Assessing Water Quality Index in River Basin : Fuzzy Inference System Approach

1 Herman Umbau Lindang, 2 Zamali Hj Tarmudi, 3 Ajimi Jawan 1School of Biological Sciences, Faculty of Applied Sciences, Universiti Teknologi MARA Shah Alam, 40450 Selangor. 2 Department of Mathematics, Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA Sabah, Locked Bag 71, 88997 Kota Kinabalu, Sabah. 3 Department of Biological Sciences, Faculty of Applied Sciences, Universiti Teknologi MARA Sabah, Locked Bag 71, 88997 Kota Kinabalu, Sabah 1 Corresponding Author : email: hermanumbau@gmail.com Tel.: 60013-8362140, Fax: 6088-325164

Cite this article as: Assessing Water Quality Index in River Basin : Fuzzy Inference System Approach.Herman Umbau Lindang, Zamali Hj Tarmudi, Ajimi Jawan / Mal.J. Geo 1(1) (2017) 27-31 2.0 PROBLEM IDENTIFICATION Nowadays, different water quality assessment proposed by international researchers shows that each assessment have their own emphasize based on the water bodies main function.The current range implemented by Department of Environment through DOE-WQI for classi ication of parameters indexes showed less l exibility in judging the concentration of a parameter that is close to the boundary.Inputs of hydrological data into interval sets of data had shown the uncertainties of inputs in evaluating the data.Uncertainties of inputs can be evaluate effectively by using fuzzy logic in water assessment.Therefore, this paper proposes to assess the water parameter by applying into FIS to execute the water assessment 3.0 THE BASIC CONCEPT AND PROPOSED METHOD 3.1 Fuzzy Inference Systems Hydrology data exist in a crisp data and being classi ied into certain range respective to its parameter.The existence of interval data in classifying the water quality shows the existence of subjectivity in the classi ication process.Similarly Fuzzy Inference System (FIS) has been an effective and holistic tool to evaluate and execute any forms of subjective data into a single output.In this paper, the current water quality classi ication will be adapted into FIS for the execution of water status.The constructed outcome from the .

Proposed Methods of FIS into Water Quality Assessment
Step 1: Fuzzy Inputs This paper proposes the uses the inputs of hydrology data into the membership functions as the irst steps.It was implemented in the inputs of the FIS as membership functions.Every classi ication of the parameters with regards to its possible class were represented using membership functions.A membership functions (μ) transforms the real value obtained into a [0,1] value.Trapezoidal membership functions (TrapMFs) de ine the input transformation of the FIS and were represented as in expression (1).
where x is a water quality variable; a, b, c and d are membership parameters.
Table 3 shows the value of each membership parameters to be adapted into expression (1).
Cite this article as: Assessing Water Quality Index in River Basin : Fuzzy Inference System Approach.Herman Umbau Lindang, Zamali Hj Tarmudi, Ajimi Jawan / Geo.Behav.. 1(2) (2017) 27-31 Step 2: Fuzzy Operators The membership degree of each part of rule antecedent is computed after the inputs are fuzzi�ied.Three fuzzy operators as had shown in expression ( 2), ( 3) and (4) were used.The operators are union (OR), intersection (AND) and negation (NOT).
Step 3: Inference Rules (Reasoning Process) Subjectivity may refer to the speci�ic interpretations of any aspect of experiences.Likewise in this paper, it refers to the possibilities of the crisp data input in the classi�ication of data in the forms of interval set used in evaluating the water.As reported in the annual report of Department of Environment (2014), water experts' uses linguistic expression such as Class 1, Class 2, Class 3, Class 4 and Class 5 to represent the status of the water.The sets of classi�ication constructed used in this paper were described.Dissolved Oxygen The terms representing each set have the following meaning: C1 as Class 1, C2 as Class 2, C3 as Class 3, C4 as Class 4 and C5 as Class 5.As an illustration of application on River A, if the dissolved oxygen (DO) in the water is Class 1, the ammoniacal nitrogen (NH3N) level is Class 1, the pH is Class 1 and the level of turbidity is Class 1, then the expected water quality is excellent.These linguistic forms of information can be interpreted into fuzzy language.The robustness of the systems also depends on the number and quality of the rules constructed for the evaluation using FIS.As demonstrated in this paper, there were 250 rule constructed and it re�lects the possible inputs of the total parameter involve in the assessment.To illustrate some of the sets constructed to represent the parameters used in this paper, the �irst 6th rules and the 250th rules were described as follows.Step 4: Aggregation The membership function will be aggregated and produce a single output after the being used different set of rules and being matched with fuzzy outputs (μ_R).The combination of the rules is called aggregation.The aggregation used to fuzzy union all output in the FIS is the maximum methods (Carbajal-Hernández et al., 2012).
Step 5: Defuzi�ication Next, the different water quality condition obtained in a graph will have be obtained.Centroid function (CF) returns the center of area under the curved formed by the output fuzzy function according to expression 6: The output of the center of area by centroid function determines the input value to be classi�ied into the classi�ication of water status from Poor to Excellent accordingly.The different water quality status from poor to excellent can be within this range and normalization of results was done using expression 7. The output value of the �inal evaluation was in the range of [0,1].

4.0
IMPLEMENTATION AND DISCUSION To demonstrate our proposed method applied, we adapted one of our water quality data taken from Inanam River, Sabah.The data was tabulated in Table 4 * Average range of each parameters Given a situation that obeys the rule constructed in Rule 196,197,198 and Rule 199, having their parameters DO, NH3N, pH and Turbidty and their values of 1.347 mgl-1 ,0.673 mgl-1,8.84mgl-1 and 643.667NTU respectively.
Using the propose methods stated in expression (1) untill ( 7), the water quality index can be evaluate using the FIS.The execution of assessment was computed using Matlab 2015b as had shown in Figure 3 Figure 3: Fuzzy inference diagram for the water quality problem with four parameters and 4 rules.Rules 196,197,198 and 199 were used to exemplify the defuzzi ication process.
Based on the calculation computed using the data taken, the river in Inanam Likas River at the speci ied location was 0.155 and classi ied as Poor.It indicates the river is at it worst water quality.As re lected in Table 2, Class 5 of water status was not suitable as habitat for the aquatic ecosystem and utilize by humans.Poor condition of the river shows that the river is unhealthy.It affects the food web and the natural function of the river (Aweng, Imis, & Maketab, 2011).Even though the numerical example only implies only on fourselected parameters, the expected results can be derived using other parameters as long as the representing the range of parameters involves is adapted into the TrapMFs.5.0 CONCLUSION In this paper, we have applied the Fuzzy Inference System (FIS) to evaluate the water assessment by using our own hydrological data obtained from the Klombong Industrial area.It is clearly seen that the proposed method are capable to evaluate the status of the water and the process are less complex and straightforward.Furthermore, this reduces the time required to analyze the hydrological data to determine the status of the water.In short, FIS have shown to be one of the effective and less complex tools to assess the quality of water in a river basin.In the future, the proposed method can be validated using Sensitivity Analysis implemented in this paper to solve the evaluation of data on the boundary of a range was resolve as well by using FIS.That is to say, FIS is a process of formulating a mapping from a given multiple input to a single output using fuzzy logic.The process of fuzzy inference involves three important concepts: membership function, logical operations and If -Then rules(Carbajal-Hernández, Sánchez-Fernández, Carrasco-Ochoa, & Martínez-Trinidad, 2012;Ocampo-Duque, Osorio, Piamba, Schuhmacher, & Domingo, 2013)

Figure 2 :
Figure 2: Membership functions for Water Quality Index Figure 1 shows the representation of classi�ication of dissolved oxygen (DO), ammoniacal nitrogen (NH3N), pH and turbidity in TrapMFs. Figure 2 shows the representation of Water Quality Index in TrapMFs.The constructed TrapMFs was adapted from the DoE WQI that was still used in Malaysia.
If DO is C1 AND NH3N is C1 AND pH is C1 and Turb is C1 then WQI is Excellent.Rules 2: IfDO is C1 AND NH3N is C1 AND pH is C1 and Turb is C2 then WQI is Excellent.Rules 3: If DO is C1 AND NH3N is C2 AND pH is C1 and Turb is C1 then WQI is Excellent.Rules 4:If DO is C1 AND NH3N is C2 AND pH is C1 and Turb is C2 then WQI is Very Good.Rules 5: If DO is C1 AND NH3N is C3 AND pH is C1 and Turb is C1 then WQI is Excellent.Rules 6: If DO is C1 AND NH3N is C3 AND pH is C1 and Turb is C2 then WQI is Very Good.Rules 250: If DO is C5 AND NH3N is C5 AND pH is C5 and Turb is C5 then WQI is Poor.The output fuzzy rule then computed using the fuzzy operator AND, j, k and l are the different levels of concentration (Class 1, Class 2, Class 3, Class 4, Class 5 respectively) depends on each parameters.