Introduction
This chapter focuses on providing an overall introduction to this study about modeling surface water quality in Sri Lanka within a multilevel framework.
Water accumulated or flowing on the surface of the earth is considered as surface water. The water on the surface can originate from rains or from groundwater. In order to be aware of surface water, it is necessary to understand the magnitude, quality, and status with respect to spatial and temporal occurrences. Therefore, research on surface water resources would require searching for knowledge or carrying out systematic investigations to establish facts about surface water.
1.1 Background
The multilevel or hierarchical structure has been considered for a number of years in social science, biological science, education and medicine when modeling data. However, its application in the environmental science is relatively recent. (Rackhow et.al, 2009). If the units are grouped at different levels, then the units have hierarchical structure having different characteristics in different levels. Simply, the children from the same family tend to be more similar in their physical and mental characteristics than children chosen randomly from the population (Goldstien, 1999). Furthermore, as an example, water quality measurements are taken from individual water bodies, water bodies are located in different climate zones. Therefore, it can be said that water bodies are grouped within each climate zone. Hence level of the water quality would depend on the attributes of different climate zones as well as other factors. Rather than either taking all of the data together or considering these separately, taking hierarchical structure is more effective for data modeling. Analyzing hierarchical or clustered structured data using traditional methods have many problems. However, multilevel analysis allows group wise characteristics of such data to be included in models accounting for individual behavior.
1.2 Motivation of the study
The world is facing a serious problem of natural resource scarcity, especially that of water with rapid growth of population and economic development. However, water is an essential factor for life on earth. All known forms of life need water for their existence. The global availability of fresh water is estimated to drop to 5,100 m3 per capita per year by 2025. This quantity would be enough to meet individual human needs if it were spread equally among the world’s population (Shiklomanov, 1999). Pure water is considered as water, which have low suspended or dissolved solids and obnoxious gases as well low in biological life (Gorde&Janhav, 2013).The quality of water usually described according to its physical, chemical & biological characteristics. But most of the fresh water bodies all over the world are getting polluted, thus portability of water is decreasing regularly. However, chemically and physically total pure water does not exist for an appreciable length of time in nature.
Water pollution occurs when a water body is adversely affected due to the addition of large amounts of waste materials. There are two types of pollutants existing such as point source and nonpoint source. When harmful materials are released directly into a water body, water pollution occurs as a point source. A nonpoint source carries pollutants to the water body indirectly through environmental changes. As an instance, nonpoint source of water pollution happens when fertilizer used in cultivation is carried into a water body by rain. Although point sources of pollution can be monitored and regulated by using some technologies, nonpoint sources are much more difficult to investigate and control.
Climate change influences are a key area of concern, especially when water is the underlying subject. Uncertainty about the potential climate change is the factor which impact to the water quality. The changes to precipitation and its pattern, changes to soil moisture due to temperature variations and changes to quantities evaporated from irrigated lands and irrigation reservoirs have been reasonably recognized as climate changes. Projected changes in rainfall and air temperature could affect river flows. Then the mobility and dilution of contaminants could happen. Also high temperatures will affect chemical reaction kinetics (Whitehead, 2009).
Due to use of contaminated water, human population suffers from water borne diseases. Diseases such as hepatitis and dysentery may be caused by contaminated water. Any change in water quality due to natural reasons or any other reasons, will affect human life as well as other lives. So it is essential to make sure that the chemical and physical ingredients are within the prescribed limits in drinking water supplies. This is the motivation to undertake this study because it will be of immense value in the light of the need for water management planning for the immediate future.
1.3 Research Objectives
The major objective of this study is to develop multivariate multilevel models for physical and chemical water quality separately to assess the potential factors in order to improve the drinking water production in Sri Lanka.
Since there are several missing values in the dataset, a great loss of information might be happened from ignoring the records with missing values. Therefore, in addition to the above mentioned aim this study also seeks to carry out a missing value imputation within the multilevel framework.
1.4 Data for the study
The main data set was obtained from the National Water Supply and Drainage Board, Galle Road, Ratmalana. It contains monthly surface water quality details of 68 island wide water intake locations during the period 2012-2014. Climate data was obtained from Meteorological Department, Colombo 07. It consists of monthly mean rainfall (mm), mean temperature (oC) and mean humidity(%) records for all the districts from 2012-2014.
After combining all the data, the final dataset consists of 13 variables. There are seven response variables of interest named Colour, Turbidity, pH, Electrical Conductivity, Chloride, Total Alkalinity and Total Hardness. These measurements were taken on a monthly basis under standard laboratory conditions. Mainly these data are obtained to investigate the relationship between water quality and potential factors such as source type, monthly mean rainfall, mean temperature and mean humidity. In addition, "year" and "month" are considered as explanatory factor variables.
The data use in the study was gathered across three main levels. Level 1 units consist of 68 water intake locations. Since locations are grouped as districts, districts make the level 2 units. Further, districts are grouped across the province wise and level 3 units can be considered as provinces. There are a number of parameters which explain water quality collectively on different dimensions. So methods for statistical analysis of such data should be under the umbrella of multivariate multilevel statistical methods.