ICROS-SICE International Joint Conference 2009

August 18-21, 2009, Fukuoka International Congress Center, Japan

Flood Monitoring and Early Warning System

Integrating Object Extraction Tool on Web-based
Ponthip Limlahapun1 and Hiromichi Fukui2
1
Keio Research Institute at SFC, Keio University, Kanagawa, Japan
(Tel : +81- 466-49-3497; E-mail: [email protected])
2
Faculty of Policy Management, Keio University, Kanagawa, Japan
(Tel : +81-466-49-3497; E-mail: [email protected])

Abstract: The paper will develop a flood monitoring and warning system by integrating meteorological and
hydrological data. The mechanism for gathering, detecting, managing, examining, and distributing data will be
established in order to convey urgent messages, during the time of crisis. The goal is to prevent catastrophe occurrence.
The development of algorithm to extract the inundated areas will be operated over the user-friendly web based interface.
It is a significant improvement method for the flood alert and warning system. It overcomes the fix cost of operation and
platform issues. It will reduce the operating expenses and time consuming. Furthermore, it does not require technical
expertise to predict the rise of flood disasters.

Keywords: Flood, Meteorological, Hydrological, Spatial Data, Monitoring, Early Warning, Object Extraction

1. INTRODUCTION such as Google Map, Google Earth and World Wind,

offers the map visualization to the public. Millions users
Natural disasters around the world cause widespread from all over the world make use of the service which is
destruction of property, physical injuries and deaths. easy, hasty and effective. Such web mapping services
Even though natural disasters can be unpredictable applications provide directions, interactive maps, aerial
occurrences, mitigation measures can reduce the severity images, bird’s eye view, and so forth. They are efficient
of damages. The effective preventive strategy can lessen in viewing areas or finding location(s) in normal
the impact of natural hazards. Thus, information, through situations. However, those services do not function as a
data accumulation from reliable sources, is vital. real time disaster-warning tool. They do not serve for
Effective and timely data delivery is required for risk the purpose of natural disaster warning such as floods.
assessment and risk management in the community. The reason is because the procedure involves updated
Floods are one of the most common natural hazards in related data, satellite imagery, land use/land cover,
the Asian region. Increasing in human activities causes inundated areas and etc. which mostly recorded by the
higher risk of flood hazards and their level of impact. local organization.
Therefore, the integration of spatial data, meteorological Immediate response to a flood event is the key
and hydrological data acquired through the Internet command for disaster management program. This
communication can be operated as an early warning includes methods for accessing urgent data and prompt
system to reduce potential damage. communications in three stages, which are prior to an
emergency period, during the phase of an emergency
2. PROBLEM STATEMENT period, and the post emergency recovery phase. Satellite
images are promising for flood disaster locations due to
The problem in the disaster management is not lack the frequency of retrieving images. A prompt manner to
of technology [1] or existence of the relevant detect the inundated areas from the images is also vital.
information. It is often the lack of accessibility of the Several approaches to object oriented image
information. The capability to efficiently use information processing have been proposed. Commercial software
technologies to discover and manage information, the such as Definiens eCognition provides an image
ability to critically evaluate, and appropriately apply processing, the object based. Traditional image
such information to solve a problem, are the key spirit of classification at a single pixel based or known as pixel
investigation and determination to find out disaster based classification. Each single pixel with similar
recovery solutions. Therefore, the means of identifying, spectral characteristics will be grouped into the same
locating and accessing data and determining if the need class. The object-based image is different from the
to efficiently respond to the problem has been adequately pixel-based image in its meaningful image objects and
met, is crucial. their mutual relationships [2]. Neighboring pixels likely
The Internet technology service is broadly recognized present the same land cover class. This results in
as a method for distributing information. It is served as carrying out the pixel spatial, texture, form and shape of
two-way communications between service providers image information. The use of object based imaged
and users. As a result, a large number of web-based processing has been increasing for different applications,
applications and services including web-mapping for example, transportation, urban mapping and etc.
services have been increasing. The Map service provider,

- 2639 - PR0002/09/0000-2639 ¥400 © 2009 SICE

Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on April 05,2023 at 08:07:16 UTC from IEEE Xplore. Restrictions apply.
 3. FRAMEWORK OF THE STUDY Satellite images, and In-situ sensor data into
Web-based System.
3.1 Framework 3.2.4 Establish a test-bed flood early warning system
The system engaging in automated collection, to observe, process and send an emergency
monitor, processing and timely spatial data warning message.
dissemination of flood disaster is an urgent need. One of
the main functions, which has not been found, is to It is essential to have a receiving data system to detect
develop the algorithm to detect the inundated areas the changes in critical variables, timely disseminate
through an interactive system. It is an essential means emergency notifications, and provide early warnings [3].
for spatial information system, which can develop It could cause serious problems if such system is not
useful applications in the events of flooding crisis. It available to the public. It could be the cause of the loss of
serves to ensure the increased accessibility to spatial life and properties in some sensitive areas.
data and its information and to enhance the Spatial data is an essential data to identify the areas
effectiveness and efficiency in responding to a flood where flooding takes place. Additionally, processing
application. satellite images require remote sensing software and
technical knowledge. The attempt is to establish the
system allowing users to pre-detect the inundated areas
through web based which can reduce time and avoid the
software expenses. The images result can be downloaded
for further analysis.
This research utilizes the weather data, one of the main
data set supporting flood monitoring provided by
National Oceanic and Atmospheric Administration
(NOAA). Thus, implementing a systematic process to
retrieve and analyze data based on meteorological and
hydrological forecast is required.
The test-bed system through this research paper is
customized to observe, investigate and send emergency
communications and early warning signals to the
relevant organizations and authorities. The system will
Fig.1 Workflow Overview adopt a broad array of communication methods, such as
setting up a web page, sending email to inform of an
Fig 1 shows the workflow of the study. The whole emergency evacuation, and passing warning messages
scheme is to setup technological infrastructure to and other critical disaster information. This approach is
facilitate harmonious decision-making, and to to ensure the increasing accessibility of disaster
implement and develop algorithm for geospatial data information and to enhance the effectiveness and
processing. In situ and sensor data will be gathering efficiency in response to the disaster.
from the source and stored in data disseminating The finding of this research paper can be used to
platform which will centralized all the relevant prevent natural hazards. It can be applied to any areas
information. The initial processing to identify the where flooding occurs. It will be beneficial to technical
inundated areas from the satellite image is expanded and non-technical users.
into the system.
In addition, security system is considered at the 4. METHODOLOGY
beginning stage, design system, to ensure that database
and all activities through the system has been verified At the beginning stage, each module has been
and secured. Security system is applied to information designed in order to program in systematic.
flow control, inspection, communication, and user Module 1: The management system module is to
access control. identify role of authorized person to login to the system.
The system administrator responsibility to create the
3.2 Objectives internal user account in order to authorized the related
persons into system. Common user can view and
The objectives of the study are defined as follows; receive data at the necessary level for example flood
3.2.1 Develop algorithms for the feature extraction warning, inundated areas and etc.
to identify the inundated areas. Module 2: The Observing system module receives
3.2.2 Develop meteorological and hydrological data meteorological data for example rainfall from the
receiving system. NOAA ftp site and store to the database. An automated
3.2.3 Integrate spatial information from different receiver, a mean to collect meteorological observation
geo-spatial data sources, for example: GIS data, data, is accomplished to communicate between the
system and the sources of data provider. Though the
data is less flexibility than recording by human, it is

- 2640 -
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on April 05,2023 at 08:07:16 UTC from IEEE Xplore. Restrictions apply.
 much consistent in a time series and a better observe in gradient magnitude above the threshold value
a world dynamic environment changes these days. correspond to edges. In the other word, the pixel
Moreover, the system can react when a specify event is location where the magnitude of the gradient is larger
detected. For example, the system can be linked to alert than a threshold value, these locations typically occur at
in a case of emergency more rapidly and better decision the boundaries of the object. Filtering has been applied
making. The system will produce the report and before structuring the image for the final output shows
summarize in the table format. the inundated areas.
Module 3: The Monitoring system module starts once The inundated areas extraction will serve for both
the urgent situation occurs. In the case of emergency, technical and non-technical users. General users who
the system will send the email and mobile message to has satellite image in hand can avoid troublesome of
monitoring staff and disaster analyst. Monitoring staff(s) image processing hardware and software cost and
and disaster analyst will be the responsive persons who technical knowledge. Knowledgeable user can benefit
receive alert message if the data records exceed the from this service as well. It can reduce time at the office
level. desk and be able to manage at anywhere, anytime where
the internet access is available.

5. RESULTS

5.1 Observing and Restoring Data

The case study was conducted between April 18th and
10th May, 2008 at the Imja Glacial Lake, Nepal [4]. The
field server with meteorological and hydrological sensor
data was set up at the end of moraine to monitor the
Fig.2 Verified Situation change of temperature, water level and etc. Dynamic
and near real-time data had been gathered from in-situ
Module 4: The Communication system module is to sensor data to National Agriculture Research Center
convey alert message from authorized person to the (NARC). Keio University retrieves data from NARC
public users. Director is the main actor who approve for server and stores into the database. The process of
warning situation while disaster analyst takes action storing, retrieving and analyzing information is
promptly. Disaster analyst will identify the risk and completed by database management module.
affected areas and display on the web mapping. In addition to the in situ data, the researcher also
retrieved meteorological data from NOAA ftp in order
to strengthen and cover the entire geographically area.
The interactive web browser allows users to browse and
query the data instantly. Current and historical data are
available for viewing in the table mode. Graph for each
weather station can be visualized in each parameter
which users interest. Summarized data generated from
the minimum, the average, and the maximum amount of
each meteorological data. It gives users an overview of
condition at the remote site. Charts are generated in
form of visually appealing and easy to understand.
Fig.3 Decision Making Additionally, the system allows the users to customize
the summary records as desired. The result from the
Module 5: Object based image processing, the observing components, immediate and constantly data
additional module embedded to the system, defines as release are vital mean of monitoring disaster events. It
the image processing which focuses on objects for helps the research organization enhances efficiency of
example shape texture and the relations to other objects. crisis events management.
In this research, Matrix Laboratory (MATLAB) The Internet is a path to the worldwide connection,
employs as the backend prototyping system. It is a which enables global communications these days.
programming environment which integrates Near-real time receiving system can achieve its goal
computation, visualization and programming for data through the Internet network. Computers networks that
analysis, graphic visualization, and application connect to various types of communication channels are
development. MATLAB has been proved to be one of a mechanism to transmit data/information from one
tools for academic research in mathematics, engineering, place to another. Global connection technology is the
science and industry. Algorithm is developed to extract fastest and the most effective means of communication.
the inundated area. Edge Detection employed to define Worldwide communications become more accessible to
the edge from an input image by estimating the gradient public at large nowadays. Local organizations can stay
magnitude. Threshold value will be identified which alert, act promptly to emergency events, and
consequently reduce the loss of lives and properties

- 2641 -
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on April 05,2023 at 08:07:16 UTC from IEEE Xplore. Restrictions apply.
 from tragic disasters. 5.4 Spatial Web Portal
User-friendly web interface is the interactive means
5.2 Early Warning System
of communication, which is simple, efficiency and
Following the observation in Imja Glacial Lake, effective. It allows smooth transfer of spatial
warning system test was conducted as one of the main information to the end users. It is developed to provide
activities for the second field survey from April 18th to easy access to all users who can make query, and view
10th May, 2008. The first field survey was performed in GIS data over the Internet. Accordingly, this interactive
November of 2007 in the setting of the meteorological Spatial Web mapping will be comprised of necessary
monitoring device. There was no warning system tool customized function tools such as zoom in, zoom out,
set up in the area during the first field trip. The early query information, and etc.The spatial web portal
warning system was successfully established and proved system will support the areas in which extracted from
sending the alert text messages from e-mails to mobile the satellite images from the inundated extraction
communication devices and e-mails. The constrained module.
coverage for mobile networks makes it unable to send Integrated data can give the overall and up-to-date
message from mobile to mobile via SMS gateway. picture of the events as opposed to providing an isolated
Web-based technology is the service with low cost. It picture like in the old days. Observation, analysis and
does not require specialists’ knowledge in the planning will be achieved more precisely and effectively
communication field. The well-designed system depending on these sets of integrated data. Geo-Spatial
facilitates the entire process from receiving data system, Web based system has been established for collecting
detect changes in critical variables, publish the and distributing geo-information. The purpose of
emergency notifications, to provide early warnings. exploiting Web based GIS system is to overcome the
problem with users’ technical skill. It is the goal to
5.3 Inundated Areas Extraction upgrade from a non-interoperability operating system to
At the initial stage, the inundated areas, which were an integrated sharing system [6]. Various geo-spatial
pre-define through the web page individual module, data sources will be accessible through the web-based
were tested. The Landsat satellite image in coastal GIS system. Consequently, it will help reduce
Vietnam was used (data source: Geoinformatics Center investment cost and eliminate duplicate data collection
[5] through mini-project training program, 2007). It if data exists in a more efficient way.
ensures that the inundated extraction module can be A Geospatial web-based system is an emergency
applied to any areas. The algorithms have been safeguard mechanism. It is prepared for emergency
developed with different threshold value combination. communications. It serves as pre–and
The inundated areas can be detected, additionally, some post-communication methods that respond to the
locations that is not watering areas also appears in the emergency situation, including an emergency recovery
final output report. In order to obtain the optimal result, plan. As a result, it will improve decision-making
the final approach study will be expanded to the process and better planning will enhance disaster
interactive web based. This approach can benefit preparedness scheme.
various users in different geographical areas by variety
of disaster monitoring applications. 6. CONCLUSION

The development of a flood-monitoring tool and

information system is expected to be a practical strategy.
The integrated inundated areas extraction tool is used to
identify and establish infrastructure to derive inundated
information from satellite images. This study will use
object based extraction techniques, which serves as an
alternative approach for the users.
Observation system will be worthless if the
forewarning program has not been established. The
proposed early warning system is the transformation of
traditional communication into web-based system,
which is more effective because it has a better
coordination information management system. It
ensures increased accessibility to inspect the emergency
event and enhance effectiveness and efficiency in
responding to a catastrophic incident. The spatial data
sharing, real time monitoring, and early warning
mechanism, is targeted to be the most beneficial system
to the community for decision-making and evacuation
Fig.4 The result of testing extract inundated areas planning purposes.
The originality and cores of this research is to

- 2642 -
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on April 05,2023 at 08:07:16 UTC from IEEE Xplore. Restrictions apply.
 develop the creative tools to analyze and immediate actual plan needs a strong collaboration and awareness
extract information through the web based system. The among the authorities and the local people in order to
goal is to create easy use web interface and image generate timely warnings. The active communication
processing cost saving which is efficiency enough to site of the entire system should be implemented by
identify the flooded areas at the beginning stage. The providing practical and productive information and
proposed system can be widely utilized by technical and encouraging the users to utilize the system.
non-technical users as well. The system concept can be
applied to various disaster monitoring capacities. Its REFERENCES
concept will broaden research study area for future
development of real time disaster monitoring. [1] Louhisuo et al., (2005), “A disaster information and
monitoring system utilizing earth observation”,
7. DISCUSSION AND FUTHER WORK Management of Environmental Quality: An
International Journal. Vol. 18 No. 3, 2007 pp. 246-262.
The study of flood monitoring and early warning Emerald Group Publishing Limited.
system, by an integrated object extraction tool on [2] Martin Baatz et al., 2001. "eCognition User Guide
web-based system, can be useful for certain research 2.0 : Object Oriented Image Analysis." Definiens
project such as the Sentinel Asia Project [7]. It will Imaging GmbH, Trappentreustrasse 1, 80339 Munchen,
enhance efficiency of crisis events management. Area Germany.
network expansions as well as the availability of [3] P Limlahapun, et al. Early Warning System for
compromised data are vital. The integration of the Glacial Outburst Flood, Imja Lake. Asian Association
meteorological and hydrological information from on Remote Sensing (ACRS), 2008, Colombo, Sri
in-situ data and satellite images contribute significantly Lanka.
to a more precise flood analysis. It makes forecasting [4] Keio University, 2008. Imja Glacial Lake
more effective, reliable and accurate. The Global Flood Monitoring,
alert system (GFAS) [8] was developed by the http://de04.gsec.keio.ac.jp:9080/glacier/index.html
International Flood Network [9]. The system provides [5] Geoinformatics Center, http://www.geoinfo.ait.ac.th/
heavy rainfall information, which is one of the main [6] Klopfer M, 2005. Interoperability & Open
components for flood forecasting. Architectures: An Analysis of Existing Standardisation
However, there are factors to fulfill the gap in the Processes & Procedures. July 27, 2005.
disaster management system. For example; hydrological [7] Sentinel Asia, Disaster Management Support System
data such as the amount of water runoff and the water in the Asia-Pacific Region,
level in a river, which cause flooding, will need to be http://dmss.tksc.jaxa.jp/sentinel/
considered; the satellite image acquisition and the [8] Global Flood Alert System
supporting data, which is generally collected by the local http://gfas.internationalfloodnetwork.org/gfas-web/
organization such as land use/land cover, road, etc., [9] International Flood Network (IFNet),
should be automated received and included in the data http://www.internationalfloodnetwork.org/
for the evacuation and management plan. The followings [10] Simon Cox (2007) Observations and Measurements
are issues that need to be further discussed. – Part 1 – Observation schema, Open Geospatial
- The Sensor Web Enablement (SWE): It is a Consortium (OGC), Reference number OGC 07-022r1.
working group initiated by Open Geospatial Consortium, Final Version, December 2007.
Inc. (OCG). The study group will focus on how to
implement the standard specifications to connect the
earth observation devices to web based [10]. The further
development of the attempts to request satellite images,
through sensor observation system, by applying OGC
observation and measurement standard, will be
monitored. The purpose is to access to sensor data and
serve the user, which has no images in hand.
- Hydrological Data: The water level and
rainfall from the field data will be collected in order to
calculate the discharge to the river. This will be
implemented the method to integration the information
from in-situ data and satellite images.
- Areas extraction algorithms: The inundated
areas extraction algorithm has to be further investigated
to improve certain flooded boundary to achieve better
results. Satellite images from various providers such as
aster images and other areas around the world need to
be tested.
- Active communication and publication: The

- 2643 -
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on April 05,2023 at 08:07:16 UTC from IEEE Xplore. Restrictions apply.