Coastal Quality Along South West Coast of India Using Regression... ASSESSMENT OF COASTAL WATER QUALITY ALONG SOUTH WEST COAST OF INDIA USING MULTILE REGRESSION ANALYSIS ON SATELLITE DATA

The coastal waters being the ultimate receiver of all the wastes, shows a declining trend in its quality. It is of immense importance to know the extent of pollution for its monitoring and management. Measurement of dissolved oxygen (DO), biological oxygen demand (BOD), pH and fecal coliform (FC) are vital in water quality monitoring and assessment studies. Usually these parameters are determined by analysing water samples collected from various locations. Since this is tedious and expensive, it is limited to small scales. In this paper, an effort has been made to quickly assess the quality of coastal waters of Kerala directly from the satellite imagery by estimating National Sanitation Federation Water Quality Index (NSFWQI) along with DO, BOD, pH and FC. Multiple linear regression is used to develop statistically significant models using Sea Surface Temperature (SST) and Remote Sensing Reflectance (R rs ) from Moderate Resolution Imaging Spectroradiometer (MODIS) and in-situ data available on DO, BOD, pH and FC. The models when validated showed good correlation between in situ values and predicted values with r values ranging from 0.73 (p=0.001) for DO to 0.89 for NSFWQI (p=0.018).Spatial maps are generated showing the distribution of these parameters along the coast. The parameters in the study are checked to see if they are in compliance with the standards. The study gives models to estimate the daily distribution of these parameters along the coast using MODIS data. Thus, appropriate control measures could be adopted to limit the effect on susceptible rural population.


Introduction
The activities on land are the major contributors of pollutants in the oceans. Over 80 per cent of marine pollution, comes from oil spill, fertiliser, sewage and garbage disposal, toxic chemicals, and ballast water from ships (Vikas and Dwarkish, 2015). Marine creatures and plants are affected by toxic substances from dumped materials (Caroline, 1996). The consumption of affected fishes can cause health issues to human beings. Moreover, fishes being one of the crucial parts of aquatic food chains show a declining trend in its production (Islam and Tanaka, 2004). This affects the livelihoods of the rural fishing community.
A huge amount of effluents are being disposed daily into the rivers of the State from the cities, towns and industries. These pollutants finally reach the coastal waters through the estuaries (Chattopadhyay and Franke, 2006). The conventional method of water quality monitoring involves the collection of samples from various locations and its analysis. Though this method may give accurate results, it can only be employed for small scales as it is expensive and timeconsuming. A solution for the above problem is the usage of satellite remote sensing. Sometimes the temporal and spatial resolution of sensors might be inadequate. So, a combination of both these methods might be effective in tackling the problem. This study aims to estimate DO, BOD, FC, pH and NSFWQI using MODIS sensor data and thereby quickly ascertaining the water quality along the coastline of Kerala.

Study Area
Kerala is situated along the south western coast of India, between 8°17'30" and 12°47'40" north latitudes and 74°27'47" E and 77°37'12" E longitudes. Kerala with an area of 38,863 km² has got a coast 580 km in length, while its width varies from 35 to 120 km. The State is thickly    (1)  Table 1 shows the equations for determining the sub-index of the water quality parameters used for surface waters. Table 2 suggests the description for water quality based on NSFWQI. validation of the predictive model. Figure 3 shows the sampling locations used for formulation of algorithm.

Results and Discussion
Values  Table 3.

Development and Validation of Algorithm:
In the case of FC and NSFWQI, the algorithms

NSFWQI
The values of NSFWQI used for formulation of algorithm are given in Table 3.
NSFWQI=74.247-0.503*SST+0.674* R rs The scatter plot of actual and predicted values of FC with a coefficient of correlation of 0.78 and coefficient of determination of 0.605 is given in Figure 6.

Dissolved Oxygen:
The equation (4) The above equation shows that the variation pH increases with the reflectance in band 488 and it decreases with reflectance in band 678. This is in agreement with the finding (Sheela et al., 2013) that pH is related to reflectance of both green and red bands.  In the algorithm, BOD value increases with sea surface temperature. Figure 9 shows the plot of actual values of BOD against predicted ones.    implies improved water quality moving away from the shore.

Conclusion
The present study gives models to quickly estimate the water quality along the entire coast of Kerala using multiple linear regression by estimating the water quality index. Significant correlations were observed between optically inactive parameters (DO, BOD, pH and FC) dealt in the study and remote sensing reflectance. The dissimilar response of these parameters to statistical analysis for the development of prediction algorithm lead to unique equations.
These algorithms are validated for their useful future application.
The application of these algorithms to develop spatial map showing the distribution of NSFWQI on 03-01-2016 suggests that the overall water quality falls in the category of medium to good. The spatial maps of DO, BOD, pH and FC show anthropogenic impact on their distribution. DO and FC values falls above the prescribed CPCB standards. The rise in organic matter and faecal coliform and reduction in dissolved oxygen can be observed near the coast.
Huge coastal pollution affects the marine organisms namely fishes on which the rural fishing community depend on for their daily living (Bhuyan and Islam, 2016). The models developed may help in effective daily monitoring and prediction of marine water quality of Kerala, thus helping in resolving the problem in rural coastal communities which are the most vulnerable group of coastal pollution.