TL;DR
This paper tackles the challenge of prioritizing dam construction sites using a Fuzzy TOPSIS method, specifically addressing the common issue of missing environmental data. By employing Monte Carlo simulations to generate plausible missing information, the study demonstrates how this approach can robustly rank potential sites, even with data gaps, offering a valuable tool for strategic environmental assessments.
What Problem Does This Paper Address?
Strategic Environmental Assessments (SEAs) are crucial for long-term planning, especially for large infrastructure projects like dam construction. In South Korea, the Long-Term Master Plan for Dam Construction (LPDC) relies on SEAs to identify suitable sites. However, a significant challenge in these assessments is the presence of missing environmental data. This data gap can lead to uncertainty and potentially flawed prioritization of potential dam sites, impacting the effectiveness of the entire planning process. Previous studies have attempted to address missing data, but this paper specifically investigates how the performance of a Fuzzy TOPSIS method is affected by different ways of generating missing information, particularly when multiple environmental criteria are involved.
Methods
The core of this study is the application of the Fuzzy TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) method. TOPSIS is a well-established multi-criteria decision-making (MCDM) technique that ranks alternatives based on their distance to an ideal positive and negative solution. The "Fuzzy" aspect allows for the incorporation of linguistic variables and uncertainty, which is particularly useful for qualitative environmental criteria.
To handle missing information, the researchers employed the Monte Carlo Simulation method. This involved generating random data for the missing values using both uniform and normal distributions. This was done for seven specific instances of missing data across four environmental categories: National Preservation (NP), Endangered Species (ES), Water Quality (WQ), and Toxic Environment (TE). The study then used these generated datasets to run the Fuzzy TOPSIS analysis and compare the resulting priority rankings.
Datasets
The study utilized data from ten potential dam sites in South Korea, located in three major river basins: Han River, Geum River, and Nakdong River. These sites were evaluated based on four socio-ecologic-environmental assessment categories:
- National Preservation (NP): Including natural, historic, and scenic preservation.
- Endangered Species (ES): Covering amphibians, reptiles, fish, birds, mammals, insects, and plants.
- Water Quality (WQ): Measured by parameters like total phosphorus (TP), total nitrogen (TN), chemical oxygen demand (COD), and biological oxygen demand (BOD).
- Toxic Environment (TE): Assessed by the number of abandoned mines.
The paper explicitly notes that there were seven sets of missing information within this dataset, specifically four in the NP category and three in the TE category.
Key Insights and Ideas
- Impact of Missing Data Complexity: The study highlights that the number of fuzzy sets associated with missing information significantly impacts the results. When only one fuzzy set is involved (as in the Toxic Environment category), the generated data has a less pronounced effect on the Fuzzy Positive-Ideal Solution (FPIS) and Fuzzy Negative-Ideal Solution (FNIS) estimations. However, when multiple fuzzy sets are involved (as in the National Preservation category), the impact on FPIS, FNIS, and the final priority ranking becomes much more substantial.
- Distribution Choice Matters (Sometimes): While both uniform and normal distributions were used for data generation, their impact varied. For categories with more complex missing data (like NP), the choice of distribution could lead to notable differences in rankings. However, for simpler cases (like TE), the distributions produced very similar results.
- Robustness in Certain Basins: The study found that in the Han River and Geum River basins, the priority rankings of the proposed dam sites remained relatively stable regardless of the missing information generation method. This suggests a degree of robustness in the prioritization for these specific regions.
- Nakdong River Basin Sensitivity: In contrast, the Nakdong River basin showed more sensitivity to the missing information generation method, with different rankings emerging depending on whether uniform or normal distributions were used. This indicates that careful consideration of data generation is particularly important for this basin.
Key Results
- Top-Ranked Sites: Across all river basins, the Bocheong stream 2 (GB2) site in the Geum River basin consistently emerged as the highest priority site. Other highly ranked sites included Dal stream 1 (HD1) in the Han River basin and the Gohyun stream (NG) site in the Nakdong River basin.
- GB2 and GB1 Dominance: The GB2 site was selected as the most suitable site over 80% of the time, while GB1 was selected less than 20% of the time, highlighting GB2's strong preference.
- Han River Stability: In the Han River basin, HD1 was consistently the most suitable site across all missing information generation scenarios.
- Nakdong River Nuances: In the Nakdong River basin, the NG site generally showed higher suitability than other sites, though the difference between NG and NH varied with the generation method. The NI site was consistently the least suitable.
- Fuzzy TOPSIS Applicability: The study concludes that Fuzzy TOPSIS is a viable method for evaluating priority rankings even in the presence of missing information, aligning with previous MCDM studies.
Limitations and Future Work
- Uncertainty in Actual Distribution: The authors acknowledge that the study's results are "still uncertain because the study applied Monte Carlo Simulation for missing information with uniform and normal distribution. Moreover, we do not know the actual distribution of absent information." This is a critical limitation, as the generated data is based on assumptions.
- Future Research Directions:
- Explore Diverse Missing Information Scenarios: Apply various types of absent information to better estimate the precise performance of Fuzzy TOPSIS.
- Investigate Theoretical Approaches: Develop theoretical frameworks for Fuzzy TOPSIS performance in absent information cases.
- Compare MCDM Approaches: Explore and compare different MCDM methods to find the most suitable one for absent information scenarios.
- Consider Other Distributions: Investigate the impact of other probability distributions for data generation.
Summary updated