Application of spectrometer in prediction of total suspended solids in Singapore's coastal waters

• Main Author: Leong, Chee Kai
• Other Authors: Law Wing-Keung, Adrian
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Environmental engineering
• Online Access: http://hdl.handle.net/10356/78611
• Institution: Nanyang Technological University

Singapore has been involving in many reclamation projects in the past decades. During the reclamation process, the government will carry out their due diligence to monitor the water quality to ensure the coastal environment is not adversely impacted. There are several ways to conduct water quality monitoring such as field sampling, underwater monitoring station, constant surveillance by boats or remote sensing from satellite imaging. However, there are several limitations from each of this method such as time consuming, high cost or the accuracy. Thus, this study is started and aims to find out a more effective way to monitor water quality in the coastal area. Field samplings were conducted in February and March 2019 and water samples were collected from 10:00 to 14:00. The TSS concentrations were measured using the conventional filtration method in NTU Environment Laboratory. Light reflectance was taken using a handheld Blue-Wave Miniature Spectrometer (StellarNet Inc) during the field sampling at different integration time due to the different intensity of sunlight. The results from this study show that Blue-Wave Miniature Spectrometer has the capability to collect reflectance data in coastal area which possesses a considerably good curvilinear relationship to TSS concentration. Analysis suggests that the optimal correlation comes from the combination of green and red bands. The optimal logarithmic model is found to be y = 58.464*ln (Red + Green) - 124.35 and is validated by field sampling data with the highest coefficient of determination or R2 of 0.9347 and the lowest RMSE of 7.695 mg/L. The R2 of predicted and observed data is 0.8267 and RMSE of 4.6 mg/L which is relatively accurate. This study also finds that wave is one of the key factors that is not considered sufficiently in past literature. Hence, it would be useful to further consider the impact of waves on the reflectance and examine if there is some way to include this parameter in the model to the accuracy.