# Impact of Population Growth on the Water Quality of Natural Water Bodies

^{*}

## Abstract

**:**

## 1. Introduction

_{3}

^{−}, DO, chemical oxygen demand (COD), BOD, and the population in five watersheds. The results of the first step were used to find the most correlated parameters with the population. The classification model was developed as the second step. TC, DO, and BOD were used to qualitatively define the population ranges using the BN classification model. Upon developing the BN, the probability distribution table of the population node in the BN were analyzed to quantitatively derive the sustainable population ranges to maintain a certain level of water quality.

## 2. Materials and Method

#### 2.1. Study Area

^{2}and is the third largest watershed in Sri Lanka. Ranking as the fourth longest river in the country, it stretches from the Sri Pada mountain range to Colombo. Colombo is one of the most highly urbanized areas in the Asian region [20]. The Kelani River supplies approximately 80% of the water used in the Colombo district, and it is a primary source of drinking water for the Colombo district. Some municipalities of the Colombo district are in the L1 and L2 regions as shown in Figure 1. However, it is the most polluted river in Sri Lanka due to the rapid growth of industries located in close vicinity to it [14,21]. The annual rainfall distribution in these regions varies from 2001 to 3000 (mm) [13]. Flood encroachment is another disaster which frequently occurs in the Kelani River. A major contributory factor to the flooding has been the rapid illegal construction (e.g., illegal construction of buildings, filling of marshlands for development) taking place in the lower reach area of the Kelani River (i.e., Colombo, Gampaha and Kegalle). This has increased the sediment loadings, organic and inorganic loadings in the river and resulted in the frequent floods [22]. We used the data from five sampling sites located in five different watersheds: Raggahawathta Ela (Biyagama), Maha Ela (PallewelaOya), PusweliOya, WakOya, and PugodaOya. The points L1–L5 in Figure 1 respectively denote the above-mentioned sampling sites.

_{5}, TC and NO

_{3}

^{−}were used to analyze the correlation between water quality and population, representing a strong dataset [23]. There were some missing values and data errors due to both human and technical errors in the analysis, the recording of results and the failure to collect samples. Thus, we have 564 monthly data records of water quality in total.

#### 2.2. Analysis of Correlation

_{b}is the actual water quality parameter value for each sample. The number of records pertaining to each of the five actual water quality parameters in each sampling site L1, L2, L3, L4 and L5 are 116, 111, 117, 106 and 114, respectively. C

_{0}is the value of water quality standard. Here we used five water quality parameters, namely DO, COD, BOD

_{5}, TC and NO

_{3}

^{−}and water quality standards values for each parameter for the drinking category: 6 mg/L, 15 mg/L, 3 mg/L, 5000 MPN/100 mL and 5 mg/L, respectively. The m is the number of monitoring parameters. The urban and regional design, planning and disaster management mostly considers the urbanization levels when they want to make decisions by comparing the different regions such urban, suburban and rural. When categorized, with the most similar regions in one group according to density, it is easier to understand the impact of urbanization, rather than comparing the effect of the individual population in each region [1]. One of the crucial factors in defining the urbanization level is the population size and the density [25]. Therefore, considering the population distribution shown in Table 1, the five watersheds in Kelani River were classified into three categories as urbanized level one (UL1), which has a higher population, urbanized level two (UL2) which has intermediate population, and urbanized level three (UL3) which has lower population. Figure 1 shows the five watersheds according the above-mentioned three urbanization categories. Next, we compared the PI of each category of urbanization in every year. Then the correlation coefficient of each water quality parameter with population was calculated separately to identify the greatest effect.

#### 2.3. The Development of the Classification Model

_{0}on the given observed data x

_{1}, x

_{2}, …, x

_{n}in Equation (2).

_{0}is a variable representing the unobserved class CV and x

_{1}, x

_{2}, ..., x

_{n}are the set of variables of TC, BOD, DO and POP. The proportion ∝ holds because we assume the inputs are given. The BN implicitly encodes joint distributions and the probability of n attributes of x

_{i}can be decomposed as a product of the joint probability distribution as shown in Equation (3).

_{1}, ..., x

_{n}given x

_{0}in Equation (3). The Bayesian network with TAN learning algorithm shows some dependencies between variables other than the class variable.

_{1}, r

_{2}, ..., r

_{y}, ..., r

_{Q}}). D is the monthly data-set of water quality parameters containing DO, BOD and TC from the five sampling sites from 2003 to 2013 excepting 2007. The r

_{y}is the yth record of the data-set. Q is 564 the total number of records in the data-set (D). Equation (5) explains the MDL score algorithm. Y is the number of variables; LL is the log-likelihood and q is the number of records in data set (D). The value of LL is negative and the best structure should have the minimum score.

#### 2.4. The Comparison of Classification Models

## 3. Results and Discussion

#### 3.1. Analysis of Water Quality

^{2}) of each watershed in 2003 were 3916, 3399, 913, 1791 and 2642, corresponding to L1–L5, respectively, while in 2013 they were 4385, 4150, 1133, 2137 and 3099, respectively. The average values of the above population densities were 4151, 3774, 1023, 1964 and 2871 corresponding to the watersheds L1–L5, considering that L1 and L2 are categorized as UL1, L4 and L5 as UL2, and L3 as UL3, as defined in Section 2.2. These watersheds cover 2.65%, 2.63%, 4.87%, 4.02% and 2.23% of the entire watershed area of the Kelani River, respectively [13]. The research conducted on the Shanghai estuary of the Yangtze River in China used a similar comparison to identify the impact of the population on water quality [1]. They categorized the watershed area as urban, suburban and rural as per the population density of each.

^{−}(Table 3), can be used as a model in development. However, only three parameters (TC, BOD and DO) were selected from the above, given that their correlation coefficient was greater than 0.5. The result of water quality in three urban areas in Nepal, India and Bangladesh also showed positive correlations of BOD and TC with population and a negative correlation of DO with population [4]. Further, less correlation between population density and NO

_{3}

^{−}has been shown in other research conducted in Sierra Nevada, California by Dylan S. Ahearn et al. [29]. The research conducted on the Jinshui River Basin of the South Qinling Mountains, China, predicted the most correlated water quality parameters, which have strong correlation with population, by defining the linear equations [10]. Comparison of the results for both rivers clearly illustrates the relationship between population and the water quality of the river basin.

#### 3.2. The Development of the Classification Model

#### 3.3. Quantitative Population Range

#### 3.4. Proposed Method

## 4. Conclusions

## Acknowledgments

## Author Contributions

## Conflicts of Interest

## References

- Wang, J.; Da, L.; Song, K.; Li, B.-L. Temporal variations of surface water quality in urban, suburban and rural areas during rapid urbanization in Shanghai, China. Environ. Pollut.
**2008**, 152, 387–393. [Google Scholar] [CrossRef] [PubMed] - Sabry, H. Thousands of Lives at Risk Kelani River Contamination. Sunday Lead. 2015. Available online: http://www.thesundayleader.lk/2015/09/27/thousands-of-lives-at-risk-kelani-river-contamination/ (accessed on 10 July 2017).
- International Decade for Action ‘Water for Life’ 2005–2015. 8 September 2015. Available online: http://www.un.org/waterforlifedecade/ (accessed on 6 April 2017).
- Karn, S.K.; Harada, H. Surface Water Pollution in Three Urban Territories of Nepal, India, and Bangladesh. Environ. Manag.
**2001**, 28, 483–496. [Google Scholar] - Qin, H.; Su, Q.; Khu, S.-T.; Tang, N. Water quality changes during rapid urbanization in the Shenzhen River Catchment: An integrated view of socio-economic and infrastructure development. Sustainability
**2014**, 6, 7433–7451. [Google Scholar] [CrossRef] - Groppo, J.D.; de Moraes, J.M.; Beduschi, C.E.; Genovez, A.M.; Martinelli, L.A. Trend analysis of water quality in some rivers with different degrees of development within the São Paulo State, Brazil. River Res. Appl.
**2008**, 24, 1056–1067. [Google Scholar] [CrossRef] - Peters, N.E.; Meybeck, M.; Deborah, V.C. Effects of Human Activities on Water Quality. In Encyclopedia of Hydrological Sciences; Wiley Online Library, Wiley & Sons, Ltd.: New York, NY, USA, 2016. [Google Scholar]
- Athukorala, S.W.; Weerasinghe, L.S.; Jayasooria, M.; Rajapakshe, D.; Fernando, L.; Raffeeze, M.; Miguntanna, N.P. Ananlysis of Water Quality Variation in Kelani River, Sri Lanka Using Principal Componet Analysis. In Proceedings of the SAITM Research Symposium on Engineering Advancements, Malabe, Sri Lanka, 27 April 2013; pp. 129–135. [Google Scholar]
- Pompeu, P.S.; Alves, C.B.M. The Effects of Urbanization on Biodiversity and Water Quality in the Rio das Velhas Basin, Brazil. Am. Fish. Soc.
**2005**, 47, 11–22. [Google Scholar] - Bu, H.; Liu, W.; Song, X.; Zhang, Q. Quantitative impacts of population on river water quality in the Jinshui River basin of the South Qinling Mts., China. Environ. Earth Sci.
**2016**, 75, 292. [Google Scholar] [CrossRef] - Peters, N.E. Effects of urbanization on stream water quality in the city of Atlanta, Georgia, USA. Hydrol. Process.
**2009**, 23, 2860–2878. [Google Scholar] [CrossRef] - Zhao, H.; Duan, X.; Stewart, B.; You, B.; Jiang, X. Spatial correlations between urbanization and river water pollution in the heavily polluted area of Taihu Lake Basin, China. J. Geogr. Sci.
**2013**, 23, 735–752. [Google Scholar] [CrossRef] - Edussuriya, C.H.; Pathirag, S. Natural Resource Management and Monitoring Unit Published Final Report of Medium to Long-Term Multi-Stakeholder Strategy and Action Plan for Management and Conservation of the Kelanni River Basin; Central Environment Authority: Battaramulla, Sri Lanka, 2016. [Google Scholar]
- Ileperuma, O.A. Environmental pollution in Sri Lanka: A review. J. Natl. Sci. Found. Sri Lanka
**2010**, 28, 301–325. [Google Scholar] [CrossRef] - Wijesinghe, T.W.A.W. Status of Water Quality of Kelani River; Central Environmental Authority: Battaramulla, Sri Lanka, 2010. [Google Scholar]
- Edussuriya, C.H.; Pathirag, S. Preliminary Perception Survey on Water Safety: Kelani River Basin; Central Environment Authority and IUCN Sri Lanka Office: Colombo, Sri Lanka, 2016. [Google Scholar]
- Day Sharon, K.; Geoff, T.; Eileen, P.; Jone, M. Effects of Population Growth on Water Quality in a Mountain Watershed; Colorado Convention Center: Exhibit Hall, Hong Kong, 2002. [Google Scholar]
- Abistado, K.G.; Arellano, C.N.; Maravillas, E.A. Weather forecasting using artificial neural network and Bayesian network. ResearchGate
**2014**. Available online: https://www.researchgate.net/publication/288746408_Weather_forecasting_using_artificial_neural_network_and_Bayesian_network (accessed on 6 July 2017). - Rosly, R.; Makhtar, M.; Awang, M.K.; Rahman, M.N.A.; Deris, M.M. The study on the accuracy of classifiers for water quality application. Int. J. u-Serv. e-Serv. Sci. Technol.
**2015**, 8, 145–154. [Google Scholar] [CrossRef] - Kawamura, M.; Jayamanna, S.; Tsujiko, Y.; Sugiyama, A. Comparison of urbanization of four Asian cities using satellite data. Doboku Gakkai Ronbunshu
**1998**, 1998, 97–105. [Google Scholar] [CrossRef] - Liyanage, C.P.; Marasinghe, A.; Yamada, K. Comparison of Optimized Selection Methods of Sampling Sites Network for Water Quality Monitoring in a River. Int. J. Affect. Eng.
**2016**, 15, 195–204. [Google Scholar] [CrossRef] - Fowze, J.S.M.; Liyanage, P.P. Flood Hazard Mapping in Lower Reach of Kelani River, Sri Lanka (PDF Download Available). ResearchGate
**2008**. Available online: https://www.researchgate.net/publication/228772034_Flood_hazard_mapping_in_lower_reach_of_Kelani_River_Sri_Lanka (accessed on 6 July 2017). - Liyanage, C.P.; Yamada, K. Comparison of Water Quality Classification Models Using Artificial Neural Network and Bayesian Network. In Proceedings of the 2016 Joint 8th International Conference on Soft Computing and Intelligent Systems and 2016 17th International Symposium on Advanced Intelligent Systems, Hokkaido, Japan, 25–28 August 2016. [Google Scholar]
- Sri Lanka: Provinces & Cities—Population Statistics in Maps and Charts. City Popul. 2017. Available online: http://www.citypopulation.de/SriLanka-Cities.html (accessed on 6 July 2017).
- Dahly, D.L.; Adair, L.S. Quantifying the urban environment: A scale measure of urbanicity outperforms the urban-rural dichotomy. Soc. Sci. Med.
**2007**, 64, 1407–1419. [Google Scholar] [CrossRef] [PubMed] - Witten, I.H.; Frank, E.; Hall, M.A. Data Mining: Practical Machine Learning Tools and Techniques, 3rd ed.; Morgan Kaufmann: Burlington, MA, USA, 2011. [Google Scholar]
- Neapolitan, R.E. Learning Bayesian Networks; Pearson Prentice Hall: Upper Saddle River, NJ, USA, 2004. [Google Scholar]
- Statistical Correlation—Strength of Relationship between Variables. Available online: https://explorable.com/statistical-correlation (accessed on 6 July 2017).
- Ahearn, D.S.; Sheibley, R.W.; Dahlgren, R.A.; Anderson, M.; Johnson, J.; Tate, K.W. Land use and land cover influence on water quality in the last free-flowing river draining the western Sierra Nevada, California. J. Hydrol.
**2005**, 313, 234–247. [Google Scholar] [CrossRef] - El-Sheekh, M.M. Impact of Water Quality on Ecosystems of the Nile River. In The Nile River; Springer: Berlin, Germany, 2016; pp. 357–385. [Google Scholar]
- Altman, I.; Wohlwill, J.F. Behavior and the Natural Environment; Plenum Publishing Corporation: New York, NY, USA, 1983. [Google Scholar]

**Figure 1.**The five watersheds in the Kelani River with their water sampling points and defined three levels of urbanization.

**Figure 2.**Structures of the Bayesian network based on learning algorithms. (

**a**) Structure of BN based on the K2 learning algorithm; (

**b**) Structure of Bayesian Network (BN) based on Tree Augmented Naive Bayes (TAN) learning algorithm.

**Figure 4.**Relation between the integrated pollution index and the level of urbanization considering the increment of population in 2003, 2008 and 2013.

Watershed | Area (km^{2}) | Population 2003 | Population Density 2003 | Population 2013 | Population Density 2003 |
---|---|---|---|---|---|

L1 | 61.74 | 241,752 | 3916 | 270,752 | 4385 |

L2 | 61.46 | 208,910 | 3399 | 255,038 | 4150 |

L3 | 113.67 | 103,746 | 913 | 128,794 | 1133 |

L4 | 93.67 | 167,756 | 1791 | 200,138 | 2137 |

L5 | 52.14 | 137,752 | 2642 | 161,601 | 3099 |

Parameter | Drinking Water with Simple Treatment | Bathing Water | Fish and Aquatic Life |
---|---|---|---|

Class-A | Class-B | Class-C | |

TC as MPN/100 mL | <5000 | <5000 | <20,000 |

DO mg/L | >6 | >5 | >3 |

BOD_{5} mg/L | <3 | <4 | <4 |

Water Quality Parameters | Correlation Coefficients (p < 0.01) |
---|---|

TC | 0.687 |

BOD | 0.745 |

COD | 0.29 |

DO | −0.699 |

NO_{3}^{−} | 0.400 |

Method of Model | Accuracy out of 564 (%) | Micro Average of Recall | Micro Average of Precision | Computational Time (S) |
---|---|---|---|---|

BN with K2 leaning algorithm | 548 (97.16%) | 0.972 | 0.973 | 0.02 |

BN with TAN learning algorithm | 555 (98.43%) | 0.984 | 0.984 | 0.01 |

ANNs | 551 (97.69%) | 0.977 | 0.977 | 3.62 |

Forecasted Quality | |||||
---|---|---|---|---|---|

A | B | C | D | ||

Actual quality | A | 69 | 0 | 0 | |

B | 0 | 36 | 0 | 0 | |

C | 0 | 0 | 287 | 6 | |

D | 0 | 1 | 2 | 163 |

Water Quality Classification Class | Probability of Class Variable in Each Density Range Defined by BN Models (... × 100%) | ||
---|---|---|---|

0 to 2375 | 2375 to 2672 | Above 2672 | |

(POP-1) | (POP-2) | (POP-3) | |

A | 0.816 | 0.064 | 0.121 |

B | 0.867 | 0.013 | 0.12 |

C | 0.576 | 0.188 | 0.236 |

D | 0.104 | 0.021 | 0.875 |

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Liyanage, C.P.; Yamada, K.
Impact of Population Growth on the Water Quality of Natural Water Bodies. *Sustainability* **2017**, *9*, 1405.
https://doi.org/10.3390/su9081405

**AMA Style**

Liyanage CP, Yamada K.
Impact of Population Growth on the Water Quality of Natural Water Bodies. *Sustainability*. 2017; 9(8):1405.
https://doi.org/10.3390/su9081405

**Chicago/Turabian Style**

Liyanage, Chamara P., and Koichi Yamada.
2017. "Impact of Population Growth on the Water Quality of Natural Water Bodies" *Sustainability* 9, no. 8: 1405.
https://doi.org/10.3390/su9081405