Evaluating the effect of sediment color on water radiance and suspended sediment concentration using digital imagery

Abstract

Satellite remote sensing has been extensively used for estimating the suspended sediment concentration (SSC) in coastal and inland waters. The application of this method to smaller waterbodies, however, encounters several limitations, such as the coarse temporal and spatial resolution of satellite images, cloud coverage intercession, and inaccessibility during emergency periods. Conventional digital cameras are used for monitoring the water quality parameters in inland waters, but their application for remote sensing of SSC in inland waters is still under development. The empirical models developed to estimate SSC from digital imagery are usually site-specific and do not consider the effect of the sediment properties such as grain size and shape, sediment color, and types of minerals. The performance of digital cameras for estimating SSC is assessed in this study, and the effect of sediment color on the accuracy of this method is investigated. A series of laboratory experiments were conducted using four different colors of sediment: light gray, dark gray, light brown, and dark brown. The results showed that the red waveband reflectance of digital imagery was more sensitive to the variation of SSC than that of blue and green waveband. The sediment color evidently affected the correlation between the measured SSC and waveband reflectance, even with the same particle size distribution. Sediment with darker color showed the lowest reflectance values, whereas lighter-color sediment has the highest reflectance values. The performance of the developed SSC-reflectance regression models for all sediment colors was assessed at two river sites, i.e., Village Creek, and the West Fork Trinity River. The correlation between measured and predicted SSC for Village Creek, that had the same sediment size and color as the light gray soil, was high (R2=0.98). The correlation was relatively low (R2=0.55) for the West Fork Trinity River and dark gray soil water samples which had a similar color and different sediment sizes. The difference in sediment sizes could explain why the SSC-reflectance model derived from the dark gray soil exhibited poor performance in predicting the SSC for this site. This study demonstrated that consideration of sediment color effect in developing a remote sensing algorithm from digital imagery would result in more accurate estimations of SSC in riverine environments.