Analysis of suspended sediment loads in streams and rivers using linear regression and Pearson correlation

Abstract

Sediment is the number one pollutant in the United States. Sediment in streams and rivers also carries along other pollutants such as nutrients (phosphorus and nitrates) and harmful bacteria. This study employs linear regression to predict the suspended sediment load, a dependent variable, as a function of the stream water discharge, an independent variable in seven U.S. Rivers and streams. The major objectives of the study are to investigate the effect of the sediment record length on the accuracy of the prediction, and to investigate the correlation of the suspended sediment load with nutrients and fecal coliform. The linear regression results showed that sediment sources/sinks and hydrologic variations that take place throughout the year play a vital role in the regression analysis of suspended sediment data. Five of the seven investigated rivers produced accurate predictions of the suspended sediment load. The sediment record length can affect the value of the correlation coefficient between the streamflow and suspended sediment rate. The percent deviation between the predicted and suspended sediment is less likely affected by the sediment record length and most likely affected by the hydrologic variations and sediment sources/sinks. Two of the investigated rivers show that significant hydrologic variations and sediment sources/sinks can increase the percent deviation, regardless of the sediment record length. This study also includes a Pearson correlation between the suspended sediment load, nutrients and fecal coliform. The Pearson correlation results showed no correlation between the suspended sediment loads and nutrients but high correlation was seen between the fecal coliform and the suspended sediment loads in the investigated rivers. The two investigated rivers for Pearson correlation are influenced by flow, which greatly determines the other parameters, even though the two rivers have different physical conditions.