Groundwater Journal's Special Section on Integrated Hydrologic Modeling

Groundwater’s Special Section on Integrated Hydrologic Modeling

The following is the press release from NGWA’s website announcing the issue.

January 17, 2019

Recently published scientific articles detail fresh perspectives regarding the future of groundwater modeling.

The articles were published in a special section within the January-February 2019 issue of NGWA’s flagship technical journal, Groundwater®.

The special section focuses on integrated hydrological modeling. In the past, saturated groundwater flow, unsaturated near-surface flow, and streamflow were all simulated independently from each other by separate computer programs. However, four research papers and two methods notes in this special section show the advances that have been made and future areas for research and development with integrated models representing the hydrological processes.

Within the section, there is a timely research paper on a coupled surface-subsurface hydrological model that includes winter processes to assess the impact of climate change. Another paper details a coupled surface-subsurface model for multiple flood events.

The section was put together with the aid of guest editors, Dr. Edward A. Sudicky of the University of Waterloo and Dr. Steven J. Berg, P.Geo. from Aquanty Inc., a science and technology firm in Waterloo, Ontario, Canada.

Groundwater Editor-in-Chief Dr. Henk Haitjema says, “guest editors Sudicky and Berg have been leaders in the field of integrated hydrological modeling. They collected four research papers and two methods notes in this special section to illustrate the state of the art as well as the challenges of this integrated approach to modeling. I hope that this special section will inspire young researches to further develop integrated hydrological modeling and bring it to maturity.”

Since 1963, Groundwater has published a dynamic mix of papers on topics focused on groundwater such as flow and well hydraulics, hydrogeochemistry and contaminant hydrogeology, application of geophysics, management and policy, and the history of hydrology.

HGS RESEARCH HIGHLIGHT - Estimating the Spatial Extent of Unsaturated Zones in Heterogeneous River-Aquifer Systems

AUTHORS: OLIVER S. SCHILLING, DYLAN J. IRVINE, HARRIE-JAN HENDRICKS FRANSSEN, AND PHILIP BRUNNER

The presence of unsaturated zones at the river-aquifer interface has large implications on numerous hydraulic and chemical processes. However, the hydrological and geological controls that influence the development of unsaturated zones have so far only been analyzed with simplified conceptualizations of flow processes, or homogeneous conceptualizations of the hydraulic conductivity in either the aquifer or the riverbed. We systematically investigated the influence of heterogeneous structures in both the riverbed and the aquifer on the development of unsaturated zones. The three fundamentally different states of connection resulting from the different degrees of saturation underneath the riverbed are conceptually illustrated below, including examples of the probability distributions of hydraulic conductivity of the riverbed and the aquifer that may lead to these states of connection (the saturated parts of the aquifer underneath the riverbed are illustrated in blue):

The investigations were based on a large number of numerical flow experiments using HydroGeoSphere. One of the goals of the study was to develop a simple method to predict the spatial extent of the unsaturated zone underneath a riverbed for heterogeneous river-aquifer systems, without the need to undertake complex numerical simulations. For this purpose, simulations of the following degrees of complexity were carried out:

Based on the results of the numerical simulations with HydroGeoSphere, a stochastic 1-D criterion that takes both riverbed and aquifer heterogeneity into account was developed using a Monte Carlo sampling technique. The approach allows the reliable estimation of the upper bound of the spatial extent of unsaturated areas underneath a riverbed. Through systematic numerical modeling experiments, we furthermore show that horizontal capillary forces can reduce the spatial extent of unsaturated zones under clogged areas. An example of these simualtions is provided below:

This analysis shows how the spatial structure of clogging layers and aquifers influence the propensity for unsaturated zones to develop: In riverbeds where clogged areas are made up of many small, spatially disconnected patches with a diameter in the order of 1 m, unsaturated areas are less likely to develop compared to riverbeds where large clogged areas exist adjacent to unclogged areas. A combination of the stochastic 1-D criterion with an analysis of the spatial structure of the clogging layers and the potential for resaturation can help develop an appropriate conceptual model and inform the choice of a suitable numerical simulator for river-aquifer systems.

Link to the published article.

HGS RESEARCH HIGHLIGHT - Efficient Uncertainty Quantification in Fully-Integrated Surface and Subsurface Hydrologic Simulations

AUTHORS: K.L. Miller, S.J. Berg, J.H. Davison, E.A. Sudicky, and P.A. Forsyth

Although high performance computers and advanced numerical methods have made the application of fully-integrated surface and subsurface flow and transport models such as HydroGeoSphere common place, run times for large complex basin models can still be on the order of days to weeks, thus, limiting the usefulness of traditional workhorse algorithms for uncertainty quantification (UQ) such as Latin Hypercube simulation (LHS) or Monte Carlo simulation (MCS), which generally require thousands of simulations to achieve an acceptable level of accuracy. In this paper we investigate non-intrusive polynomial chaos for uncertainty quantification, which in contrast to random sampling methods (e.g., LHS and MCS), represents a model response of interest as a weighted sum of polynomials over the random inputs. Once a chaos expansion has been constructed, approximating the mean, covariance, probability density function, cumulative distribution function, and other common statistics as well as local and global sensitivity measures is straightforward and computationally inexpensive, thus making PCE an attractive UQ method for hydrologic models with long run times. Our polynomial chaos implementation was validated through comparison with analytical solutions as well as solutions obtained via LHS for simple numerical problems. It was then used to quantify parametric uncertainty in a series of numerical problems with increasing complexity, including a two-dimensional fully-saturated, steady flow and transient transport problem with six uncertain parameters and one quantity of interest; a one-dimensional variably-saturated column test involving transient flow and transport, four uncertain parameters, and two quantities of interest at 101 spatial locations and five different times each (1010 total); and a three-dimensional fully-integrated surface and subsurface flow and transport problem for a small test catchment involving seven uncertain parameters and three quantities of interest at 241 different times each. Numerical experiments show that polynomial chaos is an effective and robust method for quantifying uncertainty in fully-integrated hydrologic simulations, which provides a rich set of features and is computationally efficient. Our approach has the potential for significant speedup over existing sampling based methods when the number of uncertain model parameters is modest ( ≤ 20). To our knowledge, this is the first implementation of the algorithm in a comprehensive, fully-integrated, physically-based three-dimensional hydrosystem model.

Link to the published article.

Read full manuscript here.

HGS RESEARCH HIGHLIGHT - Wetlands and Flood Mitigation in Ontario: Natural adaptation to extreme rainfall

AUTHOR: MASON MARCHILDON, P. Eng. Oak Ridges Moraine Groundwater Program

Wetlands are often recognized for their flood control value, but little research exists specific to Ontario, where extreme weather causing flooding poses ever-greater threats to urban areas. Ducks Unlimited Canada has undertaken new research to better understand the role of wetlands in storing and attenuating flood flows in an urban/rural watershed. The second phase of this research, reported here, employs advanced hydrologic modelling to address the questions of where and how wetlands are most effective at retaining water, what consequences further wetland loss may have on flooding, and what potential wetland restoration could have to improve flood storage within a watershed. The modelling was accomplished using fully-integrated, three-dimensional variably saturated hydrologic model built for the entire Credit River watershed at a high spatiotemporal resolution.

For more information click here.

HGS RESEARCH HIGHLIGHT - Limits of heat as a tracer to quantify transient lateral river-aquifer exchanges

AUTHORS: YUEQING XIE, AND JORDI BATLLE-AGUILAR

The application of heat as a tracer for assessing river-aquifer exchanges has been mainly limited to vertical flow through the riverbed. Lateral river-aquifer exchanges may be more important than vertical riverbed exchanges if the river is deeply incised into an aquifer. Few studies have examined lateral river-aquifer exchanges and the ability of heat to constrain such exchanges. This study aims to perform a robust assessment of the limits of heat as a tracer to quantify lateral river-aquifer exchanges. It is largely based on a section of the Meuse River in Belgium (Figure 1), a river predominantly gaining in the studied area that only becomes intermittently losing in the winter time.

Figure 1. Location of the study site adjacent to the Meuse River, monitored wells U5 and U3 (114 and 207 m distance from the river, respectively) and groundwater head contours as measured in April 2006.

A site-based transect model established with HydroGeoSphere was first calibrated using both hydraulic head and temperature time series from two monitoring wells over 100 m away from the river (U5 and U3). However, the temperature time series were not helpful in calibrating the model because of the large distance from the river and the gaining nature of the river.

The best-calibrated model was then utilised as the base case for assessing the usefulness of temperature at closer distances from the river. We extracted both head and temperature time series at a number of locations much closer to the river (i.e. 4, 5, 6, 7, 8, 9, and 10 m from the river) from the best-calibrated model. Then we analysed how the use of different synthetic heat and temperature time series as calibration targets impacts on the uncertainty of integrated river-aquifer exchange volume using the Monte Carlo approach (the river-aquifer exchange volume uncertainty is attributed to parameter uncertainties). Our results suggest that the ability of heat to reduce the uncertainty of lateral river-aquifer exchanges is directly proportional to the distance of the monitoring location from the river. In our case, the uncertainty range of the net exchange volume was reduced by approximately a factor of 3 from 4 m to 9 m (Figure 2a). This ability of course is limited to a certain range. For instance, heat cannot be used at 0 – 4 m in our case because of the occupation of the river bank, and was not useful beyond 8 m as the effect of river temperature becomes insignificant. The optimal distance is where groundwater temperature variation is no longer affected by river temperature (8 m in this study), or temperature variation is below the resolution limit of the temperature sensor. Our study also indicates that heat alone cannot constrain lateral river-aquifer exchanges better than the commonly used hydraulic head (compare Figures 2a and 2c). However, once combined with hydraulic head, heat can reduce the uncertainty of lateral river-aquifer exchanges significantly (compare Figures 2a and 2e). A factor of 3 – 6 reduction in the net exchange volume was observed in our synthetic case.

Figure 2. Net river-aquifer exchange volume statistics for using hydraulic head and temperature time series in different manners. The left panel shows results when time series were used at individual locations (i.e., 4, 5, 6, 7, 8, 9, 10 m from the river), whereas the right panel includes results when time series were used at specific ranges of locations (e.g., 4-5 m indicates locations at both 4 and 5 m, and 4-6 m indicates locations at 4, 5 and 6 m). In each boxplot, the upper and lower bounds show maximum and minimum values, the top and bottom of the box indicate 75 and 25 percentiles, and the red bar within the box is the median value. The dotted lines show the net exchange volume for the base case model.

Read full manuscript here

Australasian Groundwater Conference 2017

Aquanty is proud to support the Australasian Groundwater Conference this year. Swing by our booth to talk to us. We will also have two talks durring the Wednesday session on Groundwater Modelling.

River basin-scale integrated surface-subsurface hydrologic modelling to support agricultural risk management.

Simulating complex surface water/groundwater interactions during flood events with a fully-integrated physics based hydrologic model.

HGS RESEARCH HIGHLIGHT - Integrating hydrological modelling, data assimilation and cloud computing for real-time management of water resources

Authors: Wolfgang Kurtz, Andrei Lapin, Oliver S. Schilling, Qi Tang, Eryk Schiller, Torsten Braun, Daniel Hunkeler, Harry Vereecken, Edward Sudicky, Peter Kropf, Harrie-Jan Hendricks Franssen, and Philip Brunner

Online data acquisition, data assimilation and integrated hydrological modelling have become more and more important in hydrological science. In this study, we explore cloud computing for integrating field data acquisition and stochastic, physically-based hydrological modelling in a data assimilation and optimisation framework as a service to water resources management. For this purpose, we developed an ensemble Kalman filter-based data assimilation system for the integrated hydrological model HydroGeoSphere, which is able to run in a cloud computing environment. A synthetic data assimilation experiment based on the widely used tilted V-catchment problem showed that the computational overhead for the application of the data assimilation platform in a cloud computing environment is minimal, which makes it well suited for practical water management problems. Advantages of the cloud-based implementation comprise the independence from computational infrastructure and the straightforward integration of cloud-based observation databases with the modelling and data assimilation platform. 

For more information click here.

HGS RESEARCH HIGHLIGHT - Incorporating Surface Water Operations in an Integrated Hydrologic Model: Model Development and Application to the Lower Republican River Basin, United States

AUTHORS: A. Brookfield, C. Gnau, and B. Wilson

Few river systems remain unaltered by engineered water management structures. Yet research investigating the interdependence between natural and engineered components of a hydrologic system within a modeling framework is limited. Most current models either focus on the natural system, incorporating only a portion of the engineered structures and often excluding elements like reservoir operations, or focus on the engineered system, simplifying the temporal and spatial variations of the surface water/groundwater flow system. The objective of this work is to link an object-oriented model of surface water operations to a physically based, fully integrated surface/ subsurface hydrologic model to capture the effects of water management decisions on the groundwater and surface water flow systems. The capabilities of the new linked modeling framework are demonstrated in the heavily managed Lower Republican River Basin (LRRB) in portions of Nebraska and Kansas in the central United States. This area of the basin contains two storage reservoirs and a network of surface water canals from seven irrigation districts, in addition to thousands of diversions from groundwater and surface water irrigators. The linked model was able to reasonably represent the surface and groundwater flow conditions and demonstrated the interdependence between the surface water operations and the groundwater/surface water flow system. The temporal variability of groundwater/surface water interactions has a significant impact on reservoir operations and streamflow. This work demonstrated how integrating surface water operations and groundwater/surface water flow components into a model improves the representativeness of the simulated results and better captures the temporal and spatial variations in hydrologic processes occurring within the domain.

OASIS lower Republican River model schematic

Read the full manuscript here.

HGS RESEARCH HIGHLIGHT - Combined analysis of time-varying sensitivity and identifiability indices to diagnose the response of a complex environmental model

AUTHORS:  Mehdi Ghasemizade, Gabriele Baroni, Karim Abbaspour, and Mario Schirmer

Physically based models for simulating environmental processes are usually criticized due to having many parameters. This issue leads to over-parameterization and can finally reduce the uncertainty (reliability) of the simulated outputs. Sensitivity and identifiability analyses are common diagnostic tools to address over-parametrization in complex environmental models. In this study, we performed a temporal global sensitivity and identifiability analyses of HydroGeoSphere (HGS) model parameters. HGS was used to simulate daily evapotranspiration, water content, and recharge based on high quality data of a weighing lysimeter. Figure below shows the schematic of the lysimeter as well as the conceptual model for simulating the lysimeter. The model has four soil layers in addition to a preferential flow component. We found that identifiability of a parameter does not necessarily reduce output uncertainty. It was also found that the sensitivity of the model parameters is required to allow uncertainty reduction in the model output. 

Schematic of the weighing lysimeter.

Conceptual Model.

Read the full manuscript here.

HGS RESEARCH HIGHLIGHT - On the effects of preferential or barrier flow features on solute plumes in permeable porous media

Authors: Megan Sebbena and Adrian Werner

Discrete flow features (DFFs) such as fractures, faults, sand lenses and clay layers are common geologic features in groundwater systems. DFFs can provide preferential pathways (i.e. ‘preferential flow features’; PFFs) or act as barriers (i.e. ‘barrier flow features’, BFFs) to fluid flow and solute transport. Considerably less research attention has been paid to the role of DFFs in modifying groundwater flow and solute transport where the host rock is permeable, compared to low-permeability rocks. A numerical investigation was conducted within this study to explore how the distributions of solute plumes in permeable aquifers are influenced by a DFF.

HydroGeoSphere was used to evaluate the impact of 2D flow effects within DFFs, which were treated as thin bands of porous media. We show the changes to solute plumes that occur where both BFFs and PFFs are encountered in otherwise permeable rock aquifers (e.g. sandstone and limestone). Additionally, the potential role of ‘back dispersion’ (i.e. the anomalous movement of solutes from the PFF back into the matrix against the direction of groundwater flow) on predictions of PFF effects are explored. The numerical simulations were used to quantify the displacement and widening (or narrowing) of a steady-state solute plume as it crosses a DFF in idealised, 1 × 1 m aquifers. A simple analytical expression for the advective displacement of a solute plume encountering a DFF is provided.

The outcomes of this study suggest that PFFs typically have a more significant influence on plume distributions than BFFs, and the impact of DFFs on solute plumes generally increases with increasing aperture. Plumes crossing a PFF are less symmetrical, and peak solute concentrations beneath PFFs are considerably lower than plumes in BFF cases.

Steady-state salinity distributions for PFF Cases A, B and C (decreasing variability between the PFF and matrix hydraulic conductivities), and Scenarios 1, 2, 3 and 4 (increasing PFF aperture). PFFs are indicated by transparent white lines.

Steady-state salinity distributions for BFF Cases D, E and F (increasing variability between the BFF and matrix hydraulic conductivities), and Scenarios 1, 2, 3 and 4 (increasing BFF aperture). BFFs are indicated by transparent white lines.

Read the full manuscript here.