A thorough understanding of fluvial sediment transport is critical to addressing many environmental concerns such as exacerbated flooding, degradation of aquatic habitat, excess nutrients, and the economic challenges of restoring aquatic systems. Fluvial sediment samples are integral for addressing these environmental concerns but cannot be collected at every river and time of interest. Therefore, to gain a better understanding for rivers where direct measurements have not been made, extreme gradient boosting machine learning (ML) models were developed and trained to predict suspended sediment and bedload from sampling data collected in Minnesota, United States (U.S.), by the U.S. Geological Survey. Approximately 400 watershed (full upstream area), catchment (nearby landscape), near‐channel, channel, and streamflow features were retrieved or developed from multiple sources, reduced to approximately 30 uncorrelated features, and used in the final ML models. The results indicate suspended sediment and bedload ML models explain approximately 70% of the variance in the datasets. Important features used in the models were interpreted with Shapley additive explanation (SHAP) plots, which provided insight into sediment transport processes. The most important features in the models were developed to normalize streamflow by the 2‐year recurrence interval and quantify the rate of change in streamflow (slope), which helped account for sediment hysteresis. Generally, this study also showed a combination of mostly watershed and catchment geospatial features were important in ML models that predict sediment transport from physical samples. This study is a promising step forward in making fluvial sediment transport predictions using machine learning models trained by physically collected samples. The approach developed here can be used wherever similar datasets exists and will be useful for landscape and water management.
Measuring suspended sediment in fluvial systems is critical to understanding and managing water resources. Sampling suspended sediment has been the primary means of understanding fluvial suspended sediment. Specialized samplers, sampling methods, and laboratory methods developed by select U.S. Federal Agencies are more representative of river and stream conditions than commonly used grab sampling and total suspended solids (TSS) laboratory methods but are not widely used because they are expensive, time consuming, and not required as part of water quality standards in the United States. A new suspended-sediment sampling method called a depth-integrated grab (DIG) was developed by combining certain elements from both grab and depth-integrating sampling methods and suspended-sediment concentration (SSC) laboratory methods. The goal of the DIG method was to provide more accurate results than Grab-TSS while being easier and cheaper to sample than specialized samplers and methods. Approximately 50 paired comparison samples were collected at 9 sites in Minnesota from 2018 through 2019. Results showed no significant difference between the DIG and specialized sampling methods and a significant difference between both methods and the Grab-TSS method. The DIG-SSC provided an improved alternative to the Grab-TSS method, but additional research and testing is important to evaluate if this method is appropriate in different conditions than were observed in this study.
Accurate measurements of suspended sediment, a leading water-quality impairment in many Minnesota rivers, are important for managing and protecting water resources; however, water-quality standards for suspended sediment in Minnesota are based on grab field sampling and total suspended solids (TSS) laboratory analysis methods that have underrepresented concentrations of suspended sediment in rivers compared to U.S. Geological Survey equal-widthincrement or equal-discharge-increment (EWDI) field sampling and suspended sediment concentration (SSC) laboratory analysis methods. Because of this underrepresentation, the U.S. Geological Survey, in collaboration with the Minnesota Pollution Control Agency, collected concurrent grab and EWDI samples at eight sites to compare results obtained using different combinations of field sampling and laboratory analysis methods.Study results determined that grab field sampling and TSS laboratory analysis results were biased substantially low compared to EWDI sampling and SSC laboratory analysis results, respectively. Differences in both field sampling and laboratory analysis methods caused grab and TSS methods to be biased substantially low. The difference in laboratory analysis methods was slightly greater than field sampling methods.Sand-sized particles had a strong effect on the comparability of the field sampling and laboratory analysis methods. These results indicated that grab field sampling and TSS laboratory analysis methods fail to capture most of the sand being transported by the stream. The results indicate there is less of a difference among samples collected with grab field sampling and analyzed for TSS and concentration of fines in SSC. Even though differences are present, the presence of strong correlations between SSC and TSS concentrations provides the opportunity to develop site specific relations to address transport processes not captured by grab field sampling and TSS laboratory analysis methods.
Despite limitations in reproducing complex bedload sediment transport processes in rivers, formulas have been preferred over collection and analysis of field data due to the high cost and time-consuming nature of bedload discharge measurements. However, the performance of such formulas depends on the hydraulic and sedimentological conditions they attempt to describe. The availability of field measurements provides a unique opportunity to test bedload transport formulas to better guide formula selection. Hydraulic parameters and bedload discharge data from the Lower Minnesota River and two of its tributaries were used to evaluate nine bedload transport formulas using three different indices. The bedload data for the different sites were collected by the United States Geological Survey (USGS) from 2011 through 2014, with bed material varying from very coarse to medium sand. The formulas calculated higher bedload rates than were measured due to a combination of site-specific physical characteristics, including the presence of bed forms (dunes), and sampling uncertainties. Because of the lack of reproducibility of the tested formulas, five power functions, based on the relation between the specific unit power (independent hydraulic variable) and the USGS measured data (dependent variable), were derived as provisional equations to estimate the bedload discharge on the Lower Minnesota River and tributaries.
1 April 2022 in Earth Surface Processes and Landforms
Lake Pepin is the largest naturally formed lake on the Mississippi River and has complex management needs to satisfy economic, environmental, and cultural demands. Lake Pepin is filling in with sediment at a rapid rate compared to conditions before settlement by European immigrants and intense agricultural cultivation. Accordingly, the Minnesota Pollution Control Agency has developed aggressive plans to prioritize sediment sources, understand transport mechanisms, and implement large-scale strategies to reduce sedimentation in Lake Pepin.The Minnesota River is the primary sediment source to Lake Pepin, and reductions in sediment loading from the Minnesota River are needed to reduce sedimentation in Lake Pepin. Current loading estimates were calculated from grab sampling and total suspended solids laboratory methods that greatly underestimate the actual concentrations in the rivers when compared to U.S. Geological Survey width and depth integrated sampling and laboratory methods for determining suspended-sediment concentration (SSC). Therefore, the U.S. Geological Survey, with funding from the Environment and Natural Resources Trust Fund and in cooperation with the U.S. Army Corps of Engineers, Lower Minnesota River Watershed District, Minnesota Pollution Control Agency, and Minnesota Department of Natural Resources, collected SSCs and acoustic backscatter data from acoustic Doppler velocity meters over a 2-year period at nine sites. The purpose of the study was to improve understanding of sediment-transport processes and increase accuracy of estimating SSCs and suspended-sediment loads for the lower Minnesota River and the Mississippi River compared to traditional measures.The study results indicated that acoustic backscatter worked well in estimating SSCs at sites not regulated by locks, dams, and lakes. The results also confirmed previous studies that determined most of the suspended-sediment loading into the Mississippi River is from the Minnesota River and the largest sediment sink is Lake Pepin. Suspended-sediment loading from site to site and year to year was often variable when compared to streamflow, which has been traditionally used to estimate SSC. As a result, this study demonstrates the value in having high temporal and spatial resolution of continuous sediment monitoring from acoustic devices to help manage the sources of sediment into Lake Pepin.