Mapping of Total Suspended Sediments Using Meteosat Second Generation

Abstract

High sediment load significantly reduces the optical transparency of coastal waters and causes establishment of sediments shoals that affects navigation. In addition, Total Suspended Sediments (TSS) affect adversely primary productivity and ecology by transporting chemicals and re-suspended pollutants. Therefore, estimation of the sediment concentration levels and dynamics of sediment distribution along coastal environment is essential for improvement of integrated resource management especially in maintaining safe navigation routes, harbour access and dredging operations. Whereas most studies have been successful in retrieval of the suspended particulate matter, the rapid fluctuations in TSS concentrations at coastal environment calls for higher temporal resolution which has not been widely addressed. The current study envisaged to achieve this using SEVIRI Instruments on board Meteosat Second Generation (MSG) satellite using channel one and two (visible bands). With the objective of retrieving TSS, an algorithm comprised of three models (Gordon, 1975; 1988 and Morel and Prieur (1977) was formulated in Matlab 7.6.0 software to compute water leaving radiance; normalized water leaving radiance; the exact water leaving radiance; subsurface irradiance reflectance; and bio-optical modeling (parameterization of inherent optical properties) and finally TSS concentration, respectively. Data preparation involved cloud and sun glint masking. The coded algorithm determined TSS concentration from Global MSG images, which were validated along the Kenyan Coast (Sabaki River Estuary). The results from Gordon (1975) model resulted in negative values because backscattering was larger than absorption. A scatter plot of the TSS concentration (lab) versus Gordon et al., (1988) model for small particles showed two clustered plots corresponding to neaptide and springtide; a regression analysis of 85%, correlation coefficient of 0.92 and absolute Root Mean Square Error (RMSE) of 3.44mg/m3. Though the large particles indicated the same correlation coefficient of 0.92, RMSE was 2.73mg/m3. Regression analysis of Morel and Prieur, (1977) model was 85%, correlation coefficient of 0.92 for both large and small particles. However, the absolute RMSE showed significant differences between small (2.81 mg/m3)and the large (12.08 mg/m3) particles. A spatial-temporal assessment of the TSS along the Kenyan Coast indicated three distinct trends in TSS concentration: marked low concentration for sites close to the land; an increasing sediment concentration seawards and an anomalously high concentration along the river mouth with a gradual decrease in seaward direction. A statistical test showed that there was no significant difference between the modelled and insitu TSS concentrations. These results attest to the novelty of the models used in the algorithm for monitoring TSS concentration along the coastal environment at near-real time-resolution of 15 minutes. Consequently, it was concluded that it is possible to monitor TSS concentration using SEVIRI instruments on board Meteosat Second Generation.