Water Quality Prediction of Buriganga River Using Remote Sensing Techniques

Surface water pollution is one of the most malignant subjects in Bangladesh, particularly in the Dhaka district. Significantly, the amount of pollution in the Buriganga River is alarming to our environment and ecosystem, primarily due to untreated industrial waste and municipal discharge. Several studies show the degradation of the water quality of this river. A handful of them include remote sensing techniques. Any reservoir's water quality can only be determined by conducting numerous, time-consuming, and expensive in-situ measurements in addition to laboratory testing. However, we can cut down on both time and cost by utilizing remote sensing techniques. The purpose of this study is to use remote sensing methods to determine the Buriganga River's water quality. To achieve this goal, 53 water samples were taken from the Buriganga River in April 2023. Using Javascript codes, a Sentinel 2 level 2A image from the same date of the water sample collection was downloaded from Google Earth Engine (GEE). SNAP software was used to extract the surface reflectance values from the Sentinel 2 image. The following parameters were measured in the water samples: pH, COD, DO, turbidity, electrical conductivity (EC), total suspended solids (TSS), and total dissolved solids (TDS). Mostly, there was positive correlation between the water quality parameters and the reflectance values of the Sentinel 2 satellite image bands (B2-B12). The linearity of each parameter was examined further by testing it against the reflective bands. A statistically significant overall effect was found between them using multiple linear regression analysis. The R-square values for each of the individual parameter tests were as follows: pH (0.258), DO (0.013), COD (0.218), EC (0.551), TDS (0.549), Turbidity (0.508), TSS (0.131), and pH (0.258). The R-square value indicates that between 1% and 54.9% of the variation in the dependent variables (EC, TDS, TSS, turbidity, pH, DO, and COD) can be explained by the predictor variables (B2-B12). Aside from EC, TDS, and turbidity, the R-square values of the other parameters were too small for the satellite image to predict. Thus, additional investigation was not done on these parameters. The empirical models that can be used in the future to estimate a specific level of pollution in a river without always needing in-situ testing were constructed using multiple linear regression. The models were then utilized in conjunction with the raster calculation function of ArcMap software to produce maps of the river's anticipated water quality parameters. Once more, ArcMap is used to generate interpolation maps for the whole Buriganga River using the water quality parameter values derived from field data and laboratory tests. The degree to which remote sensing can assess the pollution level in a body of water is discussed through a comparison of the two types of maps.