Distribution of Total Dissolved Solids in Drinking Water by Means of Bayesian Kriging and Gaussian Spatial Predictive Process

Abstract

High level of total dissolved solids (TDS) in drinking water may cause problems in human body such as, stiffness of the joints, kidney stones, gallstones, and it may also cause hardening of arteries and sometime blockage of arteries in severe conditions. The TDS level of groundwater with spatial domain Punjab province, Pakistan is considered as response variable (random field) in present paper. Model-based ordinary kriging (OK), Bayesian ordinary kriging (BOK), and Gaussian spatial predictive process are used to model the spatial distribution of response variable. These methods ultimately estimate predictions maps and probabilities with specified threshold limit. The BOK performs better than OK and Gaussian spatial predictive process in terms of root mean square prediction error. This study reveals that the elevated TDS level crossing WHO permissible limit (1,000 mg/l) found in 24 % sample. Bahawalpur, Dera Ghazi Khan, Hafizabad, Khushab, Nankana Shaib, Okara, and Toba Tek Singh districts of Punjab province are observed of great concern.