Models and Simulations for Natural Disasters

Regional program ICT-Asia

 


 

 


1. Project summary

 

We purpose to develop studies, models and simulations, for risk analysis and disaster prediction, in environmental sciences.

 

In such assessments, there are typically two steps: the first one analyses and describes what currently occurs, as precisely as possible (with the help of RS, GIS, etc.); the second step tries to build up a predictive model based on the previous analyses, and according to various assumptions.

 

Therefore studied situations will be the matter for probabilistic approaches: random sets, or heuristic random data.

In all cases, simulations can be produced, and serve as go-between from model to reality.

 

 

 

The project takes place downstream image processing in RS and GIS, since it aims to predict from such images, and not to develop techniques for image processing.

 

All domains of remote sensing and GIS are concerned (agriculture, urban areas, forests, floods, tsunamis, meteorology), as soon as they are the subject of risks and disasters.

 

This sort of results are eagerly awaited by policy makers, they can serve as a guide to define strategies for preventing disasters, and also for managing them. They can help in orienting some economical and tourism choices

 

The proposed project aims to reach the following three objectives:

 

  1. to establish a better understanding of global patterns of disasters;

 

  1. to yield local and overall predictions of disasters from the tracking of convenient parameters;

 

 

  1. to serve as platform for modellers and policy makers.

 

 

2. Project members

 

Asian partner A : The Chinese Academy of Survey and Mapping

 

The Chinese Academy of Survey and Mapping, established in 1959, is affiliated to the State Bureau of Surveying and Mapping of China (SBSM). CASM is the largest multi-disciplined comprehensive research institute in China in the field of Geoinformatics, and mainly engaged in the basic theoretical study and application research on Geoinformatics. CASM provides also technical support to national fundamental mapping programs, major engineering survey projects, and important GIS projects. Over more than forty years, CASM has already formed its strong research capabilities, and is taking the leading role in China, in the areas of Geodesy and Geodynamics, Photogrammetry and Remote Sensing, Cartography and GIS, and Spatial Decision Support System. CASM has received over 100 national awards for its excellent research contribution since its establishment. CASM has a group of scientific researchers with reasonable knowledge structure and of strong competence. Among them, there is one academician of CAE (Chinese Academy of Engineering), twenty-three professors, and seventy-seven associate professors. CASM provides educational courses both for master and doctoral degrees as well. Under the CASM, there are a number of high tech companies to provide technical solutions and services.

 

CASM has involved in the researches on natural resources and environment issues since 1980’s. It took the advantages of remote sensing and GIS techniques to study flood control, mainly on flood modeling, risk analysis, and loss evaluation. CASM has been the coordinator of the 12-year project, national land use change monitoring with remote sensing, sponsored by the Ministry of Land Resources of China (MLR) since 1998. Recently, CASM was sponsored by the Ministry of Science and Technology of China (MOST) to do some researches on monitoring techniques on environment change and security of Three Gorges Reservoir area, which include three aspects: land cover/use information extraction, landslide monitoring/alerting, and water source and its environment monitoring. As to the landslide monitoring/alerting, some work has been done on spatial-temporal data series analysis for landslide body information interpretation and automatic extraction, landslide triggering factor derivation models, as well as the process model of landslide event simulation and alert.

 

 

 

Asian partner B : The Centre for Applied Electromagnetics - MMU

 

The Centre for Applied Electromagnetics-MMU (28 members of which 12 are research officers) is established to focus on research and development in Microwave Remote Sensing Applications, Radar Sensor Development, Electromagnetic Interference/Compatibility, and Microwave Circuits and Systems. The research group has advanced expertise both in theoretical and computation analyses as well as experimental studies. Most of the research projects are supported by external sources. Both hardware and software activities are covered in these academic and applied research projects.

 

 

Asian partner C : The Earth Observation Centre (EOC) - UKM

 

The Earth Observation Centre (EOC) is a centre of research under the  Faculty of Social Sciences and Humanities situated at the Universiti  Kebangsaan Malaysia. The Centre is involved in multi-disciplinary research areas that the cover topics such as global change to the environment, conservation biology, geology, impact studies and water resource studies. 
Its research members come from the social sciences, engineering, geology and biological background.
 
  The EOC was established to conduct research in the field of application earth resource satellites and areas associated with issues of global change. The EOC has links to the Southeast Asian Regional Research Information Network (SEARRIN), which is a research organization that consists of member scientists in the ASEAN countries.
 

 

 

Asian Partner E: The Asian Institute of Technology

 

The Asian Institute of Technology promotes technological change and sustainable development in the Asian-Pacific region through higher education, research and outreach. Established in Bangkok in 1959, AIT has become a leading regional postgraduate institution and is actively working with public and private sector partners throughout the region and with some of the top universities in the world. 
 
Recognized for its multinational, multi- cultural ethos, the Institute operates as a self-contained international community at its campus located 40 km (25 miles) north of Bangkok, Thailand.
 
Besides the usual labs and academic buildings, the main campus includes housing, sports, and medical facilities, a conference center, and a library with over 230,000 volumes and 830 print and on-line periodicals.
 

 

 
French Associated partner A : Team Clime - INRIA
 

Clime team has been created to study the links between data on models (data processing, data assimilation) for environmental applications.

 

The complexity of the environmental phenomena as well as the operational objectives, necessitate a growing interweaving between physical models, data processing, simulation and database tools.

This situation is met for instance in atmospheric pollution, an environmental domain whose modelling is gaining a widening importance, either at small (air quality), regional (transboundary pollution) or global scale (greenhouse effect). In this domain, modelling systems are used for operational forecast (short or long term), detailed case studies, impact studies for industrial sites, management of different spatial and temporal scales, coupled modelling (e.g. pollution and health, pollution and economy). These scientific subjects strongly require coupling the model with all available data; these data being either of numerical origin (e.g. models outputs), or coming from raw observations (e.g. satellite acquisitions or information measured on a spatial grid), or obtained by processing and analysis of these observations (e.g. chemical concentrations retrieved by inversion of a radiative transfer model).

 
French Associated partner B : research group Espace - University of Nice

 

The research group Espace-Nice (30 members of which 24 are research-fellows) is commited into research and development activities of two kinds :

i)        Environmental management: forest fires, hydrology and climate, flooding risk, erosion and pollution. Most of the studies concern Mediterranean areas.

ii)       Territorial and spatial dynamics studies: socio-economic urban dynamics, urban risks, processes of coastal urbanization, coastal landscape changes, sustainable transport in metropolitan urban areas of southern France and northern Italy.

In these fields, ESPACE-Nice has established strong partnerships with national and international administrations and local decision makers. The group is specialised into the use and the development of spatial analysis, methods using geostatistical tools, image analysis, GIS and modelling. Most of the researchers are geographers familiar with quantitative as well as systems analysis. Currently, much is done on 3D analysis applied to urban and coastal areas, where sustainable development issues are studied.

 

The research group Espace is the UMR (Unité Mixte de Recherche) of code number
UMR 6012 at CNRS. It is joined to the two CNRS departments which are intitled : Man and Society ;  Environment and sustainable development.

 

 

French leading partner : Laboratoire A2SI – ESIEE

The The A2SI Laboratory is a team of 32 researchers, under the Direction of Prof. G Bertrand. It comprises sixteen Phd students and is organized around the five following topics :

1.       Discrete mathematical structures : discrete topology, graph theory, ordered sets, and discrete models for imaging, information processing and physics.

2.       Imaging : medical and biological imaging, image compression and coding, microscopic image analysis, and image segmentation.

3.       Algorithms and modeling : combinatorial optimization, physical and biological algorithms, distributed and parallel algorithms, stochastic algorithms, and Markov models.

4.       Architecture and software design : machine architectures, mapping algorithms on architectures, and dedicated circuits, and distributed databases, object-oriented languages, and software development methods.

The laboratory A2SI  is the UMR (Unité Mixte de Recherche) of code number UMR 8049 at CNRS, intitledInstitut Gaspard Monge”.
:

 

3. Project description

 

The proposed project is the concern of an “oriented research” and focuses on disasters analysis and prediction. In such assessments, there are typically two steps: the first one analyses and describes what currently occurs, as precisely as possible (with the help of RS, GIS, biologists, etc.); the second step tries to build up a predictive model based on the previous analyses, and according to various assumptions.

As it involves simulations and predictions, the project has to develop some theory (e.g. partial derivative equations, and random sets). Note that a same approach may apply to various types of disasters. Moreover, the descriptions and the predictions that we hope to find for some territories can be extended to other regions.

Note also that all partners have already some experience, and are often currently working on risk questions. Therefore the conditions for a cross fertilization are fulfilled.

 

 

Before telling what we want to do, we would like to indicate what is the matter. It will help to understand the strategy we propose.

 

What is a natural disaster ?

 

Natural disasters that occur on terrestrial environment are those that are triggered due to:

(i)                  internal disturbances/adjustments (e.g. tectonic activity) such as earthquakes, and volcano eruptions. The direct effects of such disasters of endogenic origin include Tsunamis, and landslides/snow avalanches; Atmospheric pollution.

(ii)                atmospheric disturbances (e.g. weather related) include storms, hurricanes, typhoons, sandstorms, tornadoes, and ground water depletions. The effects due to these disturbances include ozone depletion, drought/famine, global warming, and floods.

(iii)               neither endogenic nor exogenic origins, but are essentially due to anthropogenic activity (e.g. CO2 emissions, forest fires, disease and epidemic spread).

 

 

How to study natural disasters?

 

Natural risks and disasters are relevant of the typical three phases, or three steps approach depicted in table 1

 

input data

Processing

Results and feed back

Three types:

1-GIS type data (maps, statistics, etc.)

2- Remote sensing observations

3- Direct in situ information (measures, interviews, etc.)

based on a double distinction :

1- rare or frequent events;

2- dominant feature or not.

 

interpolation;

prediction;

anticipation;

security rules.

Table 1: the three steps approach to disasters analysis

 

In the processing step, the choice of a pertinent approach mainly depends on the fact that some disasters occur rarely (e.g. tsunamis), and others relatively frequently (e.g. some flooding). In the first case, the method is deterministic, and  based on the physics of the phenomenon, in the second one a probabilistic approach can be an alternative. Moreover, when one feature is dominant, then the framework of partial derivative equations is sound, whereas when several aspects interweave, then the stochastic framework can be better adapted. Table 2 summarizes this classification.

 

 

 

 

Rare  events

  Frequent  occurrences

Dominant feature

deterministic approach

behaviour based simulations

partial derivatives; probabilities

Multiple context

multivariate analysis;

deterministic models

stochastic models

(random sets and functions)

Table 2: Double switch classification of the approaches

 

A few examples are given in annex below, for illustrating the three phases depicted in table 1 and the approach differences shown in table 2. They indicate recent researches of the partners (2004-2005), and are proposed as seeds for the applications of the project.

 

Project reach

 

Although the examples of the annex are obviously finalized, their approaches can be transposed to other situations. A landslide in the middle of an Asiatic city, or a severe breakdown of the electric power  and net would provoke the same type of traffic jam as a flooding in Nice. The  propagation of some epidemic might be relevant of the random spread model developed for forest fires, etc.

 

The partners are aware of this versatility, and would like to exploit it. Most of them know their respective works, have experience of same methods (in particular with mathematical morphology), that they appreciate, and sometimes have already worked together.

 

Note also that the Asiatic partners meet practically all types of natural disasters listed in section B2-1. For example, China’s cultivated land decreased by around 12 million ha (over 122 million ha of cultivated land) between 1988 and 2003 due to conversions, settlements... and natural disasters of various types (ref. [&à] in the annex) . Another example is given by  the National Disaster Committee of the Government of Malaysia who has declared five priority disasters, namely forest fires, floods, landslides, oil spills, and hot-installations.

 

Therefore we do not fix on all specific themes at the moment, but four of them only. If the project is accepted, we will rapidly meet and confront the strengths and weaknesses of the various methods, versus the types of disasters. During this meeting, the thing to settle will be the specific cases of study, with corresponding approaches and sub-teams (surely, we shall find much more cases than what we can undertake in two years...). 

 

It is clear that the main horses of the cab (flood and forest fire, presented in the annex) are to be kept. This leads to the following themes

 

a/ Forest fires (ESIEE-UKM) The random spread model (see annex) allows us to predict the burnt scares and to locate them, but up to now, it does not permit short term predictions: if the fire starts here today, what will happen elsewhere in one day, two days, etc.? Also, situations with wind have not been modelled, underground propagation either. These tasks define the matter for one or two Phd theses.

 

b/ Urban traffic and disaster (Un. Nice) . The approach used for the city of Nice (see annex) has to be re-designed in the case of other types of urbanism, and also when the cause or the traffic thrombosis is no longer a river flood but a landslide. Again such a work could be the matter of a Phd thesis.

 

cTide and pollutants (MMU). Tidal systems in which the flooding pattern follows simple propagations can be solved by means of shallow water equations. The study of tidal flats through simulations are based on coupled dynamic systems, as the tidal morphology of the basins is ocean tide dependent. The understanding of the pollutant dispersal on the flat surface is a phenomenon which requires tide simulations in the discrete space.

 

d/ Extreme events (INRIA). The prediction of the meteorological extreme events, such as storm and cyclones, builds on conceptual models. The performances of these highly non-linear models are limited, mainly because they lack observation data. Satellite acquisitions allow the observation of precursors of these events as coherent structures in images. The aim is to establish methodologies for the assimilation of image structures into conceptual models, which will allow improving their performances by constraining their non-linearity.

 

e/ Landscape dynamic under endogenic/exogenic forces (MMU). Landscape dynamics- under the influence of various non-linear effects including perturbations originated from endogenic (e.g. upliftment) and exogenic (e.g. weathering) processes- can be studied using geodesic morphological transformations. The understanding of the landscape evolution under perturbation can be expanded to the study of other terrain related impacts, in particular the delineation of potential zones/paths of landslides.

 

 

Expected outcomes

 

The expected results will be the monitoring, assessment and warning of risks and disasters. The analyses result firstly in maps and information similar to those presented for particular cases in the annex (e.g. prediction of the forest fires places, 3-D maps of the costal areas for beach risks). Secondly, it provides simulations either drawn from assumptions, or from the current data of an actual disaster (e.g. to predict the extension at day j+1 of a fire existing at day j).

 

This sort of results are eagerly awaited by policy makers, they can  serve as a guide to define strategies for preventing disasters, and also for managing them. They can help the end users  (State, economy, tourism, etc.) in orienting some choices.

 

 

4. Four examples of the state of the art among the partners

 

The following examples illustrate the three phases depicted in table 1 and the approach differences shown in table 2. They indicate also recent researches of the partners, which could be seeds for applications for the project.

 

1/Impact of the tsunami on Phi-phi island (X.Chen, AIT)

 

input data : record of the changes by aerial photographs, by direct measures of the height and the directions of the tide, and by witnesses interviews.

Processing : the previous information, and the surrounding maps allow to simulate the two waves (see figures)

                   

           a) topography before the tsunami               b) simulation of the flood provoked by the two waves

 

Results and feed back : the study helps to recover tourist economy by making 3-D tourists maps, by providing input to simple warning systems, and by becoming consultant with hazard maps. Moreover, it yields a better understanding about the tsunamis.

 

2/ Flooding of the city of Nice (Ch. Voiron-Canicio, Un. Nice)

 

This second example looks like the previous one as it deals again with a rare event, but with an event that never occurred up to  now.

input data : In this urban context, the  input data are, on the one hand, the population densities at various moments of the day, and the other one the typology and the traffic of streets and roads

Processing : the previous information, and the surrounding maps allow to simulate the two waves (see figures)

Results and feed back : the simulations permit to estimate the time necessary for  fire service vehicle to travel to senior homes, when flooding. By generating vulnerability maps the method detects the Low accessibility areas .

 

 

 

 

 

 

 

 

 

 

 

 


        a) input data                                                                               b) traffic simulation

 

3/ Random model for Malaysian forest fires  (J. Serra, ESIEE, M.D.H. Suliman, UKM)

 

This study analyses the forest fires in the state of Selangor (Malaysia) during the period 2000-2004. The phenomenon is not a rare event (hundreds of hot spots detected), and it is not governed by one main feature (fires depend on the dryness, on  the ground and underground moistures, on  the type of fuel, on the spread rate, on atmospheric conditions, etc.). Therefore approach can be convenient.

 

input data :All hot spots detected by satellite over the territory of Selangor, and  two risks maps (non represented here), for the spread rate and for the total fuel consumption.

Processing : Definition of the stochastic model of the random spread. By theoretical derivations, one can predict the probability that the fire extinguish after one day, two days, etc. as well as the sizes and locations of the burnt scars (see figures). One can also use the model for simulating the spread of a given seat through the time.

   a) hot spots during 2000-2004           b) burnt scars in the same period    c) Model prediction of the burnt scars

 

Results and feed back : The model predicts the zones where hot spots induce extended fires. Therefore, it orients the warning on specific places, and allows to compare various techniques used to control fire propagation.

 

 

 

References

 

  1. Kim Siang ANG,  Conceptual Model of Malaria Surveillance System for Peninsular Malaysia Master thesis, Un. Technology Malaysia, Dec. 2001
  2. Xiaoyong CHEN Detailed 3-D Mapping of Disaster Affected Areas, DMAI' 2005, X. Chen (Ed.), 4-6 Nov. 2005 A.I.T. Bangkok, Thailand.
  3. B.S. DAYA  SAGAR, Discrete simulations of spatio-temporal dynamics of small water bodies under varied stream flows discharges. Non linear process in Geophysics (2005) Vol.12, pp. 34-40
  4. Isabelle HERLIN,
  5. Jean SERRA and Mohd Dini Hairi SULIMAN  Random Spreads and Forest Fires, submitted to Int. Journal of Remote Sensing Jan. 2006.
  6. Christine VOIRON-CANICIO, Florence OLIVIER, SIG, simulations et détection des espaces à enjeux, SAGEO'2005, 27 June 2005, Nice, France, pp. 1-13.
  7. Jixian ZHANG, Yonghong ZHANG,

 

  1. National Program of Land Use change Monitoring by remote Sensing in China, DMAI' 2005, X. Chen (Ed.), 4-6 Nov. 2005 A.I.T. Bangkok, Thailand.

 

 

 

4/ Discrete simulations of floodplain water bodies under non-linearly controlled streamflow discharges (B.S. Daya Sagar, MMU)

 

Similarities that exist between floodplain and tidal environments include the following aspects:

 

(1)    In both environments, the topographic elevation slope is less than 2 degrees.

(2)    Certain phenomena (e.g. water bodies, channels) in these environments exhibit time dependency in their geometric and/or topologic organisation.

(3)    These phenomena are controlled by exogenic forces (e.g. fluctuating streamflow discharge and tide level patterns).

 

Hence studying certain phenomena is similar to investigating the coupled spatial dynamics. In this study the fluctuating pattern of tide level shows direct impact on the spatio-temporal behaviour. The example shown below is a simulated case of spatio-temporal behaviour of surface water bodies, essentially from floodplain zone, under the influence of simulated streamflow discharges. This is the basic study to have an idea to model and simulate the impacts of periodically fluctuating tide levels on the spatial channels in tidal environment.

 

Input data: Spatially distributed floodplain water bodies over floodplain region of Gosthani river, India, and time-series of streamflow discharge data simulated via first order non-linear difference equation

Processing: The geometric and topologic organisations of floodplain water bodies vary with environmental stresses, such as floods and drought, that occur in various possible patterns (e.g. periodic, aperiodic and chaotic). The flood and drought are the two kinds of cascade transformations that are performed by morphological dilations and erosions by synchronising the simulated time series data of streamflow discharges. One can use this approach to model and simulate the propagation of water in the tidal environment synchronising periodically behaving tide levels.

Results and feedback: The model generates various geomorphic-attractors that predict the spatio-temporal behaviour of water bodies.

 

            

 

Spatio-temporal organization of the surface water bodies under the influence of various streamflow discharge behavioral patterns at the environmental parameters at (a-f) l = 1, 2, 3, 3.46, 3.57, and 3.99 are shown up to 20 time steps. In all the cases, the considered initial Mean Streamflow Discharge, A0 = 0.5 (in normalized scale) is considered under the assumption that the water bodies attain their full capacity. It is illustrated only the overlaid outlines of water bodies at respective time-steps with various ls.

 

 

References

 

  1. Kim Siang ANG,  Conceptual Model of Malaria Surveillance System for Peninsular Malaysia Master thesis, Un. Technology Malaysia, Dec. 2001
  2. Xiaoyong CHEN Detailed 3-D Mapping of Disaster Affected Areas, DMAI' 2005, X. Chen (Ed.), 4-6 Nov. 2005 A.I.T. Bangkok, Thailand.
  3. B.S. DAYA  SAGAR, Discrete simulations of spatio-temporal dynamics of small water bodies under varied stream flows discharges. Nonlinear process in Geophysics (2005) Vol.12, pp. 31-40
  4. Zhengjun LIU, ChangyaoWANG, Aixia LIU,and Xiangming XIAO, 2005. Statistical Ratio Rank Ordered Differences Filter for SeaWiFS Impulse Noise Removal. Photogrammetric Engineering & Remote Sensing, vol.71(1), pp.89-96.
  5. Zhengjun LIU, Aixia LIU, Changyao WANG and Zheng NIU, 2004, Evolving neural network using real coded genetic algorithm (GA) for multispectral image classification, Future Generation Computer Systems, Vol.20(7): pp.1119-1129.
  6. Mastura MAHMUD, Forest Fire Monitoring And Mapping In South East Asia National Seminar On LUCC and GOFC (NASA/EOC), 12 Nov. 1999, Bangi Selangor Malaysia.
  7. Jean SERRA and Mohd Dini Hairi SULIMAN  Random Spreads and Forest Fires, submitted to Int. Journal of Remote Sensing Jan. 2006.
  8. Christine VOIRON-CANICIO, Florence OLIVIER, SIG, simulations et détection des espaces à enjeux, SAGEO'2005, 27 June 2005, Nice, France, pp. 1-13.
  9. Jixian ZHANG, Yonghong ZHANG,

 

  1. National Program of Land Use change Monitoring by remote Sensing in China, DMAI' 2005, X. Chen (Ed.), 4-6 Nov. 2005 A.I.T. Bangkok, Thailand

10.  Jixian ZHANG,  Guosheng LI, Yu ZENG. The study on automatic and high-precision    rectification and registration of multi-source remote sensing imagery. Journal of Remote Sensing, 2005 vol. 9(1), pp 73-77