Suspended Solids Mixing in Large River Confluences: A Remote Sensing Perspective

Abstract:

Mixing processes control spatial and temporal patterns of water quality in large rivers. Studies of mixing at river confluences have mainly been limited either to laboratory junctions or to field investigations at small confluences. Detailed studies of mixing downstream from confluences in large rivers are difficult due to the expense and effort required to conduct such studies. The objectives of this research are: 1) to develop an empirically based model, based on spectral reflectance information, for predicting total suspended solids concentrations (SSCs) at and downstream of confluences, and 2) to use the predicted values of SSCs to better understand the dynamics of mixing downstream from confluences. The study site for our research is the Mississippi River at and downstream of its confluence with the Missouri River. Data consist of SSCs measured by the USGS in the Mississippi River between 1988 and 2003 and atmospherically corrected imagery obtained by Landsat 5 and Landsat 7 satellites during the same period. Data on spectral reflectance for pixels with measured sediment concentrations are used in a RandomForest regression tree analysis to develop a predictive model relating spectral reflectance to SSCs. This model is used to estimate SSCs downstream of the confluence for different dates of imagery. Cross-channel patterns of SSCs illustrate the degree of transverse mixing at various distances downstream from the confluence and are useful for ascertaining how changes in channel characteristics influence rates of transverse mixing. Preliminary results indicate that the magnitude of differences in SSCs between the two rivers, the momentum flux ratio between the rivers, and changes in channel width are important factors controlling rates of transverse mixing. These results advance our understanding of mixing in large rivers and help inform water quality management practices for such rivers.