Investigating the water quality index (WQI) using Landsat 8 satellite images and the application of univariate and multivariate models in Sefidroud River in northern Iran.
Yousef pourhabib 1 , ebrahim fataei 2 , Fatemeh Nasehi 3 , Behnam Khanizadeh 4 , Hossein Saadati 5
1 - Department of Environmental Sciences and Engineering, Ardabil Branch, Islamic Azad University, Ardabil, Iran
2 - Department of Environment, Ardabil Branch, Islamic Azad University, Ardabil, Iran
3 - Department of Environmental Sciences and Engineering, Ardabil Branch, Islamic Azad University, Ardabil, Iran
4 - Department of Chemistry, Sarab Branch, Islamic Azad University, Sarab, Iran
5 - Department of Environmental Sciences and Engineering, Ardabil Branch, Islamic Azad University, Ardabil, Iran
Keywords: : water quality index (WQI), Landsat 8, Sefidroud, multiple linear regression,
Abstract :
In this research, the water quality of Sefidroud River during the years 2013-2018 using Landsat 8 satellite images as well as 10 qualitative chemical and physical parameters including Ca2+, Na+,Mg2+, (Cl-,SO4-2, HCO3-, TDS, EC, TH and pH were studied in three hydrometric stations. Drinking water quality index (WQI) was calculated and its relationship with satellite bands and band ratios (28 parameters) was analyzed using univariate and multivariate regression models. The results of the univariate regression model showed that the WQI index with band 5 and the ratio of band B4/B3 had a linear and power correlation at a significance level of 1% with coefficient of determination (R2) of 0.55 and 0.51, respectively. The implementation of the stepwise linear multivariate regression model of WQI with all the studied bands and ratios showed that the three band 5 variables and band ratios B4/B3 and B6/B5 were correlated with WQI, with an R2 of about 0.80 at a 5% significance level. After preparing the spatial changes map of WQI using multivariate linear regression model, the results indicated that the water quality in the head branches of Sefidroud, that is, the Qezaluzen and Shahroud rivers, was lower compared to the lower areas and Manjil Dam Lake, although the water quality of the Qezeluzen and Shahroud rivers was lower. Shahrood was weak, but after entering Manjil Dam, it became a good water class. However, the WQI of the water coming out of the dam had gradually increased by passing through the agricultural lands, residential and industrial areas along the river until it reached the Caspian Sea, and it had weakened water class. In general, the results of the research showed that the use of Landsat 8 satellite images and multivariable regression model has a high power for water quality monitoring.
