Investigating the spatial variability of some important groundwater quality factors based on the geostatistical simulation (case study: Shiraz plain)

摘要

The management of groundwater plays a vital role in arid and semiarid regions. An assessment of groundwater suitability for irrigation is essential for a sustainable food production. In this study, efficiencies of three interpolation techniques such as inverse distance weighting, kriging and cokriging for simulation of groundwater quality indices for irrigation such as: pH, Mg2+, Ca2+, Na+, TH, electrical conductivity (EC), sodium absorption ratio (SAR), Cl- and SO42- were compared. The spatial structure results show that the variograms and cross-validation of the nine variables can be modeled with three methods, namely the inverse distance weighting, kriging and cokriging. The relevant data from 56 wells (with the depth between 30 and 60 m and diameter usually between 10 and 20 cm) in suburb of Shiraz were collected. After normalization of data, variograms were computed. Optimum variogram was selected based on least square value analysis. Then, by using cross-validation, mean error and root mean square error analysis, the interpolation model was selected. Results showed that for Mg2+, Ca2+, TH, EC, Cl- and SO42- cokriging had the lowest root mean square error, and for SAR and Na+ inverse distance weighting technique and for pH, kriging had better results comparing geostatistical method to simulate groundwater quality indices. For pH, TH, EC, and Mg2+ data; for SAR, Ca2+, Na+, and Cl- data; and for SO42- data, spherical, Gaussian model, and exponential were proved to be the best semivariogram models, respectively. Moreover, the results illustrated that cokriging method was the best due to its highest precision and lowest error. Finally, the geographic information system can fully display spatial patterns of quality factors in groundwater resources of the study area.