRISP-DM) Methodology. Focusing on understanding the flash flood data, this paper intends

to characterize data pertaining to disaster management and identify the right data that can facilitate more accurate decision making

by stakeholders. Literature review was done to determine which data are needed in the Malaysian urban setting. The research found

the critical factors for determining flash flood occurrence in Malaysia are unique due to the tropical climate and urbanization.

Therefore, it is important to understand and characterize these factors for more effective and accurate data collection and predictive

analytics later. Based on the findings, the most significant factors identified for flash flood prediction are rainfall, urbanization, and

fluvial flood which eventually lead to blocked drainage. Details of data under these categories will be analysed as part of data

understanding of flash flood occurrence. This study intends to uncover the potential of using Predictive Visual Analytics in flood

forecasting and also to discuss how prediction can bring values to the Malaysian environment and create a sustainable ecosystem.

Keywords: Flash Flood, Disaster, Predictive Analytics, Data Understanding

flood occurrence: i) rainfall and climatic changes, ii) urban

Introduction

changes and anthropogenic activities, iii) network and

catchment factors, and iv) geomorphological features.

According to (1), the climatic changes have a major impact

on the rainfall intensity. The climate is one of the major

factors that will lead to the rainfall intensity which has the

potential to cause flood. In relation to the flash flood issue in

Malaysia, this research intends to implement the big data

analytic approach in prediction of flood using the occurrence

pattern of flash flood in Malaysia. Big data has a good

potential to play an important role in generating

knowledgeable information that ignite business and

government interest in driving towards better decision

making and mitigation plan for flood situation. This is the

case when paying attention to findings of (2) indicating that

the inefficient solid waste management is also one of the

contributors to flood situation.

Flash flood is a sudden and rapid flooding of low-lying

areas. It is one of the hazards that may cause loss of life,

injury, and destroyed or damaged assets, which could affect

the society, economy, and environment (1). Due to monsoon

season and heavy rainfall, Malaysia is exposed to the flash

flood risk every year. This condition becomes worse when

rapid urbanization takes place. As an example, Kuala

Lumpur is well known as the capital and the largest city in

Malaysia. According to Department of statistic Malaysia, the

population of Kuala Lumpur is estimated at 1.78 million as

of 2019. It is among the fastest growing metropolitan regions

in South-East Asia in terms of population, economic, and

social development. Nevertheless, in terms of climate, Kuala

Lumpur is one of the equatorial regions often beset by

prolonged rainfall and storms that eventually lead to flood.

Several factors cause flash flood situation. Specifically,

in case of Kuala Lumpur that is a big city located in a tropical

country, the research found four factors contributing to flash

Big Data usually known as 5Vs which are volume,

variety, velocity, veracity, and value. As for big data

Corresponding author: Suraya Ya’acob, Razak Faculty of Technology and Informatics, Universiti Teknologi Malaysia, 54100

Kuala Lumpur, Malaysia. E-mail: suraya.yaacob@utm.my.

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

definition, these 5Vs can be translated as a big volume of

data collection that has velocity of complexities from variety

instruments and sources which has the potential veracity and

value to be used for data-driven decision makers in the

organizations. To discover the required data and to get

information, data analysis approaches such as descriptive,

predictive, and prescriptive are used. Therefore, with the

result obtained from the analysis, data are used to generate

useful information supporting better decision making. As for

this research, the implementation of Big Data Analytics is to

predict the flash flood occurrence supports sustainability of

cities and community in Malaysia. Therefore, the present

study intends to understand and characterize data pertaining

to flood situations and identify the right data that can

facilitate more accurate decision making regarding to flash

flood situation. At the end, it can help to reduce flash flood

impact on society and economy in urban areas such as Kuala

Lumpur.

unseen data or information.

1.2 Impact of Flash Flood Prediction

The application of big data to the prediction of flood can

reduce many hazards to the environment, society, and

economy. An early identification of hazards will help to

manage residual risks. Authorities can formulate more

comprehensive mitigation plans by taking into consideration

three different entities: i) citizens affected by the flash flood,

ii) the agencies that are expected to manage the flash flood,

and iii) the risk reduction experts. At the end, the initiative

practically taken can reduce risk of flash flood to the

economy, people, process, technology, society, ecology, and

environment in Malaysia.

In brief, the research objective is to reduce flash flood

risk in Kuala Lumpur. In order to do that, the researchers

intend to develop predictive visual analytics based on the

environment challenges as shown in Table 1.

.1 More demand for Flash Flood Prediction.

Table 1: Research Summary

In (3), it was found that, in recent years, more researchers

Elements Description

have started to explore and integrate the predictive analytics

techniques of data exploration into visual analytics systems.

The capabilities of predictive analytics in reducing risk,

making more intelligent decisions, and generating different

customer experiences have attracted a lot of industrial

players implement it in their business (4). Another study (5)

investigated how the visual analytics community has

recently focused on building interactive visualizations and

associating them with predictive analytics methods. It was

achieved through translating real data into the knowledge for

their business decisions. The descriptive and predictive

analytics are among four categories from business analytics.

They differ in the way the use data: the descriptive analytics

summarizes what has happened by collecting relevant data,

storing them, and presenting the information to find the

trends (4); on the other hand, the predictive analytics goes

beyond that; it purposely provides insight into what and why

will happen in the future by analysing the current and

historical data (4).

Predictive analytics uses machine learning algorithms

and statistical analysis techniques to predict flash flood

future trends (36). In more details, the authors in (6) found

that predictive analytics can provide the organization or

business with a better understanding of what people need

and, at the same time, helps to identify potential mitigation

plans.

In (37), an interactive visual analytics application was

proposed to be combined with automatic predictive visual

analytics supported by domain experts to investigate the

environmental conditions. The automatic predictive analysis

has achieved results of a high level of accuracy (5). A

predictive model can forecast the buying patterns, potential

risks, and possible prospects through providing a deep

understanding of the customers (38), and according to (39),

it is capable of foreseeing possible future stories from the

past similar cases via a predictive analysis system. The

researchers in (40) mentioned that, in the current era of big

data, machine learning, and Artificial Intelligence (AI),

many visual analytics systems have been recently used to

produce different predictive models applicable to defining

What

Why

Developing predictive visual analytics to reduce

flash flood risk in Kuala Lumpur, Malaysia.

Flash flood has risk of loss of life, injury, and

destroyed or damaged assets, which could occur

to society, economy, and environment.

When

Rainy season throughout the year.

Where

Malaysia and, more specifically, the urban areas.

In this case the research focuses on Kuala

Lumpur as a capital city and Selangor as most

developed state in Malaysia.

Who

How

Affected environment and economy, urban

regions, people’s health, and business and social

activities.

To identify and characterize factors that

contribute to the flash flood in the Malaysian

environment.

Due to the Cross-Industry Standard Process for Data

Mining (CRISP) methodology, there are six phases to

implement predictive analytics. Due to environment-related

content, this paper will focus more on phase 2, i.e., data

understanding. The research found the importance of data

understanding as to get the knowledge about the data, the

needs that the data will satisfy, the availability, the

requirements and the sources of the data. Furthermore, data

understanding is important to connect between the flash-

flood environment context and analytical preparation need to

be done later. In data understanding phase, this study will

identify potential of environmental data to be used; then, it

characterizes and describes each data for predictive

modelling in the next phase. The rest of the present paper is

organized as follow. Section 1 introduces the research as a

whole. Section 2 provide a background of the study

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

explaining the flash flood environment and discussing the

predictive analytics. Section 3 explains the methodology

used for flash flood data understanding. Section 4 discusses

the factors identified as the gist of this paper. Finally, section

root cause analysis of flood situation in Kuala Lumpur can

be divided into a few perspectives; Process, Technology,

Economy, People and Social. Wet and humid climate as well

as geographical characteristics of its location, Kuala Lumpur

has a high potential to be exposed to flooding. Despite the

existence of draining or water tunnels system in the city,

several reservoirs and rivers have been found near the city,

which will affect the water flow to the city.

provides the conclusion and the summary of future work.

Working Background

.1 Flash Flood Situation in Malaysia

Flash flood is a global issue faced by major cities around

the world. It occurs because of a complex combination of

meteorological and hydrological extremes such as extreme

precipitation and flows. Flash flood can also be the result of

human activities, including unplanned growth and

development in floodplains, or it can take place due to the

breach of a dam or an embankment that has failed to protect

planned developments (7). In Malaysia, flash flood is not a

strange phenomenon anymore. As a tropical climate,

Malaysia has high humidity and rainfall throughout the year.

Heavier rains usually fall during November to February,

especially in the East Malaysia. Furthermore, the rapid

urbanization has led to increase loss of open space and

forested land, reducing flood plain and gradual canalization

of the rivers. These factors have led to runoff and peak flow

resulting in the increase of flood occurrences, especially in

the city areas (8). In Malaysia, Department of Irrigation and

Drainage (DID) is responsible for handling, managing, and

reducing the flash flood. DID has identified and listed flash

flood locations in each state by the order of severity. The

severity is based on the flash flood frequency. Level 1 refers

to cases when the frequency exceeds 5 times a year, level 2

refers to cases when the frequency is between 2-4 times a

year, and finally, level 3 refers to cases when the flood

happens less than 1 time during a year. Furthermore, the

authors in (9) created the plotted map based on the

estimation on road map published by the Malaysian Public

Work Department (PWD) (see Figure 1). Three colours were

used to differentiate flash flood duration.

As shown in Figure 1, Kuala Lumpur is exposed to most

flash flood hazard in Malaysia. Kuala Lumpur is the capital

city of Malaysia where several flash flood occurrences have

been recorded over the past years. Between 1965 and 2016,

Figure 1: Flash Flood in Malaysia (source: (9))

6 cases of natural disasters have been recorded in Malaysia,

including landslide, epidemic, tsunami, mudflows, storm,

and wildfire; half of the cases is related to flood. The impacts

of the recorded flood include damage to properties,

disruption in social activities and transportation systems, and

losses in businesses and household assets. Concerned by the

above-mentioned impacts, in 1971, Federal Government of

Malaysia established a permanent commission for flood

control to cope with the issue. The Head of the Commission

is Minister of Agriculture, and DID has been given a

mandate to take the flood mitigation initiatives (10). To

identify the root causes of the flash flood, an Ishikawa

diagram was constructed. In addition, 5W and 1H (5W -

What, When, Where, Who and Why, 1H – How) approach

has also been used to have a clear picture on how to tackle

this issue as shown in Figure 2. Using the approach, related

.2 Predictive Analytics for Early Flood Detection

Flood is a disaster Kuala Lumpur encounters almost

every year. It often causes damage to belongings for business

and people. In certain cases, there is a need to evacuate

buildings in a certain area. At the end, the people request a

vast amount of money for repairing and replacing

necessities. Flash flood prediction is an effective solution

that can reduce the impact of disaster. Predicting flash flood

is quite challenging since the factors of prediction are not

only depending on meteorological phenomena and heavy

rain. Therefore, at the same time, it is very critical to apply

flood forecasting as real-time indication of flow rates and

water levels ahead of time.

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

Figure 2: Ishikawa Diagram identifying the issues and factors of flash flood in Kuala Lumpur

Predictive analytics is one of the element of big data

analytics. It is one of the technology categories defined in

the elements of the fourth industrial revolution (4IR) and

refers to a large collection of complex data generated by

various instruments and sources. Big data analytics has the

potential to be implemented by data-driven decision makers

in organizations. To discover required data and to get

information, data analysis approaches such as descriptive,

predictive, and prescriptive are used. Therefore, with the

result obtained from the analysis, data are transformed into

predicted information to support better decision making.

Predictive analytics will extract information from data and

using it to predict trends and behaviour patterns. In relation

to the issue of flash flood in city areas, this study will identify

pattern within data to predict future flash floods and trends.

Commonly, statistical inferences, machine learning,

artificial intelligence and data mining approaches are applied

to predictive analytics. This approach has been found

effective, especially when dealing with large datasets (11).

Hence, it is essential to develop better understanding about

what data can be used for flash flood prediction. By

characterizing and knowing the data types, then the research

will be able to identify which techniques and methods can be

used for prediction. According to (11), there are three

predictive analytics techniques known as regression,

classification and clustering. Regression is a statistical

technique to measure the relationship between variables.

Linear and Non-Linear are two types of regression

commonly uses for prediction. Meanwhile classification

covers the process of categorizing a new instance based on

information from a training dataset. By using the data

attributes features from the training data set, the classifier

can predict categories from unknown instances. Well-known

classification methods include Bayesian, decision trees,

Artificial Neural Networks and Support Vector Machine

(SVM). Classification is similar to classification which also

attempts to categorize a new observation into a class

membership, only that this technique is an unsupervised

method that discovers the natural groupings of a data set with

unknown class labels.

It is essential to understand the flash flood data based on

the prediction of the rainfall model. The research found the

major factors of Kuala Lumpur’s flash flood as follow: i)

blockage of drainage channels, ii) increase of surface runoff,

iii) uncontrolled urbanization, iv) poor drainage

maintenance, and v) inadequate drainage system. Every

flood event is followed by several physical and mental

effects. Major physical effects of flash flood in Kuala

Lumpur are damages to properties, business premises, and

roads. As for the services, flash flood interrupts

transportation services. According to [12], currently DID has

come out with a list of mitigation actions such as: i) more

frequent maintenance of drainage system, ii) construction of

flood diversion, iii) control of land use activities, iii)

installation of flood bypass, iv) limitation of deforestation,

and v) enhanced and stringent acts on flash floods.

Based on literature review, flat clustering was used to

categorize data and provide flood predictions. Flat clustering

is generally used to limit classification of data. The analysed

data is later transformed into dashboard, where the analysed

data is presented in a visual form. These data give better

insight and data understanding for the decision makers to

predict the flood situation in an early time. As proposed by

(12), inundation maps, which can be prepared in GIS, are

useful tools to determine the areas vulnerable to flood

events. The usage of geospatial data can help to construct

Flood Hazard Map, Flood Risk Map, and Flood Evacuation

Map in Malaysia (13). These maps are important to identify

the risks involved and facilitate proper land use planning.

Furthermore, the authors in (14) indicated that the usage of

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

satellite image data and geospatial data in flood forecasting

processes leads to long-term risk reduction and effective

post-disaster responses.

importantly, prediction is the catalyst for flood mitigation

plan. The presentation of the flood risk prediction will be

simplified using data visualization in order to improve the

users’ understanding about the data and make use of the

information.

Proper planning and mitigation at the identified area can

help to reduce the local economy losses, thus, economy in

the high-risk flood area will be more sustainable. Land and

property value can be appreciated when the flood can be

predicted and the risk of land disruption, landslide, and

cracks can be reduced. As such, the town planning can avoid

massive development in the high-risk flood area. This action

can help to preserve the land and greens in the city that lead

to more secure sustainable environment.

The awareness and alertness about the predictive flash

flood will help the people to make a proper planed about the

activities during the rainy season. The social and business

activities can operate as usual without worrying the risk

caused by the flood. Furthermore, the people can customize

their activities according to the season as predicted by the

result. Daily activities and events can be held with ease

without any disruption and sudden cancellation. In the other

hand, people are more alert and concern about their safety

surroundings, thus to be more responsible to prevent the any

risk and make a better living.

Advancement in tools employed to analyse complex data

such as hydro climate data can accelerate the process of data

computation (15). As such, the method of Nonlinear

Autoregressive network with Exogenous input (NNARX)

uses numerical method to examine simulated and actual data.

Through examinations in real situations, NNARX model

was found capable of predicting flood water level ahead of

time (16). Meanwhile, another study took into consideration

the non-linear least square regression as a method applicable

to the estimation of the rain intensity by measuring the size

of rain drops (17). The study on the correlation of flood

pattern (water level) and the prediction (rainfall) on its

recursive occurrence is of a high significance (18).

The rapid advancement of technology of Big Data

Analytics (BDA) can be used to help the early prediction of

flood. Study from (19, 20) indicated that, water level

prediction has improved from 5 hours of early flood

prediction to 3 hours using the same algorithm Nonlinear

Autoregressive network with Exogenous input (NNARX).

Furthermore, by using advance data analysis from collection

of live heterogeneous data sources like sensors and satellite,

process of prediction can be done in timely manner, an

improvement from what is lacking in the previous generation

of data analysis environment. Therefore, providing timely

information helps the related parties to take further actions

and implement prepared strategic plans.

Methodology

This research is carried out using the Cross-Industry

Standard Process for Data Mining known as CRISP-DM.

CRISP-DM is a widely-used methodology and its use in

projects helps researchers to be focused on delivering real

business value. CRISP-DM breaks the process of data

mining into six major phases as shown in Figure 2. The

sequence of the phases is not strict and moving back and

forth between different phases is allowed, as it is always

required. The arrows in the process diagram indicate the

most important and frequent dependencies among phases.

The outer circle in the diagram symbolizes the cyclic nature

of data mining itself.

The early flood detection and its visualization need to be

improved in future studies carried out in this field. The gaps

are identified as follow:

)

The data visualization dashboard can be

understood by the decision makers and operators

people/society); therefore, data presentation should be

(

clearly expressed and it should satisfy the requirements of

decision makers and operators. The artificial intelligence

applied to the system can be useful in improving the model.

Thus, this will increase the participation of both government

and society in taking proactive action regarding early

detection of flood preventing its damages.

)

The early flood detection model based on real-time

data is suggested, whereby it can provide the government

and society with more accurate information, improving

decision making and execution plans. Thus, the flood-

induced damages can be reduced and economy will be

secured.

)

The data analysis of early flood detection model

should include other variables rather than focusing only on

rainfall data and water level. With the other variables, it can

expand the analysis widely with more data pattern.

.3. The benefit of flash flood prediction

Literature comprises multiple variables, including

demographic, hydrological, and ecological variables, that

have the potential to enhance the decision making

capabilities in reducing water-based disaster risk (2, 21).

Early detection result of analytic data will help the users or

stakeholders to make better flood-related decisions such as

town planning, flood prevention program, identifying

strategies and actions in order to reduce damages. Most

Figure 3: CRISP-DM Cycle

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

This paper will describe only the data understanding

phase since it is our main focus in this study. The data

understanding phase involves four steps: i) Initial Data

Collection, ii) Describing data, iii) Exploring data, and iv)

Verifying data quality. Furthermore, this paper will go

deeper only regarding the initial data collection step. Initial

data collection is essential since it is the main step towards

getting the knowledge about the data, the needs that the data

will satisfy, the availability, requirements, and the sources.

This step is pre-requisite before real data collection because

real world data is often inadequate, unreliable, and/or

missing in certain behaviours or trends and is likely to

contain many errors. Therefore, the characterization of the

data in the initial step is capable to collect data from different

sources and then understand, re-organize, re-format, and

integrate the collected data (5) before the real data collection

takes place. It requires an extensive analysis to deal with data

scalability based on the selection of data requirement.

Moreover, data characterization also concerns more about

the business process and the flow of data within it. There are

lots of data and knowledge that can be extracted from

business processes and from relationship among the data.

Finally, proven and credible initial data understanding is

significant to determine the quality and trust of the analysis

results (22).

applications and establishing the information on alternate

routes (29). Thus, with the analysis of the information about

rainfall and its regime, the results can help the operators

know the current and historical rainfall water information

promptly and accurately and also help policy makers design

the right flood rescue programs, hence minimizing casualties

and property losses (27).

Thus, rainfall data can be used for prediction purposes.

The researchers believe that the fluvial flood data that has

been collected during 10 years (from 2000 to 2010) by DID

can bring insight into predicting the flash flood in city areas.

However, the research also found that rainfall data alone are

insufficient to determine flash flood. By analysing trend

using time series and historical data, the correlation result

showed a weak relationship between rainfall amount and

number of flood incidents. This study found that flood

incidents do not happen just because of rainfall; there are

also other contributing factors such as tidal rivers, climate

change, temperature, altitude, drainage and terrain changes

(23). Due to the focus on flash flood, the factors such as tidal

rivers, temperature and climate change will not be covered

in this study. Thus, the research will further investigate the

factors of altitude, drainage and terrain changes in the next

paragraph.

Table 2: Maintenance and Number of Reservoirs in Malaysia

Flash Flood Initial Data Understanding

Since flash flood takes place due to multiple factors, it is

Area

State

DID PBT Others

Total

(ha)

difficult to understand data involved and choose the right

parameters unique to the impacted area. For example, coastal

areas and vegetation areas versus a place that is nearby a

river with rapid development may have different factors to

be considered. Thus, any single model cannot offer all

factors from all aspects when looking into the best way of

understanding the data for later analysis.

Perlis

138

185

278

188

169

240

288

96.55

233.81

38.00

Kedah

Penang

Perak

214.40

Selangor

171

397 1,492.92

In Malaysia, there is two types of flood: monsoon flood

and flash flood (or urban flood). In literature, several

contributing factors for flood such as tidal rivers, climate

change, temperature, altitude, drainage, and terrain changes

have been discussed (23). Since this paper is focused on

controlling the flash flood in Malaysia, only related notable

key variables of this type of flood are discussed in the

following. Principally, flash flood occurs due to four major

factors: rainfall, blocked drainage, urbanization, and fluvial

flooding (24). Each factor will be described in the following

subsections.

Kuala

Lumpur

305.03

5.76

Putrajaya

Negeri

Sembilan

111

211

438.38

Melaka

Johor

302

130

136

350

152.12

253.95

221.64

Pahang

Terengga

412.39

.1 Rainfall

Kelantan

Sarawak

Sabah

172

186

14.63

54.17

70.43

1.68

A study into the historical data of rainfall indicates that

extreme rainfall is associated with flash floods or water river

floods. It is clear that the key factor in river flash severity

and its speed is characterised by intensity of the rainfall;

thus, it will cause risk to people’s life (25). The authors in

Labuan

,48

(

26, 27) used rainfall data, which were associated with the

hydrological data, to resolve flood disaster using an early

warning system. Another study (28) showed that

Total

200

531 2,211 4,005.85

Source: DID (Department of Irrigation & Drainage)

PBT (Pihak Berkuasa Tempatan means Local Authorities in

Malaysia)

significant relation exists between rainfall and the disaster

mapping in metropolitan and surroundings area. For

example, in Philippines, a predictive modelling technique,

i.e., the clustering technique, is used for flood projection and

warnings by the use of rainfall information through web

.2 Blocked Drainage

Drainage is the natural or artificial removal of surface

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

and sub-surface water from an area. Blocked drainage is

often caused by debris and litter, particularly around the

cities. According to The Star (30), silt traps built is very

poorly maintained; as a result, sediment and debris end up

clogging the drains, leading to flash flood after a downpour.

For years, the media has been highlighting the poor

maintenance of the city’s drains that have caused reduced

flow capacity and, thereby causing flood in the area. Data

related to drainage are collected by KL Municipal Council

or DBKL. DBKL has started to collect flash flood-related

data only since 2010.

Figure 4: Warning Level Indicator

In Selangor alone, there are about 66 water level

telemetry stations, 95 rainfall intensity telemetry stations,

103 siren telemetry stations, 30 IP cameras telemetry

stations, and 25 flood gauge telemetry stations, as part of

flood early warning and forecasting system installed by DID.

These data and information will then be fed into a portal for

monitoring (37).

.3 Urbanization

Urbanization is the outflow process of population

moving to a larger city in order to satisfy their needs for

education, cultural development and professional realization

(31). In (32), a very close relationship was found between

flooding and rapid development and urbanization. The more

populous the area, the higher the chances of flash flood to

occur (33). The Geographical Information System (GIS)-

based analysis indicates that there is a strong correlation

between the land use changes and the flood hazards (34). If

development processes are not done in a sustainable

approach, it would result in higher likelihood for flood

occurrences. The Integrated Flood Forecasting and River

Monitoring System (IFFRM) has been developed by DID in

order to forecast flood by monitoring the river within Klang

Valley.

The research found strong correlation between flash

flood and urbanization because of the development of urban

area changes the infrastructure and its surrounding. The

changes in land use associated with urbanization affect

flooding in many ways. Land use and massive human

activities influence and modify how rainfall are stored on

and run off the land surface into rivers. In cities areas which

covered more by roads and buildings, the land surfaces have

less capacity to store the rainfall. With less storage capacity

for water in urban area, urban streams rise more quickly

during heavy rain and have higher density of sediments.

Indirectly, this impacted the drainage system which

eventually leads debris and sediments to the river channel.

.4 Fluvial Flooding

Areas prone to flood is another contributing factor of

flood occurrence, especially areas situated at the lower

height, adjacent to a river called floodplain(35). Kuala

Lumpur has its own fair share of flood history. Being

situated at the confluence of Kelang and Gombak rivers has

added another factor of flash flood occurrence by having

riverine or fluvial flood in the area. DID has installed

telemetry stations to monitor river water level at the rivers in

Malaysia as part of the Integrated River Basin Management

Figure 5: Flood Gauge

Conclusion

The flood situation in the city of Kuala Lumpur

contributes major impact to the society, economy, and

environment. Climate change is one of the main concerns

that contribute to flooding. The implementation of big data

analytics approaches such as predictive and descriptive

analyses will enable users to identify the correlation between

rainfall and water level and data patterns, which result in

early prediction for flood. If prediction and preventive

actions are done properly, the damage to people and

environment can be significantly reduced.

(IRBM) Plan and Flood Mitigation Plan to reduce flood risk

of the area (36). Apart from that, telemetry stations are also

used to monitor rainfall intensity daily. To complete the

monitoring component, IP cameras were also installed to

have a better picture of the condition.

Siren will be used once the water level reaches the

warning and danger level. The siren is only to warn nearby

residents and is stationed at certain places with high track

records of flooding.

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

Figure 6: Infobanjir Portal showing Water Level & Rainfall Data

Flash flood pattern prediction is still in the early stages of

development. The evaluation of interdependent variables as

contributing factors is crucial in determining the correlations

among different variables. In Malaysia, Big Data Analytics

has been started since 2014 mandate. From that time, data

collections conducted by DID on river basins in regard to

water level have been a big help given to researchers to make

required analyses on time series and historical data collected.

Collaboration on data sharing like this has helped

researchers a lot in conducting prediction tasks using

machine learning algorithms. This indicates that

collaboration between different parties such as government

agencies, researchers, and private sectors can accelerate the

formulation of predictive models. In future, prediction

models can be completed and then can be used as a

breakthrough in predicting flash flood patterns in Kuala

Lumpur.

Ethical issue

Authors are aware of, and comply with, best practice in

publication ethics specifically with regard to authorship

(avoidance of guest authorship), dual submission,

manipulation of figures, competing interests and compliance

with policies on research ethics. Authors adhere to

publication requirements that submitted work is original and

has not been published elsewhere in any language.

Competing interests

The authors declare that there is no conflict of interest

that would prejudice the impartiality of this scientific work.

Authors’ contribution

All authors of this study have a complete contribution for

data collection, data analyses and manuscript writing

The method that integrate structural and non-structural

measures can manage the issues in a more realistic and

holistic way. Therefore, integration of various parties can

construct a better framework to overcome flash flood in

Malaysia and provide sustainable development environment

for the surrounding area.

Four flash flood incidence factors identified in the

previous section are significant clues for real data collection.

The characterization of the flash flood factors will act as the

guidelines to identify the potential data collection that can be

used to predict flash flood in Kuala Lumpur, Malaysia.

Furthermore, this research will collect the identified factors

from the agencies, describe and explore the data, and finally

it will verify again the data quality.

References

Service NsNW. Flash Flooding Definition 2019 [Available

from: https://www.weather.gov/phi/FlashFloodingDefinition.

Zambrano L, Pacheco-Munoz R, Fernandez T. Influence of

solid waste and topography on urban floods: The case of

Mexico City. Ambio. 2018;47(7):771-80.

3. Malik A, Maciejewski R, Towers S, McCullough S, Ebert DS.

Proactive Spatiotemporal Resource Allocation and Predictive

Visual Analytics for Community Policing and Law

Enforcement. IEEE Transactions on Visualization and

Computer Graphics. 2014;20(12):1863-72.

Attaran M, Attaran S. Opportunities and Challenges of

Implementing Predictive Analytics for Competitive Advantage.

International Journal of Business Intelligence Research.

018;9:1-26.

Junhua LU WC, Yuxin MA,Junming KE,Zongzhuang LI,Fan

ZHANG,Ross MACIEJEWSKI. Recent progress and trends in

predictive visual analytics. Front Comput Sci. 2017;11(2):192-

Acknowledgement

The authors gratefully acknowledge Ministry of

Education (MOE) and Universiti Teknologi Malaysia

07.

6. Lebied M. Top 11 Business Intelligence and Analytics Trends

for 2017 2016 [Available from:

http://www.datapine.com/blog/business-intelligence-trends-

017/.

(UTM) for the support in conducting this study. This work is

conducted at Razak Faculty of Technology and Informatics

and funded by UTM (RUG: QK130000.2656.17J23).

Journal of Environmental Treatment Techniques

2020, Volume 8, Issue 2, Pages: 770-778

Jha AK, Bloch R, Lamond J. Cities and Flooding A Guide to

Integrated Urban Flood Risk Management for 21st Century.

Science and Information Application Technology; 2009 4-5

July 2009.

012.

28. Althuwaynee O, Pradhan B, Ahmad N. Estimation of rainfall

threshold and its use in landslide hazard mapping of Kuala

Lumpur metropolitan and surrounding areas. Landslides.

2015;Vol 12(No 5):pg.861-75.

29. Panganiban EB, Cruz JCD, editors. Rain water level

information with flood warning system using flat clustering

predictive technique. TENCON 2017 - 2017 IEEE Region 10

Conference; 2017 5-8 Nov. 2017.

Seang SH. A Case Study of Mitigating Flood in City Center of

Kuala Lumpur. 2009.

Nizam MS, Ghani ANA, Md Ghazaly Z. The characteristics of

road inundation during flooding events in peninsular Malaysia.

International Journal of GEOMATE. 2019;16:129-33.

0. Loi HK. Flood Mitigation and Flood Risk Management in

Malaysia. 1996.

1. Assunção MD, Calheiros RN, Bianchi S, Netto MAS, Buyya R.

Big Data computing and clouds: Trends and future directions.

Journal of Parallel and Distributed Computing. 2015;79-80:3-

30. Lam Thye L. Improve city’s drainage2016. Available from:

https://www.thestar.com.my/opinion/letters/2016/05/16/impro

ve-citys-drainage/.

31. Yuliya A.Efremova, Olga N.Goryacheva , F.Kurbanova R.

Image of a City as a Factor of Strategic Development of a

Territory. Journal of Environmental Treatment Techniques.

2019;7(Special Issue on Environment, Management and

Economy):925-9.

32. Siti Fadzilatulhusni Mohd Sani, Rindam M. Analisis taburan

hujan dan impaknya kepada sumber air di Pulau Pinang. 2011.

33. Bhuiyan Tariqur R. Facts and Trends of Urban Exposure to

Flash Flood: A Case of Kuala Lumpur City. In: Reza

Mohammad Imam H, editor. Improving Flood Management,

Prediction and Monitoring. Community, Environment and

Disaster Risk Management. 20: Emerald Publishing Limited;

2018. p. 79-90.

2. Gogoase DEN, Armaş I, Ionescu CS. Inundation Maps for

Extreme Flood Events at the Mouth of the Danube River.

International Journal of Geosciences. 2011;02(01):68-74.

3. Zakaria SF, Zin RM, Mohamad I, Balubaid S, Mydin SH, Mdr

EMR. The development of flood map in Malaysia. 2017.

4. Yu M, Yang C, Li Y. Big Data in Natural Disaster Management:

A Review. Geosciences. 2018;8(5).

5. Abdullah MF, Ibrahim M, Zulkifli H. Big Data Analytics

Framework for Natural Disaster Management in Malaysia.

Proceedings of the 2nd International Conference on Internet of

Things, Big Data and Security2017. p. 406-11.

6. Rohaimi NA, Ruslan FA, Adnan R, editors. 3 Hours ahead of

time flood water level prediction using NNARX structure: Case

study pahang. 2016 7th IEEE Control and System Graduate

Research Colloquium (ICSGRC); 2016 8-8 Aug. 2016.

34. Chang H, Franczyk J, Kim C. What is responsible for increasing

flood risks? The case of Gangwon Province, Korea. Natural

Hazards. 2009;48(3):339-54.

7. Reba MNM, Roslan N, Syafiuddin A, Hashim M, editors.

Evaluation of empirical radar rainfall model during the massive

flood in Malaysia. 2016 IEEE International Geoscience and

Remote Sensing Symposium (IGARSS); 2016 10-15 July 2016.

8. Yusoff A, Din N, Yussof S, Khan S. Big data analytics for Flood

Information Management in Kelantan, Malaysia2015. 311-6 p.

9. Ruslan FA, Samad AM, Adnan R. 3 Hours Flood Water Level

Prediction Using NNARX Structure: Case Study Kuala Lumpur

35. Mohammad Ali Nezammahalleh, Mojtaba Yamani,

Abolghassem Goorabi, Mehran Maghsoudi, Mohamadkhan S.

Evaluation of a GIS-based floodplain height difference model

for flood inundation mapping, case study Rudbar, Iran. Journal

of Environmental Treatment Techniques. 2017;5(3):100-6.

36. DID DoIaD. Towards Realising Integrated River Basin

Management In Malaysia 2017 [Available from:

https://www.water.gov.my/index.php/pages/view/708.

37. Selangor JN. Pengurusan Sumber Air dan Hidrologi JPS Negeri

015.

0. Adnan R, Samad AM, Zain ZM, Ruslan FA. 5 hours flood

prediction modeling using improved NNARX structure: case

study Kuala Lumpur. 2014 IEEE 4th International Conference

on System Engineering and Technology (ICSET)2014. p. 1-5.

1. Youssef AM, Sefry SA, Pradhan B, Alfadail EA. Analysis on

causes of flash flood in Jeddah city (Kingdom of Saudi Arabia)

of 2009 and 2011 using multi-sensor remote sensing data and

GIS. Geomatics, Natural Hazards and Risk. 2015;7(3):1018-42.

2. Sacha D, Stoffel A, Stoffel F, Kwon BC, Ellis G, Keim D.

Knowledge Generation Model for Visual Analytics.

Visualization and Computer Graphics, IEEE Transactions on.

http://water.selangor.gov.my/index.php/ms/2-

uncategorised/84-pengurusan-sumber-air-dan-hidrologi-jps-

negeri-selangor.

Deshpande, P. (2018). Predictive and prescriptive

analytics in big data Era. Advances in Intelligent Systems and

Computing, 810, 123–132.

Seebacher, D., Miller, M., Polk, T., Fuchs, J., & Keim,

D. (2019). Visual Analytics of Volunteered Geographic

Information: Detection and Investigation of Urban Heat Islands.

IEEE Computer Graphics and Applications.

36.

37.

014;20:1604-13.

38.

39.

40.

Mukherjee, S. (2019). Predictive Analytics and

Predictive Modeling in Healthcare. SSRN Electronic Journal,

(June).

Yeon, H., Kim, S., & Jang, Y. (2017). Predictive visual

analytics of event evolution for user-created context. Journal of

Visualization, 20(3), 471–486.

Cashman, D., Humayoun, S. R., Heimerl, F., Park, K.,

Das, S., Thompson, J., … Chang, R. (2019). A User‐based

Visual Analytics Workflow for Exploratory Model Analysis.

Computer Graphics Forum, 38(3), 185–199.

3. Mohamed NH, Ismail A, Ismail Z, Adnan CWMSW, Raji

MFA. TREND ANALYSIS AND FORECASTING OF

RAINFALL AND FLOODS. 2014.

4. Tariqur Rahman Bhuiyan, Mohammad Imam Hasan Reza, Er

Ah Choy, Pereira JJ. Direct Impact of Flash Floods in Kuala

Lumpur City. ASM Sci J, 11(3), 145-157. 2018.

5. Archer DR, Fowler HJ. Characterising flash flood response to

intense rainfall and impacts using historical information and

gauged data in Britain. Journal of Flood Risk Management.

018;11(S1):S121-S33.

6. Garcia FCC, Retamar AE, Javier JC. Development of a

Predictive Model for On-Demand Remote River Level

Nowcasting: Case Study in Cagayan River Basin, Philippines.

016 IEEE Region 10 Conference (TENCON); Singapore,

Singapore IEEE; 2016.

7. Liu X, Hao L, Liu Y, editors. The Research on Flood Control

and Rainfall Regimen Information System of Poyang Lake

Areas. 2009 International Conference on Environmental