ERT measurements of salt intrusion in the laboratory

(1)

ERT measurements of salt intrusion in

the laboratory

(2)

2 of 56 ERT measurements of salt intrusion in the laboratory

11203677-007-ZKS-0002, 20 February 2020

ERT measurements of salt intrusion in the laboratory

Author(s)

Victor Hopman Marios Karaoulis

(3)

ERT measurements of salt intrusion in the laboratory

Client Rijkswaterstaat Centrale Informatievoorziening

Contact Marc Hartogs

Reference KPP DI01 2019 - Efficiënte Monitoring - subproject 7

Keywords

Document control

Electrical measurements, salinity, ERT tomography, salt intrusion

Version 0.3 Date 20-02-2020 Project nr. 11203677-007 Document ID 11203677-007-ZKS-0002 Pages 56 Status final Author(s) Victor Hopman Marias Karaoulis

Doc. version Author Reviewer Approver Publish

0.3

Vic Hopman Marias Karaoulis

3 of 56 ERT measurements of salt intrusion in the laboratory 11203677-007-ZKS-0002, 20 February 2020

(4)

11203677-007-ZKS-0002, 20 February 2020

Executive summary

(5)

11203677-007-ZKS-0002, 20 February 2020

About Deltares

Deltares is an independent institute for applied research in the field of water and the subsurface. Throughout the world, we work on smart solutions, innovations and applications for people, environment and society. Our main focus is on deltas, coastal regions and river basins. Managing these densely populated and vulnerable areas is complex, which is why we work closely with governments, businesses, other research institutes and universities at home and abroad. Our motto is ‘Enabling Delta Life’.

As an applied research institute, the success of Deltares can be measured by how much our expert knowledge can be used in and for society. At Deltares, we aim to use our leading expertise to provide excellent advice and we carefully consider the impact of our work on people and planet.

All contracts and projects contribute to the consolidation of our knowledge base. We always apply a long-term perspective when developing solutions. We believe in openness and transparency. Many of our software, models and data are freely available and shared in global communities.

Deltares is based in Delft and Utrecht, the Netherlands. We employ over 800 people from 40 countries. We have branch and project offices in Australia, Indonesia, New Zealand, the Philippines, Singapore, the United Arab Emirates and Vietnam. Deltares also has an affiliated organisation in the USA.

(6)

6 of 56 ERT measurements of salt intrusion in the laboratory 11203677-007-ZKS-0002, 20 February 2020

1 Management summary

Salt intrusion of surface waters in the Netherlands poses a problem for fresh water resources, for example at drinking water intake points. Currently, the tools and instruments to monitor and understand salt intrusion are insufficient. The point sensors of the monitoring network provide valuable information, but only at specific locations where measurements are made. RWS is looking for new technologies to gather (2D or 3D) information about salinity variations in fresh, brackish and salt-water bodies. Two promising techniques are Electrical Resistivity Tomography (ERT) and Distributed Acoustic Sensing (DAS) with fibre optic cables. This report describes the results of a laboratory test using ERT. The results of a laboratory test using DAS are described in a separate report (Kruiver et al., 2019).

Electrical Resistivity Tomography is a well-known geophysical technique used all over the world, mainly on land. The technique uses electricity to image sub-surface structures of electrical resistivity. In recent years there has been application of this technique for measuring salinity in waterbodies. In general, ERT is considered a suitable tool to map and monitor salt/fresh water movements both in the water column and in the subsurface.

In this study, laboratory experiments were conducted to test if the ERT technique could potentially be applied in a field setting for measuring salinity intrusion in inland waterways and to demonstrate a feasible monitoring set-up to do so. These laboratory experiments were conducted in

cooperation with another ongoing laboratory test using the IJmuiden scale model, constructed in the Hydrohal facility of Deltares. Four experiments were performed, using different configurations of ERT cables. Of these, two of the experiments, with cables on the bottom, are directly relevant for a future field test. The two additional experiments, with cables at the surface and bottom, were conducted to provide additional measurements to help evaluate the quality of the results and to demonstrate the maximum level of information that can be obtained using multiple cables.

The laboratory experiments as described in this report show that ERT can be used to monitor the intrusion of seawater into inland water systems, providing good temporal resolution and more spatial information than available from point measurements. The imaging of the salt water intrusion can be done with one horizontal electrode cable on the bottom (achievable in practice in a canal) and with two cables vertically above one another. For better spatial resolution, two or more parallel cables on the bottom can be used. This is the configuration that is recommended for a field test.

Some initial recommendations for conducting a field test, also in conjunction with a test of the DAS method, are provided. The recommendations include suggestions on how to set-up the sensors to obtain meaningful results with respect to relevant spatial and temporal resolution.

(9)

11203677-007-ZKS-0002, 20 February 2020

2 Management samenvatting

Zoutindringing van oppervlaktewateren in Nederland vormt een probleem voor

zoetwatervoorraden, bijvoorbeeld bij drinkwaterinname. Momenteel zijn de instrumenten om zoutindringing te controleren en te begrijpen onvoldoende. De puntsensoren van het monitoringnetwerk leveren waardevolle informatie, maar alleen op specifieke locaties waar metingen worden verricht. RWS is op zoek naar nieuwe technologieën om (2D of 3D) informatie te verzamelen over zoutvariaties in zoet, brak en zout water. Twee veelbelovende technieken zijn Electrical Resistivity Tomography (ERT) en Distributed Acoustic Sensing (DAS) met

glasvezelkabels. Dit rapport beschrijft de resultaten van een laboratoriumtest met ERT. De resultaten van een laboratoriumtest met DAS worden beschreven in een afzonderlijk rapport (Kruiver et al., 2019).

ERT is een bekende geofysische techniek die over de hele wereld wordt gebruikt, voornamelijk op het land. De techniek maakt gebruik van elektriciteit om structuren onder het oppervlak van elektrische weerstand af te beelden. De laatste jaren is deze techniek toegepast voor het meten van het zoutgehalte in waterlichamen. Over het algemeen wordt ERT gezien als een geschikte methode om zout/zoet water in zowel de waterkolom als in de ondergrond in kaart te brengen en te volgen

In deze studie werden laboratoriumexperimenten uitgevoerd om te testen of de ERT-techniek mogelijk zou kunnen worden toegepast in een veldomgeving voor het meten van zoutindringing in binnenwateren en om een haalbare monitoringopstelling aan te tonen. Deze

laboratorium-experimenten werden uitgevoerd in samenwerking met een andere lopende laboratoriumtest met behulp van het schaalmodel IJmuiden, gebouwd in de Hydrohal-faciliteit van Deltares. Vier experimenten werden uitgevoerd met verschillende configuraties van ERT-kabels. Twee van deze experimenten, met de kabels op de bodem, zijn relevant voor de toekomstige veldproef. De andere twee experimenten, met kabels op de bodem en aan het wateroppervlak, zijn uitgevoerd om met extra metingen de kwaliteit van de resultaten te beoordelen en om de maximale

hoeveelheid data te verkrijgen.

De laboratoriumexperimenten zoals beschreven in dit rapport tonen aan dat ERT kan worden gebruikt om de indringing van zeewater in binnenwateren te volgen, waarbij een goede temporele en meer ruimtelijke informatie wordt geleverd dan beschikbaar is bij puntmetingen.

Het is al voldoende om een enkele kabel op de bodem te plaatsen (toepasbaar in de praktijk op een kanaal) of met twee kabels verticaal boven elkaar. Voor een beter ruimtelijk inzicht kunnen twee of meerdere parallelle kabels op de bodem worden geplaatst.

Tenslotte worden er enkele aanbevelingen voor het uitvoeren van een veldtest gegeven ook in combinatie met een test van de DAS-methode. De aanbevelingen bevatten suggesties voor het instellen van de sensoren om goede resultaten te verkrijgen met betrekking tot ruimtelijke en temporele resolutie.

(10)

11203677-007-ZKS-0002, 20 February 2020

3 Introduction

3.1 Background

Salt intrusion of surface waters in the Netherlands poses a problem for fresh water resources. Mitigating measures are being designed and implemented, often in conjunction with large-scale reconstruction works, such as the new shipping locks at IJmuiden and Terneuzen. At existing locks with salt intrusion problems (e.g. Krammer, IJmuiden, Den Oever and Rhine delta) solutions such as bubble screens, pumps and narrowing the waterway are being investigated. The current models and system knowledge of salt intrusion, however, are insufficient for studying and monitoring possible solutions effectively (Schroevers, 2014). The measurement techniques and strategies currently employed by RWS are not suitable for the determination of the salt load or the shape of the salt plume within the water column.

In 2017, an initial evaluation of various techniques for measuring and monitoring salt intrusions was conducted (Schroevers et al., 2017). The most promising techniques were found to be Electrical Resistivity Tomography (ERT) and acoustic measurements with fiber optics (Distributed Acoustic Sensing, DAS). ERT in water bodies is primarily influenced by the salinity of the water and to a lesser extent by the temperature. Therefore, the ERT method is expected to be suitable to track salt water intrusions in fresh water, because the salt results in a change in electrical resistivity.

In 2018 and 2019, laboratory tests and experiments were performed for both techniques. This report describes the experiments conducted for ERT. The DAS laboratory experiment is reported in a separate document (Kruiver et al., 2019).

3.2 Scope of the test

The overall goal of the KPP project “Proof of concept new measurement techniques for salt intrusion” is to obtain a future-proof strategy for acquiring data and monitoring salt intrusion in Dutch surface waters. The purpose of the laboratory test was to demonstrate that the ERT method works in a scale model and that this method can be scaled up to a field situation in the

Amsterdam-Rhine Canal (ARK) or other similar setting. A specific aim of the test was to determine whether the differences in electrical resistivity can be used to monitor the dynamic salt water intrusion in a 2D vertical plane, specifically in the cross-section of a canal.

An additional objective was to gain insight into the applicability of the method in a field setting with respect to a number of practical aspects such as: spatial resolution, temporal resolution,

measurement resolution, detection limit, detection range, and accuracy. These factors will be relevant in assessing the possibilities of using ERT in a field situation and for designing a field test.

The setup of the experiment was dictated by another project “Selectieve onttrekking IJmuiden” (Dos Santos Nogueira et al, 2019). This setup and the approach for the different experiments conducted are described in more detail in Chapter 6.

(11)

11203677-007-ZKS-0002, 20 February 2020

3.3 Outlook for the future

When the laboratory test indicates that the dynamics of salt intrusion can be determined with sufficient resolution and accuracy, a field test is a logical next step. Such a field test could include both ERT and DAS measurements. The preferred location identified by RWS would be a site in the Amsterdam-Rijn Canal, near Diemen. The heavy ship traffic, however, can possibly influence the electrical measurements. Therefore, RWS is also considering alternative locations.

In Schroevers et al. (2017) the initial suggestions for a field test are described. These recommendations are updated in the current report, using the results from the laboratory experiments.

3.4 Reading guide

Chapter 4 describes the basic principles of ERT. Considerations regarding ERT measurements are described in Chapter 5. In Chapter 6 the ERT laboratory setup can be found. In Chapter 7 the calibration is explained and in Chapter 8 the conversion to density and salinity. Chapter 9

describes the test results. The conclusions are included in Chapter 10. The report ends with recommendations for a field test in Chapter 11.

(12)

11203677-007-ZKS-0002, 20 February 2020

4 Basic principles of the ERT method

This chapter describes the basic principles of the ERT method. Aspects such as the spatial resolution, accuracy and processing of the data are described in chapter 5.

Electrical resistivity tomography (ERT) is a geophysical technique that has traditionally been used for imaging sub-surface structures from electrical resistivity measurements made at the surface, or by electrodes in one or more boreholes. The electrical currents in the subsurface are influenced by the composition of the matrix (e.g. sand, gravel, clay) and the composition of the water in the pores. In general, sand has a higher resistivity than clay and salt water has a lower resistivity than fresh water.

A typical acquisition system for Direct Current resistivity measurements is shown in Figure 1. This figure illustrates an application to the subsurface (on land), but the basic principle is also

applicable for measurements in water. The basic setup comprises a four-electrode array with a volt meter and a current meter. For each measurement, two current electrodes are used: one to inject the current into the subsurface/waterbody and the other to retrieve the same amount of current from it (note that in effect a single electrode cannot carry current into a medium). By convention, these current electrodes are named A and B, respectively. The electric field is measured (at least) with two other electrodes (M and N), called the potential electrodes. The current electrodes cannot be used at the same time to measure the associated drop of the electric potential in the medium, due to the contact impedance across the electrodes, especially at low frequencies. Therefore, the two potential electrodes (M and N) have to be used for that purpose.

Figure 1: a) The principle of an ERT measurement, showing the currents in a cross-section; b) The effect of the spacing in the look-ahead of the measurements; c) Using 3 or 2 electrodes to increase the look-ahead of the measurements (source: Karaoulis et al., 2014)

(13)

11203677-007-ZKS-0002, 20 February 2020

Although the methods of ERT were originally used for measuring subsurface soils, they can also be applied to water systems. In this case, the electrodes are placed at the sediment bed or in the water column and are used to measure the resistivity of the water body. In Section 9.1 a field example is provided.

The resistance R (in ohm) measured across a cylindrical sample of resistivity ρ (in Ohm m), length L, and cross section A is given by R = ρ L/A, where g = A/L (in m) is called the geometric factor. For field acquisition, the geometric factor depends on the position of the electrodes as discussed below. The resistance is obtained by applying Ohm's law, U = R I, where U is the voltage

(difference of potential) in volts measured between M and N and I is the strength in amperes of the injected current. During measurements, the volt- and current meter are calibrated to check the accuracy of the measurements for a broad range of resistances.

By using multiple electrodes spread on the surface or in the water bottom (for resistivity

acquisition), a tomographic resistivity image of the waterbody can be obtained. The way in which the current and potential electrodes are arranged is called a resistivity array. A large number of electrode arrays have been suggested in the literature, but only a few such as ‘dipole-dipole’, ‘Schlumberger’ and ‘Wenner’ array, are extensively used. The main characteristic of an array is its geometrical factor, which is uniquely related to the respective distances between the probes as discussed above. The choice of a particular resistivity array for a survey is based upon

considerations regarding theoretical advantages and drawbacks of the array and its signal-to-noise ratio (Ward, 1990). More recently, approaches have been suggested to compute optimized measurement protocols, i.e., which provide best resolution, based for instance on sensitivity criteria (e.g., Stummer et al., 2004).

In the general case of a heterogeneous medium, for any possible four-electrode arrangement the geometrical factor g, when multiplied with the measured resistance R, yields the so-called apparent resistivity ρa,

a

R g



= 

The apparent resistivity ρa represents a weighted average of the true resistivity of the subsurface. By definition, the apparent resistivity is equal to the true resistivity in the case of a homogeneous waterbody. Generally, however, the true resistivity is only obtained after inversion of the apparent resistivity data (see section below).

Inverse modelling is the procedure to convert apparent resistivity data to an inverted resistivity image, also called a tomogram. Correspondingly, the approach to construct an electrical tomogram is also referred to as electrical resistivity tomography (ERT). Inverse modelling is usually performed with deterministic approaches. These approaches look for retrieving the true resistivity distribution of the ground from the apparent resistivity data. In other words, inverse modelling seeks to find a resistivity model that explains (or “predicts”) the given field

measurements. It is obvious, that the initial step of inverse modelling is to perform a forward modelling (what are the apparent resistivity data for a given resistivity distribution?). This is done by solving numerically the Poisson equation for the electric potential. There are many references on how to solve the Poisson equation in both 2D and 3D, using numerical methods like the finite-element method, the finite-difference method, or the boundary finite-element method (e.g., Coggon, 1971; Dey and Morison, 1979a, b; Rijo, 1977, 1984; Pridmore et al., 1981; Tsourlos, 1995; Xu et al., 1998). More general information about the method and its applications can be found in Revil et al., 2012.

(14)

11203677-007-ZKS-0002, 20 February 2020

5 Considerations about ERT measurements

Several considerations regarding the ERT method are described in this chapter. These considerations are relevant for the layout of the laboratory test and for a future field test.

5.1 Influence of temperature and salinity on resistivity

The resistivity of water is influenced by salinity and temperature. The effect of salinity variations on resistivity is much larger than the effect of temperature. Figure 2, left panel, shows that the resistivity is inversely proportional to the temperature, i.e. as temperature increases, the resistivity decreases. Over a temperature increase from 0 °C to 30 °C, the resistivity decreases by

approximately one half (from roughly 30 to 15 omh.m). For a temperature change of one degree, the resistivity varies only slightly. The effect of varying salinity is much more pronounced (Figure 2, middle panel): there is a sharp decrease in resistivity between fresh water (salinity < 150 mg/l) and salt water (salinity > 10,000 mg/l). The right panel of Figure 2 shows the combined effect of temperature and salinity. The curves for different temperatures are quite close together and all show the marked contrast between fresh and salt water.

Figure 2: Dependence of resistivity on temperature (Fofonoff and Millard,1983, left panel), salinity (de Louw et al., 2011, middle panel) and the combined temperature and salinity effect (de Louw et al., 2011, right panel).

5.2 Spatial resolution and coverage

The spatial resolution of the resistivity measurements is determined by the electrode spacing in the cables. The electrodes are installed in cables with fixed distances between the electrodes. The horizontal resolution of the resistivity measurements is equal to half of the distance between the electrodes. The vertical resolution decreases with increasing distance from the cable. In the inversion, the waterbody is divided into horizontal model layers of increasing thickness. The 1st layer has a thickness of half of the electrode spacing. Each subsequent layer is 10% thicker. If auxiliary data is available (in this case the conductivity sensor), the vertical resolution matches with the horizontal resolution.

(15)

11203677-007-ZKS-0002, 20 February 2020

The spatial coverage is determined by the location and placement of the electrodes. A single electrode cable provides information of resistivity of a 2D plane in the same direction as the cable. Multiple electrode cables cover a 3D volume if electrodes from different cables are used in the electrode configuration. The resolution depends on the spacing between the cables, because this dictates the distance between the electrodes. In the experiments conducted, the resolution was 0.25 m (in cable direction) x 0.4 m (between cables) x 0.2 m (vertical direction).

5.3 Measurement resolution and accuracy

Resistivity measurements are sensitive over a very large range of values. It is well within the specifications of the ERT unit to measure resistivities from 10-6_{ohm.m (wire) to ~10}3 _{ohm.m (dry} sand). The resistivities that are relevant for measuring changes in salinity and that were measured in this experiment range from 0.2 ohm.m (salt water) to ~22 ohm.m (fresh water). These values fall well within the dynamic range of the standard ERT instruments. This range makes ERT a suitable tool for fresh/salt water mapping. The instrument specifications state that the measurement accuracy is 0.05%.

5.4 Accuracy of resistivity of the inversion result

ERT provides a measurement of the resistivity of water in units of ohm.m. The difference between the resistivity of salt water and fresh water is large and thus, in principle, the measurement allows good resolution of salinity gradients. The accuracy of the inversion model result and the model grid cell size are related. Sharp contrasts, such as between fresh and saline water might not be well resolved due to the grid cell size. The values of two model cells on either side of a sharp transition might not be exact. Sharp contrasts tend to be smeared in the model when using standard processing. With an adjusted processing scheme designed for sharp contrasts, the inversion result can be improved. If auxiliary data is available (such as conductivity

measurements), the values can be further improved, and model resistivity values become closer to the true values. In the experiments conducted in this study, the transition from fresh to salt water can clearly be observed, because it is represented by a difference of two orders of

magnitude in the resistivity. Even if the data are very noisy, the transition can still be distinguished. The inversion output includes an uncertainty metric: the RMS error. This metric provides

information about the goodness of fit between the data and the inversion model result. Typically, an RMS error of less than 10% represents a very high-quality of the inversion model result. Repeated measurements and auxiliary data increase confidence in the model result.

5.5 Measuring frequency

Several types of measurement frequencies are important for time-lapse measurements. The first is the hardware measuring frequency. A typical measurement frequency is 6 Hz, meaning that current can be injected 6 times per second. The instrument of Deltares is an 8-channel system, meaning that for each current injection 8 pairs of potential electrodes can record the potential. With a measurement frequency of 6 Hz and an 8-channel system, a maximum of 48 potential measurements can be taken per second. In practice, this maximum is not reached, because most systems use mechanical switches that need a few msec to rotate, the power supply needs some extra msec to build up the power and a few msec are needed to store the data to memory. A more realistic estimation is about 16-24 measurements per second. There are systems with more channels in the market (i.e. ABEM is 12 channels, Iris is 10 channels). Custom based systems can be built, especially for applications where measurements with different cables are required.

(16)

11203677-007-ZKS-0002, 20 February 2020

Another type of frequency is the repeat frequency of each complete set of electrode combinations. It takes time (one cycle) to measure a set of electrode combinations. If the cycle is fast compared to the process that is monitored, the cycle can be regarded as one instance in time. For example, the resistivity of a system is expected to change slowly over a period of one day. A measurement cycle of 10 minutes would then result in a representative situation. If the duration of the cycle is in the same order as the expected change, the result would be somewhat smeared over time. One measurement cycle would in that case not be representative for the situation at one instance in time. The chloride concentration during the passage of a salt intrusion peak in the ARK will typically happen with one to four hours (example in Figure 3 ). The measurement cycle duration should therefore be short, relative to the intrusion peak, without missing relevant electrode combinations.

Figure 3: Example of peaks in chloride concentrations in the ARK near Diemen. Source: www.waterinfo.rws.nl The duration of the measurement cycle can be optimized. Not all combinations of electrodes are useful for the inversion. If the measuring array is optimized, it can be reduced from e.g. about 30000 combinations to 300 or less. This speeds up one cycle of measuring all electrode combinations defined in the measurement protocol. The cycles can be sped up even more if enough data is available over time, enabling the tailoring of the prediction algorithms. For example, one cycle of measurements during the first laboratory tests using a single cable with 96 electrodes took 10 minutes. If only the three bottom cables are used instead of bottom and surface cables, the measurement cycle is expected to take ~5 minutes. Further optimization might even reduce the measurement cycle to 2 to 3 minutes.

5.6 Overall QC data

The most important factor affecting data quality for ERT is the contact resistance: the contact between the sensor and the medium. In the soil this can be a problem if the soil is dry. In water, however, the contact resistance is extremely low. Therefore, current injection is easy in water bodies. We expect to acquire good quality data.

(17)

11203677-007-ZKS-0002, 20 February 2020

5.7 Processing of data

Standard ERT is usually processed using standard commercial software such as 2Dresinv or 3Dresinv (https://www.geotomosoft.com/). The special configuration of ERT in water and specifically in laboratory settings, requires custom geometrical factors. These factors can be modelled using more advanced software, such as Comsol (https://www.comsol.com/). For processing the data from the laboratory experiments, Comsol was used to simulate the exact position of the electrodes and brick walls. Also, the forward solution was modelled with Comsol. The custom-based software package IPI4D (Karaoulis et al., 2014) was used for the inversion, because it allows for any settings and any boundary conditions, e.g. due to the irregular shaped tank. Probably, the advanced processing will also be required for a field experiment.

(18)

11203677-007-ZKS-0002, 20 February 2020

6 The ERT Laboratory setup

As stated in the introduction, the overall goal of this project is to obtain a future-proof strategy for acquiring data and monitoring salt intrusion in Dutch surface waters. The purpose of the laboratory test was to demonstrate that the ERT method works in a scale model and that this method can be scaled up to a field situation in the Amsterdam-Rhine Canal (ARK) or other similar setting. A specific aim of the test was to determine whether the differences in electrical resistivity in a 2D vertical plane, can be used to monitor the dynamic salt water intrusion.

6.1 General approach

The general approach was to test a number of different electrode cable configurations in the laboratory setup. These configurations included:

• Cable(s) on the bottom only;

• Cable(s) on the bottom and at the surface.

Cables on the bottom:

The simplest configuration consists of one electrode cable on the bottom of the tank. This results in a vertical 2D cross section in the same direction as the cable. If several cables are placed in parallel on the bottom, and electrodes from different cables are used, the result is a 3D model of resistivity. In this case, cross sections in all directions can be extracted from the model, i.e. also perpendicular to the direction of the cables. These two configurations are relevant for situations where no electrode cable can be placed on the water surface, such as in the ARK because of the shipping traffic.

Cables on the bottom and at the surface:

In the laboratory situation, electrode cables can be placed at the surface. In a two-cable set-up, one at the bottom and one parallel at the surface, the result is a 2D cross-section between the cables. The most extensive configuration consists of several parallel bottom cables and several parallel surface cables. The result is a 3D model of resistivity. In principle, the spatial resolution results from the bottom and surface cables is better than that of bottom cables only. The bottom and surface configurations were added to the experiments to confirm that the bottom-only result is of sufficient quality to be used in a field experiment.

The laboratory test plan that was followed for the experiments is presented in Appendix A “Plan van aanpak ERT laboratorium experiment” (in Dutch).

6.2 The laboratory scale model configuration

The setup of the ERT laboratory test was to a certain extent dictated by another laboratory experiment focused on selective removal of salinity at the IJmuiden shipping locks (“Selectieve onttrekking IJmuiden”; Dos Santos Nogueira et al., 2019). This IJmuiden study included the set-up of a scale model to investigate selective removal of salt and included tests with salinity intrusion originating from the operation of the shipping lock near the North Sea.

Combining the ERT laboratory tests with the study in the IJmuiden scale model provided many advantages and considerable time and cost savings. The IJmuiden project created the opportunity for setting up an ERT monitoring configuration and supported all the necessary measurements. Specifically, it made it possible to conduct experiments using saline water, which require significant preparation and special permits. The ERT experiments could very conveniently make use of all these arrangements that were already in place.

(19)

11203677-007-ZKS-0002, 20 February 2020

An overview of the scale model is shown in Figure 4 including the area of interest for the ERT measurements (yellow cable). Saline water enters the scale model on the right side, which is the upstream side of the scale model. The saline tongue spreads to the downstream area where the ERT Area of Investigation is located. The Selective Removal chamber is to the left of the ERT Area of investigation. A more complete description of the IJmuiden scale model is given in appendix B.

Figure 4: Sketch of the tank location, salt water inlet on the right, ERT area of interest (yellow cable). The injected salt water flows from the inlet at the upstream side towards the selected removal chamber

downstream. This represents the situation in the ARK where regularly there is intrusion of a salt tongue. This picture shows the experiments in June 2019. One experiment (August 2019) was conducted with an opposite flow direction.

6.3 Cable Installation

For the experiments, three ERT cables were installed at the bottom of the tank, fixed to the tank floor. To measure not only the single bottom cable configuration, but also the double setup, three additional cables were installed at the water surface. The spacing between the electrodes in each cable was 0.5 m, while the interspacing between the bottom and top cables was 0.4 m. The distance between the bottom cables was 0.8 m. The distance between the surface cables was the same: 0.8 m. Fifteen electrodes were used along each cable. The installation and configuration of the ERT cables in the tank is shown in Figure 5. With this installation, it was possible to conduct different experiments, selectively collecting data from one or more of the installed bottom and top cables (see Section 6.5).

(20)

11203677-007-ZKS-0002, 20 February 2020

Figure 5: The installation of ERT cables in the tank

6.4 Geometrical factor

The ERT cables were installed in a tank with brick walls (Figure 5). The positioning of the electrodes with respect to the tank boundaries is crucial to calculate the geometrical factor correctly. In the laboratory setting, the spacing of the electrodes with respect to the desired resolution is not ideal. In a real field application, the resolution can be optimized with the electrode cable choice.

A three-step approach to calculate the geometrical factor was followed:

1 As a first step, the theoretical geometrical factor was calculated using the geophysical modelling software Comsol, based on the tank dimensions and the electrode positions (Figure 6).

2 For the second step, the tank was filled with fresh water (with known resistivity value of 22 Ohm.m at room temperature) and the geometrical factor was measured. The difference between the measured and theoretical factor was then calculated. The small differences observed originated from the small errors in the positioning of the electrodes. These differences were included in the new geometrical factor.

3 The new geometrical factor was used for each consecutive experiment.

It is important to note that in field applications the first two steps can be skipped, because the only boundaries that exist are the air-water boundary and the water/bottom boundary. In addition, small uncertainties in the positioning of the electrodes are not important because the length scales in the field are much larger and these positioning errors can be ignored.

(21)

11203677-007-ZKS-0002, 20 February 2020

Figure 6: The ERT model area as visualised in the Comsol software. Blue dots indicate the position of the electrodes, gray is the inverted volume. The total area of investigation covered by the ERT cables was 7.5m (length) x 1.6m (width) x 0.4m (height).

6.5 Measurement configurations

Since ERT was measuring over time, the data can be processed in a lapse sense and time-related changes can be observed. Two basic configurations are presented in Figure 7. The measurements were done using M and N electrodes from the same cable, and from different cables (M on one cable and N on the other cable). The measurement array encompasses

measurements using electrodes only on one cable at the bottom or measurements with electrodes in different cables.

One measurement cycle of all combinations of A&B and M&N takes a certain amount of time. To minimize the time-interval between different ERT cycles, the measurement protocols were

optimized (e.g., Stummer et al., 2004). After this process, there were 672 measurements per cycle and a time-interval between two cycles of 10 minutes.

(22)

11203677-007-ZKS-0002, 20 February 2020

Figure 7: The principle of the ERT measurements in a water tank using cables on the bottom (viewed from above): double bottom cable using current and potential electrodes from both cables (left); single bottom cable (right). Used in the simulation in Comsol.

6.6 Discretisation in Comsol for forward model

The total area of investigation covered by the ERT cables was 7.5 x 1.6 x 0.4 m (see Figure 5 and Figure 6). In the Comsol software, this is discretized into cells of 0.25 x 0.20 x 0.05 m. It must be noted that the resolution of the ERT image is dictated by the electrode spacing (0.4 to 0.8 m in this case). The typical resolution value is half the electrode spacing. The expected resolution in the laboratory experiment in the tank environment with the limited water depth (about 0.4 m) is 0.2 m. A possibility to achieve a higher spatial resolution would have been decreasing the electrode spacing by introducing several more cables at various depths. However, this was not a desired approach, as the expectation for a field test is that all cables would be at the bottom. As a solution, some optimizations to the inverse code and the pre-processing of the data and data from a close-by installed conductivity meter were used to increase the vertical resolution. Notice that those steps are not necessary for a field application.

(23)

11203677-007-ZKS-0002, 20 February 2020

6.7 Experiments conducted

In the laboratory test for ERT, four different experiments were conducted, summarised in Table 1 and described below.

Table 1: Four ERT configurations tested in the laboratory.

experiment configuration injection basin fill date Presentation in chapter 9

1 one bottom cable

salt water flow from upstream to downstream

fresh water 19 June 2019 density

2 multiple bottom cables

fresh water 19 June 2019 density

3 one double cable (bottom &surface)

salt water flow from downstream to upstream

brackisch water 16 August 2019 relative changes of resistivity 4 multiple double cables

fresh water 19 June 2019 density resistivity

1. One bottom cable (2D):

In this experiment, data from only one bottom cable was used for the inversion calculation. This configuration could be tested in the field, for example at field site ARK or another selected location, where it is not possible to have measurement cables at the water surface due to interference with shipping traffic. Because a saline water intrusion moves along the bottom it is in principle possible to use only a bottom cable to observe the salinity changes. This cable

configuration provides vertical 2D information of the salinity distribution along the length of the cable. The results from this single cable experiment are given in section 9.1. In addition to the lab experiment with the bottom cables, some results of a field experiment with a bottom cable are included. The reason for including this example is that it shows the possibilities of the method under field conditions similar to the ARK field site: a single bottom cable monitored the salt leakage underneath a dam.

2. Multiple bottom cables (3D)

Instead of using only one bottom cable for inversion, data from all parallel bottom cables were used in the inversion. This configuration is also considered to be appropriate for application in the ARK since there are no cables at the water surface. Similar as for a single bottom cable: a salt intrusion moving along the bottom makes it possible to use bottom cables only. An advantage of this set up compared to the single cable is the possibility of identifying the direction of the salt water intrusion. This setup provides a fast measurement cycle, with high spatial resolution in all dimensions, while keeping cables away from the surface. In this setup a larger area can be covered, and more detail can be achieved due to extra electrode combinations. The results from this cable experiment are given in section 9.2.

3. One double cable (surface & bottom, 2D)

In the third experiment, one cable was situated at the bottom and a second cable was just under water surface vertically above the bottom cable. This configuration is not considered appropriate for application in the ARK, due to shipping traffic. However, in the laboratory setting, it provides additional measurements which are useful to evaluate the quality of the single bottom cable configuration. Additionally, this configuration could potentially be used in a field setting if the surface cable could be placed close to shore or near a structure (such as one of the national measurement stations ‘LMW-meetpaal’ of RWS) and would therefore not interfere with shipping

(24)

11203677-007-ZKS-0002, 20 February 2020

traffic. Another possible option would be to install the cables vertically at the sides of the river, and the area of investigation is a cross-section between them. A third possible option of using this configuration is with a mobile monitoring platform, in which the cables could be towed behind a measurement ship. In this mobile system, an upper and a lower cable could be used to provide information of salt water intrusion at different locations. A single towed cable could possibly be sufficient for this purpose, but a double cable would provide more detail. This cable configuration provides 2D information of the salinity distribution. The results from this double cable experiment are given in section 9.3.

4. Multiple double cables (bottom & surface, 3D)

A fourth experiment was conducted to demonstrate the maximum information that can be obtained by ERT, with the most complete coverage of the water system. The cable configuration chosen for this test provides 3D information of the salinity distribution, in contrast to cases 1 and 3 where only 2D information is generated. To realise this, there are 3 cables set at the bottom and 3 cables just under the water surface. The results from this multiple cable experiment are given in section 9.4.

During experiment 1, 2 and 4 the tank was filled completely with fresh water and salt water was injected upstream. Experiment 3 started with a mix of fresh and salt water in the basin. The experiment started when salt water was injected at the “downstream” side of the basin (Figure 4). Because of the fast salt water injection in this experiment, the measurement scheme was

adjusted. In both situations the saline water flowed along the bottom of the tank, due to the higher density relative to fresh water.

(25)

11203677-007-ZKS-0002, 20 February 2020

7 Calibration of resistivity

In section 5.1 it was noted that the vertical resolution in the experiments is limited, because of the electrode spacing and limited water depth in the experimental setup (0.4m). To allow a higher vertical resolution, the results were calibrated with a nearby sensor that was installed near the ERT cables (see Figure 8). The conductivity sensor, called VEZO02, continuously measured the electrical conductivity and temperature over the whole water column in 5 cm steps, every few seconds (Figure 9, left panel). The conductivity is the inverse of the resistivity. It is therefore possible to calibrate the inversion results with the conductivity results of the water column. The temperature is more or less constant (Figure 9, right panel), varying between 18.4 and 18.9 °C. Because of the relatively small influence of temperature on resistivity (see Section 5.1), there were no corrections made for the temperature profile; an average temperature per time stamp was applied.

Inversion is an iterative process that seeks the model that best fits the data. The starting model is usually chosen to be a smooth model in all directions. In the cells where the conductivity sensor is situated, actual measurement values are known. These values were used to constrain the inversion model result, meaning that the model cells are forced to have the measured values. Without the conductivity meter, the vertical resolution was ~0.2 m (half electrode spacing). The conductivity meter records the conductivity every 5 cm along the depth profile. Therefore, the model cell thickness in the entire inversion model could be improved from 0.2 to 0.05 m using the measured conductivity data. In addition, the saline water flows in layers. Therefore, larger variations in resistivity in the vertical direction were allowed than in the horizontal direction.

Figure 8: Position of the ERT cable and location of the electrical conductivity sensor used for calibration of the resistivity measurement.

(26)

11203677-007-ZKS-0002, 20 February 2020

Figure 9: Depth profiles of resistivity (left) and temperature (right) at eight depths for 10 time steps, example from 19 June 2019.

(27)

11203677-007-ZKS-0002, 20 February 2020

8 Conversion of resistivity to density

The ERT method provides measurements of resistivity. Depending on the specific application, these can be converted to conductivity or density. For information about salt fluxes or the research question of the IJmuiden project on “Selectieve onttrekking”, the conversion to density is relevant. Therefore, the ERT inversion results were converted to density for all four experiments using the formulas given in Equation 8.1 (Labrique, 1964) and Equation 8.2 (Godefroy, 1981):

𝑆(𝑇, 𝜎) = (_{2.134×((8.018×10}₋₃𝜎_𝑇+1.0609)₂_−0.5911)) 1 0.92 Equation 8.1 and 𝜌(𝑇, 𝑆) = 999.904 + 4.8292 × 10−2_{𝑇 − 7.2312 × 10}−3_𝑇2_{+ 2.9963 × 10}−5_𝑇3_{+ 0.76427𝑆 −} 3.1490 × 10−3_{𝑆𝑇 + 3.1273 × 10}−5_𝑆𝑇2 _{Equation 8.2}

where S(T, σ) = Salinity, ρ(T,S)= density, σ= conductivity, T= temperature. Equation 8.2 is based on Unesco (1981), but adjusted to fit laboratory conditions by Delft Hydraulics.

The temperature at the location of the conductivity sensor was fairly constant over the depth of the profile (Figure 9, right panel). Therefore, only one average temperature value per time stamp was used in the conversion. The considerations about resolution etc. described in Chapter 5, are valid for density as well.

By using the calibration results (Chapter 7) followed by the conversion, it is possible to represent the initial resistivity measurements in terms of water density or salinity. As an example, the measured resistivity with the three bottom cables (configuration 2) is presented in Figure 10 (top). Using the conversion method, the corresponding density result is presented in Figure 10 (bottom). In both cases, the interface between the fresh and saline water can be clearly delineated. The full description of the interpretation for this case is included in section 9.4. All results are presented in Chapter 9 as density results with the exception of case 3, because this was an extra dataset.

(28)

11203677-007-ZKS-0002, 20 February 2020

Figure 10: Representation of the an ERT result expressed as resistivity (top) and as density (bottom). The example is from the experiment using three bottom cables (configuration 2), time stamp 6.

(29)

11203677-007-ZKS-0002, 20 February 2020

9 ERT results

In this chapter the results of the four measurement configurations are described. For each set up, a full cycle of measurements was collected. The data was inverted using a time-lapse algorithm (Karaoulis et al., 2011). This is an algorithm that processes all data from all time-steps in one go, rather than inverting all time stamps separately and consecutively. The appropriate time and space identifiers are assigned increase uniqueness of the models. In this way, both time‐related noise in the data (assumed to be random) and the noise associated with electrode positioning is addressed. In addition, prior information can be incorporated such as the calibration by the conductivity meter into the inversion scheme. (Karaoulis et al, 2011).

In the following sections, a selection of results at different times are shown for each experiment. Four movies containing all time stamps have been made available separately showing the complete interpreted dataset for each experiment.

9.1 Experiment 1: One bottom cable (2D)

Due to expected limitations at the intended test location in ARK, the focus of RWS in this project was on bottom ERT cables. Experiment 1 consists of one bottom cable only. The results in this subsection use a selection of 80 from the total of the 672 electrode combinations, from one cable.

The data was inverted using the time-lapse algorithm. The data RMS is a quantification showing the misfit between the measured data and the calculated (model) data (i.e. the quality of the produced model). In this case, the data RMS error was 6.8%. Typically, data RMS less than 10% represents a very high-quality of the inversion model result.

In Figure 11 the density at three selected times stamps of the experiment is shown. For each time stamp, results are shown for a vertical cross-section along the length of the cable. Between t = 0 and t = 4 no change in the density occurs, because salt water was not yet added. At t=5, a layer of saline water can be seen at the bottom. The intrusion of salt water can be followed by the changes occurring between t = 5 and t = 11. From t = 12 on, almost no more changes occur. The density increase occurs only at the bottom. This is expected, because the salt water is denser than fresh water and therefore moves along the bottom. With constraints in the processing it is possible to identify vertical direction: the salt water layer increased in thickness to the bottom 10 cm of the tank. The complete set of slices through the inversion result at all time stamps is included in appendix C. A movie of the inversion result at all time stamps is available.

The selection of electrode configurations in the one bottom cable experiment represents an array configuration that is not sufficiently dense to image in great detail. A positive aspect of this array is that it can be acquired very fast: hypothetically it will take about 2 minutes to measure all the electrode combinations. Although the array configuration was not optimized for data acquisition by one bottom cable only, it is still able to image the salt water interface.

In general, current flows in 3 dimensions while the data collected with one single cable are 2D. This means that the setup is sensitive only to changes occurring in the strike direction of the cable. There is no way to identify the direction of flow with this single cable setup. Salt water originating from the “upper mid part” or “lower mid part” (Figure 11) would have an identical response. Naturally, the same applies if changes are originating directly above the cable. In an uncontrolled processing, all three cases are identical. The results will show the change in the mid part but no information from the direction. However, in the ARK location this will not be an issue: the origin of salt intrusion is well known.

(30)

11203677-007-ZKS-0002, 20 February 2020

Figure 11: ERT results expressed as density from the experiment with a single bottom cable. Results are shown for a vertical cross-section along the length of the cable. Measurements are presented at selected time stamps (t=4, t=5, t=11). All time stamps and are included in appendix C. In general, each time stamp

(31)

11203677-007-ZKS-0002, 20 February 2020 Extra: Single bottom field test

Several years ago, Deltares conducted a field experiment about the monitoring of salt water intrusion. The results of this experiment are mentioned here, because the experiment has many similarities with the intended future field test location in ARK. In the field situation of the former experiment, salt water was leaking underneath a dam. The same cable was used in this field test as for the experiments in the laboratory. The 50-meter cable was fixed to the bottom of the water body with concrete blocks at a depth of approximately 6 meters. Continuous measurements (with remote control) were taken for several weeks to be able to fully understand the dynamics of the system.

Two time-laps images of resistivity are presented in Figure 12. Each image shows a vertical cross-section of the water body behind the dam. The differences in resistivity, especially between distances 10 and 15 m show the dynamic changes of the fresh/sea water system. In this case, data acquisition took about 20 minutes per full cycle. When applied to the ARK, this could be reduced to 10 or 5 minutes.

Figure 12: Example of salt intrusion monitoring in the field. Resistivity results between 13:15 (top figure) and 23:15 (bottom figure). The red line indicates automatic interpretation of transition zone between fresh and brackish/salt water. A dynamic change of the fresh/sea water system between distances 10 and 15 m is observed.

(32)

11203677-007-ZKS-0002, 20 February 2020

9.2 Experiment 2: Multiple bottom cables (3D)

In this example, data from the three parallel bottom cables was used. The data set consisted of 240 electrode combinations, including combinations from cable to cable using cross-hole (cable-to-cable) configurations. The data could therefore be processed to provide a 3D volume of the resistivity. Using the conductivity sensors as calibration and conversion with density, the fresh/saltwater interface was imaged. In this three-cable bottom setup, the data RMS error was 7.7%. As mentioned earlier, this number represents to a very high-quality of the inversion model result.

Figure 13 shows the density images at three selected time stamps. For each time stamp, results are shown for a bottom layer and a cross-section in the length direction of the salinity intrusion. The fresh-salt water interface can be imaged accurately, and the origin of the salt water and the direction of movement can be identified. From time stamp 0 to 4 nothing happens, because salt water was not yet injected. Between time stamp 5 and 14 the saline water entered from the upstream direction. This can be observed by the changes in density. After time stamp 15 the situation becomes stable. Therefore, time stamps 4, 5 and 14 were selected for Figure 13. The resistivity and the density model results comprise a 3D data cube. Any user-selected cross-section through this data cube can be visualised. One option is chosen for Figure 13. The complete set of slices through the inversion result at all time stamps is included in appendix D. A movie of the inversion result at all time stamps is available.

The direction of salt intrusion is indicated by the arrow in Figure 13. The bathymetry of the tank was not flat but inclined, and thus some preferentially flow paths can be identified. In addition, the very shallow water layer of 40 cm in combination with the electrode spacing of 50 cm results in some artefacts of the processing. An example of a processing artefact can be seen in a few pixels in the edge, in Figure 13. In a field situation, when the resolution and electrode spacing are optimized for the monitoring, this will not be a problem.

In a real-world scenario data, where the dynamics of an area are well known due to the resolution and the quantity of data of a monitoring period, the inversion algorithm can be optimized. It is expected that the data RMS can be reduced to below 5%.

(33)

11203677-007-ZKS-0002, 20 February 2020

Figure 13: ERT results expressed as density from the experiment with three bottom cables, at selected time stamps. Results are shown for a bottom layer and a vertical cross-section along the length of the cable. In general, each time stamp represents 10 minutes of elapsed time. The top layers are removed from the figure to show what is happening in the bottom layer. Any cross-section through this 3D volume can be visualised.

(34)

11203677-007-ZKS-0002, 20 February 2020

9.3 Experiment 3: One double cable (surface & bottom, 2D)

In this test, measurement were made along one vertical plane using one bottom cable and the surface cable above. This experiment was not explicitly planned beforehand, but the opportunity arose when IJmuiden project decided to add this experiment. Since the ERT setup was still present in the scale model, it represented an opportunity to acquire an extra data set which could be used to provide some validation of the single cable experiment. This data set was not fully processed, calibrated and converted to density. Results of the relative changes of resistivity over time are shown. This means that values less than 1 indicate a decrease in resistivity and values greater than 1 represent a higher resistivity.

The inversion results are shown in Figure 14. From time 0 to 15 minutes nothing happens,

because salt water was not yet injected. Starting form 18 minutes until about 30 minutes the saline water was injected from the gate towards the “upstream” side (Figure 4) and this can be observed by the relative changes in resistivity. After about 30 minutes a stable situation has been created. The complete set of slices through the inversion result at all time stamps is included in appendix E. A movie of the inversion result at all time stamps is available.

The situation with one bottom cable only and one double cable cannot be compared one-to-one, because the experiments were performed on different days with different laboratory settings (injection point of salt, speed of salt intrusion).

(35)

11203677-007-ZKS-0002, 20 February 2020

Figure 14: Relative changes of resistivity over time with one double cable (one at the bottom and one vertically above that at the surface) at selected time stamps. In general, each time stamp represents 3 minutes of elapsed time.

salt

(36)

11203677-007-ZKS-0002, 20 February 2020

9.4 Experiment 4: Multiple double cables (3 surface & 3 bottom, 3D)

The setup in this experiment consisted of three bottom cables and three surface cables. One measurement cycle consisted of 672 electrode combinations, resulting in the largest dataset in this project. The data were inverted using a 3D model. The measurement took place in the downstream side of the tank, while saline water was injected in the upstream side, from the bottom. A full cycle of measurements was collected every 10 minutes.

For this dataset, results for both the resistivity and density are shown in cross-sections, for different time stamps. Figure 15 presents a selection of time stamps with the original resistivity data, while Figure 16 shows the converted results for density. The resistivity and the density model results comprise a 3D data cube. Any user-selected cross-section through this data cube can be visualised.

The first 5 time stamps were used for calibration, because it represented the stable start situation. The infiltration of salt water started at t=5. After t=6, the saline water originating from the upstream is observed and it gradually spreads in the bottom 10 cm of the tank (see t =11). This is apparent from the lower resistivity shown in blue colours. Because of the 3D nature of the measurements and inversion results, the direction of the salt water intrusion could be distinguished.

The complete set of slices through the inversion result at all time stamps is included in appendix F (resistivity) and G (density). A movie of the inversion result at all time stamps is available.

The results from the experiment with multiple bottom cables (experiment 2) and multiple double cables (experiment 4) can be compared, because they were measured during the same

experiment and represent different subsets of the measured data. Vertical cross-sections for both subsets for time stamp 14 are shown in Figure 17. The images of multiple bottom cables and multiple double cables are very similar. Especially the transition between the fresh and the salt water is identical. The largest differences are found in the top layer (Figure 17, bottom panel), which is not relevant imaging of the salt. Another zone of differences is located on the left side of the model, probably as a result of the boundary of the model.

The histogram of the ratio for all time stamps and all model cells is shown in Figure 18. On average, over all time stamps and all grid cells, the ratio is 1.04 ± 0.3.

(37)

11203677-007-ZKS-0002, 20 February 2020

Figure 15: Resistivity result of multiple double cables (three bottom cables and three surface cables); sets of measurements at selected time stamps. In general, each time stamp represents 10 minutes of elapsed time.

(38)

11203677-007-ZKS-0002, 20 February 2020

Figure 16: Density result of multiple double cables (three bottom cables and three surface cables); sets of measurements at selected time stamps. In general, each time stamp represents 10 minutes of elapsed time.

(39)

11203677-007-ZKS-0002, 20 February 2020

Figure 17: Comparison between the resistivity inversion result for multiple bottom cables (top panel, experiment 2) and multiple double cables (middle panel, experiment 4). A cross-section is selected in the length direction of the cables, for time stamp 14. The bottom panel shows the ratio between multiple double cables over multiple bottom cables.

Figure 18: Histogram of the ratio between the multiple double cables resistivity and the multiple bottom cables resistivity result for all time stamps and all model cells. The average ratio is 1.04 ± 0.3.

(40)

11203677-007-ZKS-0002, 20 February 2020

10 Conclusions

In this study, laboratory experiments were conducted to test if the ERT technique could potentially be applied in a field setting for measuring salinity intrusion in inland waterways and to demonstrate a feasible monitoring set-up to do so. These laboratory experiments were conducted in

cooperation with another ongoing laboratory test using the IJmuiden scale model, constructed in the Hydrohal facility of Deltares. Combining the ERT laboratory tests with the study in the

IJmuiden scale model provided many advantages and considerable time and costs savings. Most notably, with ERT techniques, true 3D images (or 4D in monitoring setup) could be made that are based on data, rather than combining multiple 1D measurements. Specifically, it made it possible to conduct experiments using saline water, which require significant preparation and special permits.

Four different experiments were conducted, each with a different configuration of ERT cables: 1. One bottom cable (2D)

2. Multiple bottom cables (3D)

3. One double cable (bottom & surface, 2D) 4. Multiple double cables (bottom & surface, 3D)

1.One bottom cable (2D)

The single ERT cable on the bottom shows the growing of a saline layer starting from the bottom. A single cable however, cannot identify the horizontal direction of the saline water flow.

2. Multiple bottom cables (3D)

In the setup with three parallel bottom cables with electrode combinations from all bottom cables, the data was processed in 3D. The direction of the salt water intrusion could be identified in this case: the saline water could be observed to originate from the upstream side and gradually spread over the bottom 10 cm of the tank. This setup provides a fast measurement cycle, with high resolution in all dimensions, while keeping cables away from the surface. The resolution closely matches the setup of multiple bottom and surface cables (experiment 4).

3.One double cable (bottom & surface, 2D)

In this specific test with one bottom and one surface cable, the salt intrusion occurred as a fast process. It was therefore necessary to adjust the measurement cycle. It was possible to observe the saline water as it moved from the gate at the upstream side of the scale model, creating a saline layer at the bottom. In a field application, this setup represents cables installed vertically at the sides of the river and the area of investigation is a cross-section between them (not possible at the intended ARK location due to steel sheet pile walls).

4 Multiple double cables (bottom & surface, 3D)

Although this configuration is not expected to be used in a field setting, it provides the maximum data coverage and the best resolution of results. As such it provides a validation of the results obtained from the other experimental configurations. The infiltration of salt water is observed: the saline water originated from the upstream side and gradually spread over the bottom 10 cm of the tank. Because of the 3D nature of the measurements and inversion results, the direction of the salt water intrusion could be distinguished.

(41)

11203677-007-ZKS-0002, 20 February 2020

Considering the four different experiments, all of the configurations tested showed that it is possible to detect and measure a saline water intrusion into fresh water. A single bottom cable provides a 2D vertical cross section of the water column (in the direction of the cable), illustrating the thickness of the saline layer. With multiple bottom cables, additional data is collected so that a 3D image of the salinity is possible, thus providing information on the direction of saline water movement. The bottom and surface configurations with multiple cables confirm that the bottom-only results are of sufficient quality to be used in a field experiment.

The laboratory experiments showed the salt water moved along the bottom of the basin and this process could be tracked over time. ERT is therefore a suitable tool to map and monitor salt/fresh water movements in the water column. In the laboratory setting, the spacing of the electrodes with respect to the desired resolution was not ideal. Despite this, the salt/fresh water interface and direction was imaged in sufficient detail. In a real field application, the resolution will be optimized with the electrode cable choice. It is expected that the resolution will not be an issue.

Laboratory and field applications, even though they are based on the same physics, have different challenges. In a field application, the resolution issues will be solved, but manmade electrical noise present different challenges. When planning for a field test, special care must be taken to optimize the acquisition setup in order to provide the desired spatial and temporal resolution of information.

ERT measurements of salt intrusion in the laboratory