American Journal of Applied Mathematics and Statistics
ISSN (Print): 2328-7306 ISSN (Online): 2328-7292 Website: http://www.sciepub.com/journal/ajams Editor-in-chief: Mohamed Seddeek
Open Access
Journal Browser
Go
American Journal of Applied Mathematics and Statistics. 2018, 6(4), 158-169
DOI: 10.12691/ajams-6-4-6
Open AccessArticle

Multilevel Modeling of Surface Water Quality Data in Sri Lanka

Priyadarshani GDD1, and Sooriyarachchi MR1

1Department of Statistics, University of Colombo, Colombo, Sri Lanka

Pub. Date: August 19, 2018

Cite this paper:
Priyadarshani GDD and Sooriyarachchi MR. Multilevel Modeling of Surface Water Quality Data in Sri Lanka. American Journal of Applied Mathematics and Statistics. 2018; 6(4):158-169. doi: 10.12691/ajams-6-4-6

Abstract

Besides climate change impacts on water availability and hydrological risks, the consequences on water quality is just beginning to be studied. This research concerns the impacts of climate change on surface water quality through multilevel analysis. Multilevel modeling is a relatively new statistical technique in environmental science research, although its roots can be traced back to several other fields. The objective of this study was to evaluate the surface water quality, its spatial variation and its dependence on climatic parameters. The water quality data for seven parameters, namely Color, Turbidity, pH, Electrical Conductivity, Chloride, Total Alkalinity and Total Hardness collected from 2012 to 2014 from 68 locations around Sri Lanka was used for the analysis. These monthly water quality measurements had been made on two occasions nested within locations within districts and thus had a multilevel structure. Hence a multilevel regression model was adopted using the Bayesian Markov Chain Monte Carlo method. Since, neither of the 95% credible intervals for chemical composition (0.682, 4.945) and physical composition (0.203, 0.485) of water included the value zero, district level variances are significant. The chemical composition of water varies more with the districts compared to the physical composition of water. Several locations in Anuradhapura and Monaragala districts contributed to this significant difference in chemical composition and several locations in Ampara district presented a significant contribution to the difference in the physical composition as shown by the non-inclusion of the value zero in their individual 95% confidence bands. Further, it was observed that rain (P<0.01), temperature (P<0.01) and humidity (P<0.05) have an impact on both the chemical and physical composition of surface water. Source type (P<0.01) has an impact only on physical composition of water. The main conclusion of the study was that drinking water quality varied geographically and over time according to climatic conditions.

Keywords:
water quality climate change multilevel model regression Markov chain Monte Carlo

Creative CommonsThis work is licensed under a Creative Commons Attribution 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

References:

[1]  Whitehead, P. G., Wilby, R. L., Battarbee, R. W., Kernan, M., & Wade, A. J. (2015). A review of the potential impacts of climate change on surface water quality A review of the potential impacts of climate change on surface water quality, 6667 (November).
 
[2]  Gelman, A., & Park, D. K. (2009). Splitting a Predictor at the Upper Quarter or Third and the Lower Quarter or Third. The American Statistician, 63(1), 1-8.
 
[3]  Maas, C. J. M., & Hox, J. J. (2004). Robustness issues in multilevel regression analysis. Statistica Neerlandica, 58(2), 127-137.
 
[4]  Goldstein, H. (1999). Multilevel Statistical Models, London.
 
[5]  Browne, W. J. (2009). MCMC Estimation in MLwiN v2.10. Center for Multilevel Modeling, University of Bristol.
 
[6]  Maas, C. J. M., & Hox, J. J. (2004). The influence of violations of assumptions on multilevel parameter estimates and their standard errors. Computational Statistics & Data Analysis, 46(3), 427-440.
 
[7]  The SAGE Handbook of Multilevel Modeling. (2013). SAGE Publications. Retrieved from https://books.google.com/books?id=_Y1SAAAAQBAJ&pgis=1.
 
[8]  Browne, W. J. (2004). An illustration of the use of reparameterisation methods for improving MCMC efficiency in crossed random effect models. Multilevel Modelling Newsletter, 16, 13-25.
 
[9]  Bates, B.C., et al., (2008). Climate change and water. Geneva: IPCC Technical paper.
 
[10]  Ducharne, A. (2008). Importance of stream temperature to climate change impact on water quality, 797-810.
 
[11]  Hox, J. J., & Maas, C. J. M. (2005). Multilevel Analysis. Encyclopedia of Social Measurement (Vol. 2).
 
[12]  Zhang, J., & Boos, D. D. (1997). Mantel-Haenszel test statistics for correlated binary data. Biometrics, 1185-1198.