Loading...

Water Quality Monitoring Using Remote Sensing: A Case Study of Gorgan Bay

Mofidi Neyestani, Reza | 2022

181 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 55332 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Tajrishi, Massoud
  7. Abstract:
  8. Surface water quality management needs water contaminants monitoring and information about water quality parameters. Field measurement is a costly, time-consuming procedure that limits information about the whole surface of a water body. Remote sensing methods are valuable ways that prepare water quality information using different characteristics between reflectance detected from each parameter. This study aimed to investigate the water quality parameters of Gorgan Bay using field sampling and a feasibility study using Landsat-8 and Sentinel-2 images to estimate the spatial and temporal changes of optically active parameters (such as sea surface temperature (SST), salinity, electrical conductivity (EC), turbidity, Secchi depth (SD), chlorophyll-a (Chl-a), and water depth) and their use to optically inactive parameters such as total dissolved solids (TDS) and dissolved oxygen (DO). To achieve this goal, 114 points were sampled in September and March 2020 and June 2021 for the predictor model calibration and validation. In order to determine the sampling stations, the K-Means algorithm with 20 classes was used to cluster spectral reflectances. Then stations were chosen in different classes with the proper distribution. Also, the unit optical depth of the bay's water was measured using a light sensor. This study introduces Bayesian Linear Regression (BLR) as a method to estimate water quality parameters using Sentinel-2 and Landsat-8 data and analyses all of the mentioned water quality parameters temporally and spatially between 2020 and 2022. Two band series were used as the model's input to investigate the best input band series. Based on the results, Sentinel-2 is more suitable than Landsat-8 for water quality estimation, and Blue, Red, Green, and NIR bands are the best input for the estimation model. NRMSE for all of the optically active parameters in this research was lower than 0.3, and MAE for all optically active parameters except SST was lower than 1. NRMSE for DO was 0.178 and for TDS was 0.033. MAE for these two optically inactive water quality parameters was 0.320 and 0.813 mg/L, respectively. According to our exploration, this study is the first research that comprehensively investigates Gorgan Bay water quality using remote sensing technology.
  9. Keywords:
  10. Goragan Bay ; Remote Sensing ; Landsat-8 Satellite ; Sentinel Satellite ; Linear Bayesian Regression ; Water Quality ; Surface Water

 Digital Object List

 Bookmark