With the rapid development of marine economy industry, the activities for exploring and exploiting the marine resources are increasing, and there are more and more marine construction projects, which contribute to the growing trend of eutrophication and frequent occurrence of red tide. Thus, seawater quality has become the topic which the people generally cared about. The seawater quality evaluation could be considered as an analysis process which combined the evaluation indexes with certainty and evaluation factors with uncertainty and its changes. This paper built a model for the assessment of seawater environmental quality based on the multiobjective variable fuzzy set theory (VFEM). The Qingdao marine dumping site in China is taken as an evaluation example. Through the quantitative research of water-quality data from 2004 to 2008, the model is more reliable than other traditional methods, in which uncertainty and ambiguity of the seawater quality evaluation are considered, and trade the stable results as the final results of seawater quality evaluation, which effectively solved the impact of the fuzzy boundary of evaluation standard and monitoring error, is more suitable for evaluation of a multi-index, multilevel, and nonlinear marine environment system and has been proved to be an effective tool for seawater quality evaluation.
Since the 1950s, marine environmental quality has been studied by domestic and foreign scholars in detail, and many methods for seawater quality evaluation are available, including the single factor index method [
The comprehensive seawater quality level of an object According to Calculating the relative membership degree matrix: when The comprehensive relative membership degree vector of sample When When When When
In the conditions of fuzzy concept classification, using the principle of the maximum membership degree to identify the level of an object in assessment for seawater quality can easily produce an incorrect final result. The level-characteristic value proposed in the equation by Chen and Hu [
The limitation of the AHP model of putting a binary comparison of the element attributes into the comparison of the importance is analyzed, and a nonstructural decision-making fuzzy theory model was presented by Professor Chen [ An objective set Arrange the sum of According to the matrix By the relationships between different mood operators and fuzzy scales, calculate the relative membership degree of the objective to the importance of the fuzzy concepts and attain the nonnormalized weight vector
Corresponding relationship between the tone operator and fuzzy scale values.
Tone operator | Equally | Slightly | Somewhat | Rather | Obviously | Remarkably |
Fuzzy scale | 0.50 | 0.55 | 0.60 | 0.65 | 0.70 | 0.75 |
| ||||||
Tone operator | Very | Extra | Exceedingly | Extremely | Incomparably | |
Fuzzy scale | 0.80 | 0.85 | 0.90 | 0.95 | 1.00 |
Consider
For DO
In the above formula,
The weight determined by the nonstructural decision-making fuzzy theory model is an experience weight and can easily be influenced by anthropic factors. During the evaluation process, the effect of some indices may be overstated or reduced; the weight determined by the standard level method is a mathematical weight. The relative importance of some indices has not been considered. The two types of weight methods each have certain advantages and limitations.
Here referring to this paper [
The Qingdao marine dumping site was one of the first marine dumping sites for Category III dredged materials specified by the State Oceanic Administration and approved by the State Council since the implementation of the
From November 1986 to 2009, the dredged materials dumped at this site exceeded a volume of 8.00 × 107 m3. This sea area is close to the navigation channel, the aquaculture area, and the holiday resort. It is the ecologically sensitive area that is also important for economic development. Therefore, it is important to accurately and timely evaluate the current environmental conditions at the dumping site for preventing ocean dumping from damaging the ecological environment, marine resources, and the submarine landform.
To facilitate comparisons, this study utilized monitoring data (Table
Monitoring results regarding seawater quality in the Qingdao dumping site in 2003.
Sampling point | COD | OIL | DO | Inorganic nitrogen | PO4-P | Cu | Pb | Zn | Cd |
---|---|---|---|---|---|---|---|---|---|
Q1 | 0.655 | 0.051 | 7.678 | 108.500 | 9.050 | 3.183 | 1.553 | 41.425 | 0.176 |
Q2 | 0.705 | 0.049 | 7.620 | 121.000 | 6.350 | 4.393 | 1.940 | 73.767 | 0.208 |
Q3 | 0.680 | 0.027 | 7.612 | 131.000 | 8.100 | 4.768 | 4.133 | 38.775 | 0.378 |
Q4 | 0.735 | 0.034 | 7.567 | 102.500 | 8.150 | 4.023 | 1.228 | 29.725 | 0.130 |
Q5 | 0.700 | 0.034 | 7.502 | 101.500 | 6.550 | 5.982 | 2.493 | 57.383 | 0.226 |
Q6 | 0.750 | 0.029 | 7.553 | 88.000 | 6.550 | 3.475 | 2.298 | 49.400 | 0.286 |
Q7 | 0.935 | 0.100 | 7.635 | 128.500 | 6.300 | 3.265 | 2.867 | 45.925 | 0.193 |
Q8 | 0.835 | 0.049 | 7.635 | 122.000 | 7.900 | 4.393 | 1.522 | 33.958 | 0.124 |
Q9 | 0.600 | 0.062 | 7.519 | 97.500 | 5.950 | 3.143 | 2.317 | 39.675 | 0.201 |
Q10 | 0.585 | 0.018 | 7.572 | 92.000 | 5.900 | 4.957 | 1.253 | 27.350 | 0.143 |
Q11 | 0.640 | 0.024 | 7.594 | 97.000 | 6.850 | 3.407 | 2.218 | 35.150 | 0.154 |
Q12 | 0.670 | 0.028 | 7.676 | 84.500 | 6.100 | 4.517 | 2.192 | 41.200 | 0.188 |
Q13 | 0.655 | 0.021 | 7.517 | 83.500 | 4.850 | 6.967 | 2.670 | 32.792 | 0.248 |
Q14 | 0.725 | 0.023 | 7.635 | 97.000 | 6.250 | 3.660 | 2.678 | 32.542 | 0.187 |
The units for COD, oil and DO are mg/L; the units for inorganic nitrogen, PO4-P, Cu, Pb, Zn and Cd are
The data from 14 monitoring points (Table
In reference to the standard seawater quality value and the actual seawater quality conditions at the Qingdao dumping site, the attraction domain matrix, the range domain matrix, and the
Therefore, the respective attraction domain matrix, the range domain matrix, and the
If
Based on the consideration of expert opinions, the numerous studies available in the literature [
By normalizing the fuzzy measure values matrix, we obtained the weight of 9 indexes of seawater quality evaluation:
Using (
Comparison of the comprehensive seawater quality evaluation results.
Point |
Variable fuzzy evaluation method | Fuzzy comprehensive evaluation | BP neural Network | Fuzzy genetic neural Network | |||||
---|---|---|---|---|---|---|---|---|---|
|
|
|
|
Average value | Evaluation grade | ||||
Q1 | 1.464 | 1.593 | 1.309 | 1.471 | 1.46 | Between I and II, tending toward I | Grade II | Grade II | Grade II |
Q2 | 1.645 | 1.894 | 1.327 | 1.762 | 1.66 | Between I and II, tending toward II | Grade II | Grade III | Grade III |
Q3 | 1.597 | 1.761 | 1.504 | 1.694 | 1.64 | Between I and II, tending toward II | Grade II | Grade II | Grade II |
Q4 | 1.354 | 1.427 | 1.231 | 1.357 | 1.34 | Between I and II, tending toward I | Grade I | Grade II | Grade I |
Q5 | 1.614 | 1.823 | 1.465 | 1.707 | 1.65 | Between I and II, tending toward II | Grade II | Grade III | Grade III |
Q6 | 1.492 | 1.705 | 1.304 | 1.553 | 1.51 | Between I and II, tending toward II | Grade II | Grade II | Grade II |
Q7 | 1.636 | 1.802 | 1.491 | 1.716 | 1.66 | Between I and II, tending to II | Grade II | Grade II | Grade II |
Q8 | 1.451 | 1.539 | 1.370 | 1.497 | 1.46 | Between I and II, tending toward I | Grade I | Grade II | Grade II |
Q9 | 1.512 | 1.645 | 1.399 | 1.586 | 1.54 | Between I and II, tending toward II | Grade II | Grade II | Grade II |
Q10 | 1.273 | 1.352 | 1.124 | 1.229 | 1.24 | Grade I | Grade I | Grade I | Grade I |
Q11 | 1.402 | 1.507 | 1.309 | 1.510 | 1.43 | Between I and II, tending toward I | Grade I | Grade II | Grade II |
Q12 | 1.463 | 1.611 | 1.348 | 1.538 | 1.49 | Between I and II, tending toward I | Grade I | Grade II | Grade II |
Q13 | 1.476 | 1.552 | 1.453 | 1.602 | 1.52 | Between I and II, tending toward II | Grade I | Grade II | Grade II |
Q14 | 1.421 | 1.529 | 1.346 | 1.557 | 1.46 | Between I and II, tending toward I | Grade I | Grade II | Grade II |
As shown in Table
For sampling point 2, the evaluation result based on the variable fuzzy evaluation model is between I and II, tending toward II, while the result based on the BP network and fuzzy genetic neural network is Grade III, and the result of fuzzy comprehensive evaluation method is Grade II. According to analyzing the original data of point 2, the values of COD, DO, inorganic nitrogen, PO4-P, Cu and Cd are 0.705 mg/L, 7.620 mg/L, 121
Analyzing the causes of the differences between BP network, fuzzy comprehensive evaluation method, fuzzy genetic neural network evaluation method, and variable fuzzy evaluation model in Table
This paper combines the monitoring values of seawater quality indicators with the national standard to build a seawater quality evaluation model in variable fuzzy recognition model, to deal with greater subjectivity problems of water quality evaluation with limited data. To a certain extent, we will measure the ambiguity and uncertainty of water quality evaluation objectively and increase the credibility of the rank of a sample point [
The method of the seawater quality evaluation model based on variable vague set theory in this paper is able to combine linear model with nonlinear model through changes of the variable model parameters (
However, for variable fuzzy recognition model, the rational weight setting is still an important factor to determine the reliability of the evaluation results. Due to cross-iteration of the parameters of variable fuzzy recognition model and the variability of indicators weight vector, it is very important to reasonably set the indicator weight according to the nature of a real case and the importance of actual decision objective in practice.
We use weight-determination method of the comprehensive weight which combines the subjective nonstructural decision-making fuzzy weights with the objective standard level weights and provide a reference for weight setting. In the future, how to set index weight more reasonably in actual marine environment evaluation and how to determine the level of seawater quality according to the characteristic level values will be studied to improve the application of multitarget variable fuzzy recognition model for seawater quality evaluation. Each water quality evaluation method owns different emphases. Variable fuzzy recognition model can combine the linear features with the nonlinear features of the evaluation objective and provide a reference for the multi-objective decision solutions and can be promoted for the evaluation of other multi-index, multilevel, and nonlinear systems.
Using the monitoring data of the seawater quality at Qingdao dumping site (1985–2003), the comprehensive situation of Qingdao dumping site (1985–2003) is evaluated by variable fuzzy comprehensive evaluation model. The evaluation results are shown in Table
Results of the variable fuzzy comprehensive evaluation of seawater quality in the Qingdao dumping area (1985–2003).
Year | COD | OIL | DO | Inorganic nitrogen |
|
Cu | Pb | Zn | Cd | Variable fuzzy evaluation model | |
---|---|---|---|---|---|---|---|---|---|---|---|
Average | Evaluation grade | ||||||||||
1985 | 0.45 | 0.038 | 8.73 | 1.26 | 0.31 | 0.49 | 4.13 | 2.715 | 0.13 | 1.30 | I |
1991 | 0.89 | 0.015 | 7.86 | 25.9 | 6.82 | 0.36 | 5.7 | 9.8 | 0.17 | 1.33 | I |
1997 | 0.65 | 0.0745 | 7.76 | 77.3 | 14.96 | 4.74 | 1.28 | 42.5 | 0.07 | 1.59 | II |
2000 | 1.54 | 0.021 | 9.17 | 94.42 | 15.75 | 2.98 | 0.86 | 13.7 | 0.18 | 1.14 | I |
2002 | 0.48 | 0.024 | 7.99 | 65.7 | 6.5 | 2.04 | 1.49 | 11.16 | 0.28 | 1.11 | I |
2003 | 0.72 | 0.039 | 7.64 | 102 | 6.9 | 4.30 | 2.25 | 41.54 | 0.19 | 1.64 | II |
We build a seawater environmental quality assessment model based on variable fuzzy recognition model, in which uncertainty and ambiguity of the seawater quality evaluation are considered, and the monitoring values of seawater quality evaluation indicators and the standard value of seawater quality are combined. Through the application of this model for the Qingdao marine dumping site water quality evaluation and comparison in performance with other models, the model is proved to be an effective tool for seawater quality evaluation. The following conclusions can be drawn. Seawater environmental quality assessment model based on variable fuzzy recognition model considers the uncertainty and ambiguity involved in the seawater quality evaluation, combines monitoring values of seawater quality evaluation indicators and the standard value of seawater quality, and selects the right variable model of the different parameters according to the linear or nonlinear features of the evaluation objects. Therefore, the method is more flexible than other models, and the evaluation results are more stable. It can arrange the situation of water environment quality of various samples and clearly determine the water quality status that makes the evaluation results more credible; therefore, it is more suitable for evaluation of a multi-index, multi-level, and nonlinear marine environment system. Different indices in different seawater environments have different effects on the evaluation results of seawater quality. In this paper, weight-determination method of the comprehensive weight which combines the subjective nonstructural decision-making fuzzy weights with the objective standard level weights and provides a reference for weight setting. When the evaluation model is applied to other applications, it is necessary to set the index weight reasonably according to the specific conditions of seawater quality evaluation. In the future, how to determine the level of seawater quality according to the characteristic level values is an important part, which needs to be improved in the application of a multitarget variable fuzzy recognition model for seawater quality evaluation.