Prediction of River Stage Using Multistep-Ahead Machine Learning Techniques for a Tidal River of Taiwan

Time-series prediction of a river stage during typhoons or storms is essential for flood control or flood disaster prevention. Data-driven models using machine learning (ML) techniques have become an attractive and effective approach to modeling and analyzing river stage dynamics. However, relatively new ML techniques, such as the light gradient boosting machine regression (LGBMR), have rarely been applied to predict the river stage in a tidal river. In this study, data-driven ML models were developed under a multistep-ahead prediction framework and evaluated for river stage modeling. Four ML techniques, namely support vector regression (SVR), random forest regression (RFR), multilayer perceptron regression (MLPR), and LGBMR, were employed to establish data-driven ML models with Bayesian optimization. The models were applied to simulate river stage hydrographs of the tidal reach of the Lan-Yang River Basin in Northeastern Taiwan. Historical measurements of rainfall, river stages, and tidal levels were collected from 2004 to 2017 and used for training and validation of the four models. Four scenarios were used to investigate the effect of the combinations of input variables on river stage predictions. The results indicated that (1) the tidal level at a previous stage significantly affected the prediction results; (2) the LGBMR model achieves more favorable prediction performance than the SVR, RFR, and MLPR models; and (3) the LGBMR model could efficiently and accurately predict the 1–6-h river stage in the tidal river. This study provides an extensive and insightful comparison of four data-driven ML models for river stage forecasting that can be helpful for model selection and flood mitigation.

A comparative study of wavelet- and empirical mode decomposition-based GPR models for river discharge relationship modeling at consecutive hydrometric stations

River stage-discharge relationship has an important impact on modeling, planning, and management of river basins and water resources. In this study, the capability of Gaussian Process Regressions (GPR) kernel-based approach was assessed in predicting the daily river stage-discharge (RSD) relationship. Three successive hydrometric stations of Housatonic River were considered and based on the flow characteristics during the period of 2002–2006 several models were developed and tested via GPR. To enhance the applied model efficiency, two pre-processing techniques namely Wavelet Transform (WT) and Ensemble Empirical Mode Decomposition (EEMD) were used. Also, two states of the RSD modeling were investigated. In the state 1, each station's own data was used and in the state 2, the upstream stations’ datasets were used as input to model the RSD at downstream of the river. The single and integrated models results showed that the integrated WT- and EEMD-GPR models resulted in more accurate outcomes. Data processing enhanced the models capability between 25 and 40%. The results showed that the RSD modeling in the state 1 led to better results; however, when the stations’ own data were not available the integrated methods could be applied successfully for the RSD modeling using the previous stations’ data.

Assessing Surface water- alluvial aquifer water exchange using a multitracer approach and modelling

<p>Alluvial aquifers are generally highly productive in terms of groundwater and are therefore particularly exploited. The study site is a drinking water production facility located on the alluvial plain of the Rhône river, France. This site consists of several pumping wells and observation piezometers organized along the riverbank. The site is continuously supplying water to neighboring agglomerations with intermittent pumping. In this situation, the pumping produces a piezometric depression allowing leading to a water exchange from the river to the aquifer which is a common feature in the case of alluvial aquifer exploitation along a riverside.</p><p>The four pumping wells and five piezometers were equipped with continuous automatic temperature and water level measurement probes, the river stage is monitored as well. These data are used to determine the exchange (direction and magnitude) between the aquifer and the river. Although pumping is intermittent, it does not allow a sufficient recovering of the natural piezometric level, i.e. the aquifer is permanently below the river stage.</p><p>In addition to the automatic probes, additional data acquisition campaigns were carried out. During these campaigns different tracers were used such as conductivity, stable isotopes of water and radon activity. Together with the continuously measured temperature, these various tracers were used to identify hydrodynamic variables and parameters, such as Darcy’s velocity, dispersivity, transit times. A MODFLOW model was developed, integrating the site geometry and hydrodynamic context, with the Rhone River at the western boundary and the Ouveze river at the eastern boundary. Model calibration was performed using the study site piezometric records and the optimization package PEST. The flow was reproduced at the site for two situations, a natural situation without groundwater pumping, and the exploitation situation with the groundwater withdrawals. Finally, the tracer’s data were integrated into the model to reproduce the transport of different tracers, in order to quantify the exchanges and the water fractions coming from the different hydraulic boundaries.</p>

Innovative method for gathering river stage data using only the sound of the water

<p>Splosh, gurgle, burble are all terms that can be used to describe how a river sounds as we stand on the bank. We have developed a new approach that uses the passive sound generated by a river, to gauge the current stage of the river, and generate (sono)hydrographs from the safety of the river bank. Our approach offers a cost-effective, power-efficient and flexible means to install flood monitors. We have developed a method of how to take the sound from around a river and translate it into a useful gauging tool without the need to listen to individual recordings. Using an internet of things approach we have developed a system of sound monitors that can be placed anywhere in the vicinity of a river. We aim to target the lesser studied parts of a river catchment, the headwaters, which are often data scarce environments. These environments are an opportunity to identify the real time responses of sub-catchments. The ultimate goal of our research is to enable community level flood monitoring, in areas that may be susceptible to river flooding, but are not yet actively gauged.</p><p> </p><p>We hypothesise that the sound generated by a river is a direct response to the obstacles found within the channel and the turbulence they cause. Sound is generated by the increase of energy available in the channel, being transformed into sound energy through turbulence generating structures, i.e. boulders. Data gathered over a winter season from several rivers in the North East of England, during Storm Ciara and Dennis, has shown sound to be a reliable method for determining rapid changes in river stage and is comparable to what the official Environment Agency gauges measured. Through an innovative approach, we have begun to understand the limits on sound data and the calibration of sound to the channel properties. Utilising a 7.5 m wide flume at a white water course we have recreated controlled environments and simulated different discharges and their effect on sound.</p><p> </p><p>Overall, we have found that sound is an opportunity to be taken to measure river stage in areas that are seldom studied. We have identified that sound works during extreme conditions, and being placed on the banks of the channel our monitors have a lower risk of being damaged during storm events and are easy and safe to install. We present the first means of using sound from a river to actively gauge a river and the full workflow from collection, analysis and dissemination of results.</p>

Effects of hydropeaking on groundwater mixing: from laboratory experiments to field scale observations

<p>Hydropeaking, an artificial flow regime consisting on strong and frequent river stage fluctuations, is known to have important effects on groundwater-surface water interaction. It influences the transient dynamics of water flow and also of solute and energy fluxes between aquifers and rivers. In this work, we focus on the effects of hydropeaking at multiple spatial and temporal scales. We start the investigation at the laboratory scale using quasi-two-dimensional flow-through experiments in which we can characterize  flow and transport mechanisms, as well as the topology of the flow field, at high spatial and temporal resolution. We measure and model the spatial moments, the dilution index and the Okubo-Weiss parameter of a transient plume, and find a correlation between changes in flow topology and mixing enhancement. We then investigate a two-dimensional field scale cross section representative of the Adige aquifer in North-East Italy, where two rivers differently affected by hydropeaking influence groundwater flow, and we investigate the system considering hourly and mean daily fluctuations in the river stage. We characterize the transient groundwater dynamics for this and for other aquifers affected by hydropeaking using the Townley number, analyzing the potentiality of such systems for chaotic advection. Finally, at regional scale we use a three-dimensional transient model to show how the Adige aquifer is differently affected by hydropeaking depending on dry and wet years. Moreover, we apply the continuous wavelet transform to identify the main temporal scales of variability detected in the groundwater fluctuations and how they change with time. Our work therefore highlights the relevance of the effect of hydropeaking on groundwater flow and transport processes, and its impact on flow topology and mixing enhancement at multiple spatial and temporal scales.</p>