Sie sind auf Seite 1von 7

Available online at www.sciencedirect.

com

Mathematics and Computers in Simulation 80 (2010) 887893

A fuzzy soft flood alarm model


Sunny Joseph Kalayathankal a, , G. Suresh Singh b
a

Department of Mathematics, K.E. College, Mannanam, Kottayam, 686561, India


Department of Mathematics, University of Kerala, Trivandrum, 695581, India

Received 4 February 2006; received in revised form 14 October 2009; accepted 16 October 2009
Available online 12 November 2009

Abstract
A wide range of hydrological analyses for flood, water resources, water quality, ecological studies, etc., require reliable
quantification of rainfall inputs. This work illustrates a fuzzy analysis that has the capability to simulate the unknown relations
between a set of meteorological and hydrological parameters. A fuzzy approach to flood alarm prediction based on the fuzzy soft
set theory is applied to five selected sites of Kerala, India to predict potential flood.
2009 IMACS. Published by Elsevier B.V. All rights reserved.
AMS Classication: 03B52; 94D05; 68T37
Keywords: Rainfall; Fuzzy soft set; Flood; Simulation

1. Introduction
Rainfall being the dominant component in most hydrological systems, reliable quantification of rainfall is absolutely
essential for various ecological, meteorological, geo-morphological and disaster management studies. Since the
occurrence and distribution of rainfall over a region is controlled by many independent factors, reliable forecasting
becomes a complex exercise. A challenging task for catchment management and flood management in particular is
the creation of a reliable quantitative rainfall forecast. Accurate forecasts of the spatial and temporal distribution of
rainfall are useful for flood warning. A flood warning system for fast responding catchments may require a rainfall
forecast to provide sufficient lead time for early warning.
A flood warning system is a non-structural measure for flood mitigation. Several parameters are responsible
for flood related disasters and a quick-responding flood warning system is required for effective flood mitigation
measures. Major atmospheric parameters affecting floods are rainfall, wind speed, wind direction, relative humidity
and surface pressure. River and topography are two geographical parameters that directly impact rainfall and water
dissipation. This paper investigates the potential of fuzzy soft set theory in real-time flood warning. Molodtsov [6]
initiated the concept of soft set theory as a new mathematical tool for dealing with uncertainties. Soft set theory has
rich potential for applications in several directions, a few of which had been shown by Molodtsov [6] in his pioneer
work. In the present paper we use fuzzy soft sets, establish some results on them and develop an algorithm followed
by simulation for flood warning in five important locations in the state of Kerala, India.

Corresponding author. Tel.: +91 4812596606; fax: +91 4812590915.


E-mail address: sunnyjose2000@yahoo.com (S.J. Kalayathankal).

0378-4754/$36.00 2009 IMACS. Published by Elsevier B.V. All rights reserved.


doi:10.1016/j.matcom.2009.10.003

888

S.J. Kalayathankal, G. Suresh Singh / Mathematics and Computers in Simulation 80 (2010) 887893

2. Fuzzy approach to decision making


The thought process involved in the act of decision making is a complex array of streaming possibilities in which
a person selects of discards information made available from diverse sources. In doing so he is led by a meaningful
analysis of available information and optimal selection out of several apparently equi-efficient decisions. Information
is available on the basis of some criteria and the criterion-values are not crisp but fuzzy sets. In real life situations
the information available to a person would be blurred, indistinct, vague and often ambiguous. Examples of such
bits of information regarding atmospheric conditions would sound like: high surface temperature, heavy downpour,
mild wind and high humidity. Due to the presence of diverse uncertainties it would be impossible to make accurate
predictions following classical models which require exact and crisp information.
Since Zadeh [17] published the fuzzy set theory as an extension of classic set theory, it has been widely used in
many fields of application, such as pattern recognition, data analysis and system control [1,3,5,9,15]. The unique
characteristic of this theory, in contrast to classic mathematics, is its operation on various membership functions (MF)
instead of the crisp real values of the variables. This heuristic permits fuzzy theory to be a powerful tool whenever it
handles imprecise data or ambiguous non-linear relationship between the variables. This theory is extremely effective
in handling dynamic, non-linear and noisy data, especially when the underlying physical relationships are not fully
understood. As hydrologists are still uncertain about many aspects of the physical processes in the watershed, fuzzy
theory proves to be a very attractive tool enabling them to investigate these areas. The past decade has witnessed a
few applications of fuzzy logic approach to flood prediction studies [2,7,8,1014,16,18].
3. Model construction
Denition 1. [6]: Let U be an initial universe set and E be a set of parameters (real-valued variables). Let P(U) denote
the power set of U and A E. A pair (F, A) is called a soft set over U, where F is a mapping given by F : A P(U).
Example 1. Let the initial universe U = {L1 , L2 , L3 , L4 , L5 } be the five selected locations in Kerala, viz., Trivandrum,
Alappuzha, Cochin AP (Airport), Palakkad and Kozhikode and E = {P1 , P2 , P3 , P4 } be the atmospheric parameters,
where P1 , P2 , P3 , P4 are wind speed, wind direction, relative humidity and surface pressure respectively. Suppose that
F (P1 ) = {L1 , L2 , L4 },
F (P2 ) = {L3 , L5 },
F (P3 ) = {L1 , L2 , L3 },
F (P4 ) = {L2 , L3 , L5 }.
Each approximation has two parts
(i) a predicate p;
(ii) an approximate value set.
Consider F (P1 ), here predicate name is wind speed and value set is {L1 , L2 , L4 } (Table 1).
Table 1
Tabular representation of a soft set.
U

Wind speed

Wind direction

Relative humidity

Surface pressure

L1
L2
L3
L4
L5

1
1
0
1
0

0
0
1
0
1

1
1
1
0
0

0
1
1
0
1

Denition 2. [4]: Let U be an initial universe set and E be a set of parameters (real-valued variables). Let P(U)
denote the set of all fuzzy sets of U and A E. A pair (F, A) is called a fuzzy soft set over U, where F is a mapping
given by F : A P(U).

S.J. Kalayathankal, G. Suresh Singh / Mathematics and Computers in Simulation 80 (2010) 887893

889

Example 2. Consider Example 1 (Table 2). Suppose that


F (P1 ) = {L1 /0.7, L2 /0.6, L3 /0.6, L4 /0.9, L5 /0.5},
F (P2 ) = {L1 /0.7, L2 /0.8, L3 /0.9, L4 /0.6, L5 /0.9},
F (P3 ) = {L1 /0.8, L2 /0.8, L3 /1.0, L4 /0.7, L5 /0.8},
F (P4 ) = {L1 /0.9, L2 /0.9, L3 /1.0, L4 /0.8, L5 /0.9}
Table 2
Tabular representation of a fuzzy soft set.
U

Wind speed

Wind direction

Relative humidity

Surface pressure

L1
L2
L3
L4
L5

0.7
0.6
0.6
0.9
0.5

0.7
0.8
0.9
0.6
0.9

0.8
0.8
1.0
0.7
0.8

0.9
0.9
1.0
0.8
0.9

Note: lij corresponds to the location Li and parameter Pj , where lij = membership value of Li in F (Pj ).

Denition 3. (Estimation bank) It is a table in which the number of rows and columns are equal. Both are labeled by
location variables. Entries eij = the number of parameters for which the membership value of Li exceeds or equals the
membership value of Lj .
Remark 1. 0 eij n, where n is the number of parameters.
Remark 2. Each main diagonal element of an estimation bank is always equal to the constant n, where n is the
number of parameters of the system.
Denition 4. (Impact indicator). The row sum of Li s determine total cumulative impact on each location Li s
Iri =

n


eij

(1)

j=1

Denition 5. (Divider factor). The sum of each column in estimation bank tracks the prediction procedure. The divide
factor
n

Dj =
eij
(2)
i=1

Denition 6. (Discrimination factor). The discrimination factor discriminates flood probable location. Discrimination
factor (DF) is defined as
DFi = Iri Di

(3)

3.1. Discrimination algorithm


Reliable flood prediction cannot be done by subjecting available data to conventional methods of analysis. We
therefore turn to fuzzy soft sets and develop a simple but effective model (algorithm) which has been designed in such
a way as to produce reliable output in the prediction of flood possibility. The inputs are basic parameters (real-valued
variables) related to flood occurrence and fuzzy membership grade is assigned to each parameter. The model processes
the fuzzy soft set constructed from collected data and identifies the most flood prone location (the location which shows
maximum discrimination factor).

890

1.
2.
3.
4.
5.
6.
7.
8.

S.J. Kalayathankal, G. Suresh Singh / Mathematics and Computers in Simulation 80 (2010) 887893

Selection of a desired number of locations (m).


Selection of a desired number of parameters (n).
Construction of a fuzzy soft set (F, P) and tabulation.
Construction of estimation bank.
Calculation of impact indicator and divider factor.
Estimation of discrimination factors.
Construction of Ir and DF optimization table.
Identification of flood prone location from Ir and DF optimization table.

Discrimination algorithm is computed using data composed of seven pertinent parameters related to flood occurrence
and relevant conclusions are drawn.
4. Experiment and results
4.1. Study area
The area selected for the study is Kerala, a narrow segment in the south western part of Peninsular India, extending
over a distance of 560 km along the west coast with width varying from 15 to 420 km within a limited area of
38,863 km2 and presents very wide variation in its physical features. Physiographically, Kerala is subdivided into
highland (elev.= >75 m), midland (7.575 m) and lowland (<7.5 m) regions. The lowland (coastal land) is unique in
many ways, viz., high density of population, fragile nature of shoreline, presence of many rivers, estuaries, backwaters
and bays. Natural calamities like flood and coastal erosion are common events in many regions in the lowlands of
Kerala during the monsoon season.
Towards monitoring and assessing the flood system in the coastal lands of Kerala, five locations viz., Trivandrum
(N. Lat. 08:31:00 and E. Long. 76:50:00), Alappuzha (N. Lat. 09:30:00 and E. Long. 76:50:00), Cochin AP
(N. Lat. 09:54:00 and E. Long. 76:16:00), Kozhikode (N. Lat. 11:17:01 and E. Long. 75:50:00) and Palakkad
(N. Lat. 10:45:00 and E. Long. 76:45:00) have been selected (Fig. 1). The first four locations are in the low to
mid-coastal region especially vulnerable to flood and have a relatively higher density of population. Palakkad, the
Rice Bowl of Kerala, a plateau devoid of a shore face, is located in the midland. The selected locations are natural
laboratories in the tropical river systems of Kerala offering representative geographical regions enabling flood related
studies based on the model discussed (Table 3).
Table 3
Catchments descriptions.
Label

Location

Geographical description

Topography

Rivers/backwaters

L1
L2
L3
L4
L5

Trivandrum
Alappuzha
Cochin AP
Palakkad
Kozhikode

Coastal midland
Coastal lowland
Coastal lowland
Inland (midland)
Coastal midland

Moderate
Low
Low
Moderate
Moderate

River = 1, backwater = 1
Backwater = 1
River = 1, backwater = 1
River = 1
River = 1

4.2. The result


The study intends to predict the possibility and severity of floods in five selected stations in Kerala. We use fuzzy
soft set properties to establish an algorithm for a reliable prediction. Simulations are done using Java Server Pages.
The five selected locations in Kerala, Trivandrum, Alappuzha, Cochin AP, Palakkad and Kozhikode are denoted by
L1 , L2 , L3 , L4 , and L5 . The parameter set E = P1 , P2 , P3 , P4 , P5 , P6 , P7 respectively denotes wind speed , wind
direction, relative humidity, surface pressure, river contribution, topography, and rainfall amount (Tables 46). The

S.J. Kalayathankal, G. Suresh Singh / Mathematics and Computers in Simulation 80 (2010) 887893

891

Fig. 1. Catchment location map.

fuzzy soft set (F, E) = {{P1 }, {P2 }, {P3 }, {P4 }, {P5 }, {P6 }, {P7 }} where
{P1 } = {L1 /0.7, L2 /0.6, L3 /0.6, L4 /0.9, L5 /0.5},
{P2 } = {L1 /0.7, L2 /0.8, L3 /0.9, L4 /0.6, L5 /0.9},
{P3 } = {L1 /0.8, L2 /0.8, L3 /1.0, L4 /0.7, L5 /0.8},
{P4 } = {L1 /0.9, L2 /0.9, L3 /1.0, L4 /0.8, L5 /0.9},
{P5 } = {L1 /1.0, L2 /0.5, L3 /1.0, L4 /0.5, L5 /0.5},
{P6 } = {L1 /0.5, L2 /1.0, L3 /1.0, L4 /0.5, L5 /0.5},
{P7 } = {L1 /0.5, L2 /0.6, L3 /1.0, L4 /0.4, L5 /0.9}
4.3. Sensitivity analysis
Predictions based on collected data coincide with the actual flood rate. The parameters under investigation are easily
susceptible to changes in temporal and topographical conditions. Sometimes the combined effect of minute variations
of critical parameters may adversely affect the accuracy of prediction. Moreover, if the number of parameters increases,
changes in estimation bank input causes laborious computational work.
4.3.1. Critical ratio test for red spots
The chances for sudden changes in parameters often create unexpected results. The ID (impact divider) ratio gives
us some insights into the possible impacts such variations can produce. If ID ratio is greater than one, the cumulative

892

S.J. Kalayathankal, G. Suresh Singh / Mathematics and Computers in Simulation 80 (2010) 887893

Table 4
Fuzzy soft set from collected data.
U

P1

P2

P3

P4

P5

P6

P7

L1
L2
L3
L4
L5

0.7
0.6
0.6
0.9
0.5

0.7
0.8.
0.9
0.6
0.9

0.8
0.8
1.0
0.7
0.8

0.9
0.9
1.0
0.8
0.9

1.0
0.5
1.0
0.5
0.5

0.5
1.0
1.0
0.5
0.5

0.5
0.6
1.0
0.4
0.9

Table 5
Estimation bank.
U

L1

L2

L3

L4

L5

L1
L2
L3
L4
L5

7
5
6
2
5

4
7
7
2
5

2
2
7
1
1

6
6
6
7
6

5
5
7
3
7

Table 6
Optimization table.
U

Impact indicator

Divider factor

L1
L2
L3
L4
L5

24
25
33
15
24

25
25
13
31
27

Discrimination factor
1
0
20
16
3

impact of parameters affects the location. If ID ratio is less than unity the location will not be affected by such changes.
If ID ratio is unity, the location is minimally affected by the changes in the parameters. We call that location a Red
spot. This is because any fluctuation in the estimation bank input critically affects the result. In such cases instead of
constructing a fuzzy soft set for a given set of values, we take a sequence of values for each parameter which forms a
real sequence. If it converges, the corresponding parameter shows consistency. Otherwise the parameter is significant
in determining the nature of uncertainty of the prediction. If the convergent criterion for all input values are satisfied,
this estimation bank gives us reliable results.
4.3.2. Analysis of behaviors of parameters
Based on data collected from the areas of study, an estimation bank chart for the study of the behavior of parameters
has been developed. In all the fuzzy soft set tables constructed from the collected data, the topography and number
of rivers/backwaters parameters show consistency in values and are treated as consistent parameters. They are
relatively transparent and can be kept relatively steady factors in prediction. On the other hand, other parameters under
consideration show variation spatially and temporally and are treated as inconsistent parameters. The study of such
parameters requires continuous data over a long period and analytical tools are to be applied in the investigation. Of
the five inconsistent parameters, wind speed and wind direction are the cardinal ones and are primarily responsible
for the creation of Red spots.
5. Conclusion
The fuzzy soft set theory approach presented for flood warning system furnishes very promising results and
possibilities. Five meteorological parameters collected for sixteen years (19811996; source: IMD, Pune, India)
and two geographical parameters were analyzed and a membership grade was given to each parameter. The output

S.J. Kalayathankal, G. Suresh Singh / Mathematics and Computers in Simulation 80 (2010) 887893

893

generated using JSP identifies Cochin AP followed by Alappuzha as the most flood prone areas and the finding is
consistent with observed data. Greater accuracy could be obtained by the inclusion of additional parameters such as
roads, embankments and railroads, that stand in the way of flood dissipation. A sensitivity analysis of the proposed
model provides a new fillip to disaster management studies.
Acknowledgements
The author wishes to thank Dr. Sabu Joseph, Selection Grade Lecturer, Dept. of Environmental Science and
Prof. Jossie Jerome, Department of English, K.E. College, Mannanam, Kottayam, Kerala, India, for their valuable
suggestions in the preparation of this paper.
References
[1] S.M. Chen, A new approach to handling fuzzy decision making problems, IEEE Transactions on Systems, Man and Cybernetics 18 (1988)
10121016.
[2] C. Ertunga, Lucien Duckstein, Fuzzy conceptual rainfallrunoff models, Journal of Hydrology 253 (2001) 4168.
[3] R. Kruse, J. Gebhardt, F. Klawonn, Foundations of Fuzzy Systems, Wiley, New York, 1994.
[4] P.K. Maji, R. Biswas, A.R. Roy, Fuzzy soft sets, The Journal of Fuzzy Mathematics 9 (3) (2001) 589602.
[5] P.K. Maji, R. Biswas, A.R. Roy, On soft set theory, Computers and Mathematics with Applications 45 (2003) 555562.
[6] D. Molodtsov, Soft set theory-first results, Computers and Mathematics with Applications 37 (1999) 1931.
[7] P.C. Nayak, K.P. Sudheer, K.S. Ramasastri, Fuzzy computing based rainfallrunoff model for real time flood forecasting, Hydrological Processes
19 (2005) 955968.
[8] P.-S. Yu, S.-T. Chen, C.-J. Cheng, T.-C. Yang, The potential of fuzzy multi-objective model for rainfall forecasting from typhoon, Natural
Hazards 34 (2005) 131150.
[9] Z. Pawlak, Hard Set and Soft Sets, ICS Research Report, Institute of Computer Science Poland, 1994.
[10] L. See, S. Openshaw, Applying soft computing approaches to river level forcasting, Hydrological Sciences Journal 44 (5) (1999) 763779.
[11] S.-L. Chen, The application of comprehensive fuzzy judgement in the interpretation of water-flooded reservoirs, The Journal of Fuzzy
Mathematics 9 (3) (2001) 739743.
[12] M. Stuber, P. Gemmar, M. Greving, Machine supported development of fuzzy-flood forecast systems, in: A. Bronstert, C. Bismuth, L. Menzel,
(Eds.), European Conference on Advance in Flood Research, Potsdam, PIK Report 65, Reprint of Proceedings, vol. 2, 2000, pp. 504515.
[13] S.J. Kalayathankal, G. Suresh Singh, Need and significance of fuzzy modeling of Rainfall, in: Proceedings of the National Seminar on
Mathematical Modeling and Simulation, Department of Mathematics, K.E. College, Mannanam, Kerala, India, August, 2007, pp. 2735.
[14] S.J. Kalayathankal, G. Suresh Singh, IFS model of flood alarm, Global Journal of Pure and Applied Mathematics 9 (2009) 1522.
[15] S. Theodoridis, K. Koutroumbas, Pattern Recognition, Academic Press, New York, 1999, 482483.
[16] E. Toth, A. Brath, A. Montanari, Comparison of short-term rainfall prediction models for real-time flood forecasting, Journal of Hydrology
239 (2000) 132147.
[17] L.A. Zadeh, Fuzzy sets, Information and Control 8 (1965) 338353.
[18] Zekai Sen, Abdusselam Altunkaynak, Fuzzy awakening in rainfallrunoff modeling, Nordic Hydrology 35 (1) (2003) 3143.

Das könnte Ihnen auch gefallen