Effects of spectral variability due to sediment and bottom characteristics on remote sensing for suspended sediment in shallow rivers.

Advances in remote sensing techniques for water environments have led to acquisition of abundant data on suspended sediment concentration (SSC). However, confounding factors, such as particle sizes, mineral properties, and bottom materials, have not been fully studied, despite their substantial interference with the detection of intrinsic signals of suspended sediments. Therefore, we investigated the spectral variability arising from the sediment and bottom using laboratory and field-scale experiments. In the laboratory experiment, we focused on measuring spectral characteristics of suspended sediment according to particle size and sediment type. The laboratory experiment was conducted under conditions of completely mixed sediment and non-bottom reflectance using a specially designed rotating horizontal cylinder. To investigate the effects of different channel bottoms under sediment-laden flow conditions, we performed sediment tracer tests in field-scale channels comprising sand and vegetated bottoms. Based on experimental datasets, we performed spectral analysis and multiple endmember spectral mixture analysis (MESMA) to quantify the effect of spectral variability of sediment and bottom on the relationship between hyperspectral data and SSC. The results showed that optimal spectral bands were precisely estimated under non-bottom reflectance conditions, and the effective wavelengths depended on the sediment type. The fine sediments had a higher backscattering intensity compared to the coarse sediments, and the reflectance difference according to the particle size difference increased as the SSC increased. However, in the field-scale experiment, the bottom reflectance substantially decreased the R2 in the relationship between the hyperspectral data and SSC. Nevertheless, MESMA can quantify the contribution of suspended sediment and bottom signals as fractional images. Moreover, the suspended sediment fraction had a clear exponential relationship with SSC in all cases. We conclude that MESMA-driven sediment fractions could be an important alternative for estimating SSC in shallow rivers, as it quantifies the contributions of each factor and then minimizes the bottom effect.

Surrogate prediction of the breakthrough curve of solute transport in rivers using its reach length dependence.

Techniques for predicting the contaminant cloud propagation along a stream are necessary for swift action against contaminant spill accidents in fluvial systems. Due to their low computational cost, one-dimensional solute transport models have conventionally been employed, in which the complex channel characteristics are considered using model parameters. However, the determination of such parameters relies predominantly on optimization techniques based on pre-measured tracer data, which are usually unavailable for unexpected accidents. The present paper suggests an alternative method for predicting a breakthrough curve (BTC) variation along an unmeasured stream reach where no flow information is provided. In this study, we investigated the relationship between directly-measured flow properties and BTC characteristics based on field tracer experiments. Using statistical features of the tracer BTCs, we devised a regressive prediction method for estimating the BTC features as a function of travel distance, and validated the method by comparison with simulations using both a one-dimensional advection-dispersion equation (ADE) and transient storage model (TSM), whose parameters were calibrated at upstream reaches. The proposed regressive predictions were relatively accurate than those from parameter-calibrated models, and this advantage was more apparent for long-distance predictions for the unmeasured river reach.

Hyperspectral retrievals of suspended sediment using cluster-based machine learning regression in shallow waters.

Remote sensing of suspended sediment in shallow waters is challenging because of the increased optical variability of the water, resulting from the influence of suspended matter in the water column and the heterogeneous bottom properties. To overcome this limitation, in this study, we developed a novel framework called cluster-based machine learning regression for optical variability (CMR-OV), using the Gaussian mixture model (GMM) clustering technique and a random forest regressor (RFR). We evaluated the model using an optically complex dataset from a field-scale experiment. This experiment was conducted with four sediment types injected into an experimental meandering channel divided into two reaches with submerged vegetation and a natural sand bottom. We obtained high-resolution hyperspectral images using unmanned aerial vehicles (UAVs) and measured the in situ suspended sediment concentration using laser diffraction sensors. Based on optical similarity, we used CMR-OV to divide the hyperspectral dataset into several clusters. Then, we built separate RFR models for each cluster using the corresponding spectral bands that were selected using recursive feature elimination (RFE). Thus, we found that the proposed CMR-OV yielded superior results compared to the conventional RFR model, decreasing the total error score by 10.81%. The optical spectral bands of each cluster were distinguished from each other, indicating that the datasets that were spectrally discriminated from clustering enhanced the performance of the estimator. By comparing the clustered spectral dataset and physical factors, we proved the bottom type was the most critical factor in separating the clusters, even though the variability in the sediment properties also induced substantial spectral changes. Our findings demonstrated that CMR-OV accurately reproduced the spatiotemporal distribution of suspended sediment under optically complex conditions by addressing the heterogeneity of bottom reflectance in shallow water.

Identification Framework of Contaminant Spill in Rivers Using Machine Learning with Breakthrough Curve Analysis.

To minimize the damage from contaminant accidents in rivers, early identification of the contaminant source is crucial. Thus, in this study, a framework combining Machine Learning (ML) and the Transient Storage zone Model (TSM) was developed to predict the spill location and mass of a contaminant source. The TSM model was employed to simulate non-Fickian Breakthrough Curves (BTCs), which entails relevant information of the contaminant source. Then, the ML models were used to identify the BTC features, characterized by 21 variables, to predict the spill location and mass. The proposed framework was applied to the Gam Creek, South Korea, in which two tracer tests were conducted. In this study, six ML methods were applied for the prediction of spill location and mass, while the most relevant BTC features were selected by Recursive Feature Elimination Cross-Validation (RFECV). Model applications to field data showed that the ensemble Decision tree models, Random Forest (RF) and Xgboost (XGB), were the most efficient and feasible in predicting the contaminant source.

Seasonally varying effects of environmental factors on phytoplankton abundance in the regulated rivers.

This study investigates a seasonally varying response of phytoplankton biomass to environmental factors in rivers. Artificial neural network (ANN) models incorporated with a clustering technique, the clustered ANN models, were employed to analyze the relationship between chlorophyll a (Chl-a) and the explanatory variables in the regulated Nakdong River, South Korea. The results show that weir discharge (Q) and total phosphorus (TP) were the most influential factors on temporal dynamics of Chl-a. The relative importance of both variables increased up to higher than 30% for low water temperature seasons with dominance of diatoms. While, during summer when cyanobacteria predominated, the significance of Q increased up to 45%, while that of TP declined to about 10%. These tendencies highlight that the effects of the river environmental factors on phytoplankton abundance was temporally inhomogeneous. In harmful algal bloom mitigation scenarios, the clustered ANN models reveals that the optimal weir discharge was 400 m3/s which was 67% of the value derived from the non-clustered ANN models. At the immediate downstream of confluence of the Kumho River, the optimal weir discharge should increase up to about 1.5 times because of the increase in the tributary pollutant loads attributed to electrical conductivity (EC).

Validation of depth-averaged flow model using flat-bottomed benchmark problems.

In this study, a shallow water flow code was developed and tested against four benchmark problems of practical relevance. The results demonstrated that as the eddy viscosity increased, the velocity slope along the spanwise direction decreased, and the larger roughness coefficient induced a higher flow depth over the channel width. The mass conservation rate was determined to be 99.2%. This value was measured by the variation of the total volume of the fluid after a cylinder break. As the Re increased to 10,000 in the internal recirculating flow problem, the intensity of the primary vortex had a clear trend toward the theoretically infinite Re value of -1.886. The computed values of the supercritical flow evolved by the oblique hydraulic jump agreed well with the analytic solutions within an error bound of 0.2%. The present model adopts the nonconservative form of shallow water equations. These equations are weighted by the SU/PG scheme and integrated by a fully implicit method, which can reproduce physical problems with various properties. The model provides excellent results under various flow conditions, and the solutions of benchmark tests can present criteria for the evaluation of various algorithmic approaches.