There are a number of reasons that Aarhus was the perfect location for the finale of the DISC tour in Europe. The HydroGeophysics Group at Aarhus University, led by Esben Auken, has a strong track record in instrumentation as well as software for simulations and inversions, particularly for airborne EM, and more recently for induced polarization and its applications to groundwater. The group has been been instrumental in the development of the SkyTEM system and software for processing and inverting those data. SkyTEM is now a company based in Aarhus. With Aarhus having such a concentration of quality geoscientists, we were very excited for the opportunity for scientific discussion, and especially for DISC lab.
In addition to geophysics… for the days outside of the DISC, Aarhus is the European Capital of Culture for 2017, so a number of events were being held throughout the city.
Day 1: DISC Course
Aarhus had the largest turnout of participants for the DISC in Europe. The crowd included faculty, undergraduate and graduate students from the University; there were a geologists, geostatisticians, and engineers with SkyTEM who attended, as well as a number of attendees in software development with Aarhus GeoSoftware. The problems of interest primarily included environmental applications, particularly for groundwater; there was also interest in geotechnical applications and pre-construction surveys and some interest in resource exploration (eg. minerals and oil and gas).
Most of the participants who attended in Aarhus have some direct connection with EM, either developing instrumentation, working with software, or working with data to answer a geoscientific questions. The level of background in EM of this group is high, and this was clearly reflected in the quality of questions and discussion we had throughout the day. Covering the spectrum of EM methods in a single day makes for a long day; we were impressed that the participants remained engaged throughout the entire time!
From the participants
“A very elaborate overview of EM geophysics with some valuable insights and case studies” — PhD student in geophysics / geostatistics
“Very good and thorough introduction to the basics of EM. Good case histories. Inspiring to see the web options and open-source apps.”
“It gave me a deeper understanding of the different methods and inspiration to what I can do my bachelors in” — Anne, student
“Interesting from an academic point of view. It was nice to have some fundamentals recapped in an illustrative way. [The] graphical representations of currents were high quality and helpful in understanding these complex phenomena” — Thomas Steensen, Geophysicist with SkyTEM
All of the course material for the day can be downloaded at https://disc2017.geosci.xyz/aarhus.
Day 2: DISC Lab
DISC Lab day began with discussion on a couple topics we did not have time to dive into on day 1. Induced polarization, particularly in its application for hydrogeology and estimating hydraulic permeability, is a topic of active research in Aarhus. The first presentation we gave was on EM decoupling, which included an example of extracting IP from airborne EM data over a kimberlite deposit in Northern Canada (TKC).There are several approaches that can be taken for inverting for IP parameters from electromagnetic data. In this presentation, we discussed one approach, where a two-step procedure is used. The EM-dominated portion of the data is inverted for electrical conductivity. The response due to EM induction is then calculated for this model and subtracted from the data, leaving the response due to chargeable features. This residual response is then inverted to obtain a chargeability model. An alternate approach is to invert both responses at once for a complex conductivity model, typically in 1D, which helps reduce the non-uniqueness. The latter approach is applied by the Aarhus group, and we were keen to connect on the topic and compare notes. Gianluca gave us an overview during his DISC Lab presentation (more below!).
The second topic we discussed was the application of IP for delineating and monitoring landfills (from the IP slides); the case history we showed is a study conducted by Gazoty et al., 2012 for a landfill in Denmark.
Gianluca Fiandaca (AEM-IP slides, IP Parametrization slides, Hydraulic Permeability + IP slides), a professor at Aarhus University, started us off. He showed 3 presentations, all related to IP. His first presentation discussed 1D laterally constrained inversions (LCI) of airborne IP data where resistivity and IP parameters (there are three IP parameters when using a cole-cole parameterization) are inverted in an all-at-once approach. The inversion inverts each sounding individually for a layered model, but a regularization which promotes laterally smooth variation in geologic structures is applied. Inverting for resistivity and IP parameters all at once increases the non-uniqueness of the problem, but some of that is counteracted by reducing the dimensionality of the problem from 3D to 1D. Gianluca introduced a “robust” inversion scheme which includes implementation updates such as choosing a different parameterization for complex conductivity, including a floor in the noise model so that zeros in the data have appropriate error bars, and only allowing the 2 parameters with the largest impact on the data to change in the first iterations. On his final slide, he drives home the point that the choice of parameterization matters; if there are “equivalences” due to the parameterization, that makes obtaining a good model from the inversion more challenging (if we have a model consisting of 4 parameters, and two of them could be updated independently and produce a similar effect in the data, then they are “equivalent”). We saw this in the MT tutorial with the equivalence of conductance; the response due to a thin, conductive layer is approximately the same as if we were to double the thickness of the layer and halve the conductivity). If you can choose a parameterization that reduces these equivalences, then you are in better shape. In his second presentation, Gianluca, dives further into the new parameterization he is considering for complex conductivity, not only to reduce equivalences in the inversion, but to also be closer to the set of parameters quantified in lab measurements and those used by hydrogeophysicists to estimate permeability (which is the topic of his final presentation!).
Hydraulic permeability describes how pores are interconnected and it is in an important component in describing how easy it is for a fluid to flow through a rock. To model groundwater flow, hydraulic permeability must be known. Typically pump tests, slug tests, or grain size analysis would be used to obtain hydraulic permeability, but this requires boreholes, and the information obtained is limited. Maybe geophysics could be used!
Previous studies have shown that there is a correlation between the IP parameters of a rock sample in the lab and its hydraulic permeability. In this study, a well logging experiment was conducted near the Grindsted Landfill in Denmark to estimate permeability of the in-situ rocks. Three boreholes were drilled and DC resistivity and time-domain induced polarization logs were collected while drilling. In addition, grain-size analysis on 58 samples and 9 slug tests were conducted. Hydraulic permeability estimates were obtained from: (i) the IP data, (ii) the grain size analysis, and (iii) the slug test. In each of the 3 well logs, all three estimates of permeability were reasonably consistent with each other. The estimate from IP followed the trend and picked out the regions of maximum permeability compared to the grain size analysis and the slug tests. The goal with IP is to estimate permeability within an order of magnitude — this is currently the best that can be achieved in the lab, so obtaining similar results in the field is a good result. Following the borehole tests, they conducted a DCIP survey from the surface and compared the estimates of permeability to several boreholes in the region. Overall, there was good agreement between the borehole measurements and the conductivity and hydraulic permeability inverted from the DCIP data. In the inversion results, there was a region having unusually high electrical conductivity just outside of where the boreholes had been drilled; Gianluca and team suggested that this might be a polluted region… initially the contractors were not convinced (now they are!). Several boreholes were drilled in that region, and the water that came out was indeed polluted. Gianluca’s take home point is that hydraulic permeability can be mapped through IP, with the limitations that the media be saturated and unconsolidated and there there are no other significant sources of IP present (oil spills, mineralizations, etc).
Ingelise Møller (slides), with the Geologic Survey of Denmark, presentated a case study where a 3D geologic model of a contaminated stream site was built using geologic information from boreholes, chemical data and geophysical data. The site of interest in the study is near the town of Grindsted. Contaminants from a chemical factory east of the town were dumped and have since discharged into the ground. There is a clay layer 80m below the surface which is acting as a confining layer and keeping the contaminants in the overlying aquifers. Contaminants have been found in the stream that flows through the town. Multiple boreholes have been drilled, revealing that the groundwater flow system is complex: there are two aquifers, an upper and a lower aquifer that have been observed to each be contaminated with different chemical contaminants. Are there multiple sources of contamination (from the landfill and possibly somewhere else?)? In order to figure out where the contamination is coming from, a flow and transport model is needed, which requires that we have a 3D geologic model of the setting.
Several data sets have been collected over the site, including
- borehole data: lithological logs, hydraulic head measurements, and electrical conductivity measurements from water samples
- geophysical data: loop-loop electromagnetic induction, DC resistivity, time domain spectral IP
We will focus on the DC and IP data. The DC and IP data were inverted using the approach Gianluca discussed earlier in the day. In order to understand how the physical properties (resistivity and chargeability) relate to the geology, Ingelise used the borehole data to generate cross-plots and obtain the expected ranges of physical properties (resistivity and chargeability) for each of the lithologies in the area. For example, mica clays have the lowest resistivity and highest chargeability of the lithologies in the area, and will be more conductive than contaminated sands (expected to have a resistivity of ~100 Ωm). By examining the recovered resistivity and chargeability models together, clay, sand and till units were interpreted and a 3D geologic model built. The next step is to use the geologic model as an input to the flow model to model the groundwater and contaminant transport.
Soeren Rasmussen (slides), a PhD student at Aarhus University, presented work he has been doing in signal processing for time domain electromagnetic data — specifically looking at VLF (very low frequency) noise. VLF transmitters are positioned all around the world and they transmit powerful EM signals for military communication. VLF signals are well structured, binary signals, that is, they encode patterns of 1’s and 0’s that contain the radio transmission. The value of “0” is communicated by one frequency, say 200kHz, and the value of “1” is communicated by another, say 100kHz.
These signal are considered noise in a controlled source EM experiment — they are not due to the transmitter or the response of the earth. They can make a significant impact on early-time data, where the time-gates over which the signal is stacked are small (at later times, the time gates are wider, effectively acting as a low-pass filter). Linear filters are challenging to apply as they can have unintended consequences and amplify the impact of other noise sources such as spherics from lightning strikes. In his presentation, Soren introduces a non-linear approach, where the radio signal is modelled and subtracted from the data.
The first step is to try to isolate the VLF signal measured by the receivers (eg. we want to remove the signal due to the transmitter). It is typically easier to obtain a strong VLF signal in the x or y data than in the z receiver as the horizontal measurements are better coupled to the transmitter. But… the signal that we really care about in our EM survey is typically in the z-component, this is where the strongest signals due to the earth typically are. So in order to get the cleanest radio signal possible, Soren is combining data from both horizontal and vertical receivers. Once a relatively clean signal is isolated, it can be decoded into the bit sequence of zeros and ones. Isolating and decoding a signal achieves the goal of a radio transmission, but we need to go further and model how it shows up in the time domain EM data. Starting from the bit sequence, the radio-transmission signal is reconstructed. Knowing the location and orientation of receivers, the expected VLF signal measured by each receivers is estimated and it can be subtracted from the measured TEM data. The challenging aspect is not the signal decoding, but properly accounting for tuning parameters in the signal processing. Soren showed a small field example that he set up to test the approach and demonstrated that at early times, removing the VLF signal significantly reduced the noise level of the TEM data.
Maria Riis-Mygind and Anna Bondo Medhus (slides), with the engineering company COWI, showed four examples of the applications of geoelectrical (DC resistivity) and ground penetrating radar (GPR). The first example they showed was along a railway line that is planned to be upgraded. An extra, electrified track needs to be laid, and they want to avoid laying tracks on soft soils that might subside. Approximately 25km of DC resistivity data were collected along the tracks and several boreholes drilled. The soft soils, which are more electrically conductive than more competent soils, were mapped with the inverted DC data. The mapped soft soils could then be dug away and replaced with more competent material before the tracks were laid.
The second example they showed was to map the geology below a river for a horizontal, directional drilling project. To assess the risks and plan the drill path, an understanding of the sand, gravel and clay units beneath the river was needed. They conducted a DC resistivity survey, with 3 lines crossing the river and 1 line parallel to the river and combined these data with drillhole lithology data to produce a geologic map.
The final two examples both involved geologic mapping. One was a project mapping gravel (which is resistive compared to the surrounding sands) in Denmark using DC resistivity. The final example was aimed at mapping the depth to bedrock in Sweden using both DC resistivity and GPR.
Anders V. Christiansen, a professor at Aarhus University, gave the final presentation of the day on a new instrument (tTEM) for high resolution EM imaging of the shallow subsurface. The motivation for developing the tTEM system is to fill a gap in the current EM instrumentation for imaging the shallow subsurface over large areas. Right now, if you want to cover a large area, an airborne system, such as SkyTEM, would be an option. However, this has a large footprint, so the resolution is not very high. For higher resolution, one might consider a DC resistivity survey, but this is very labour intensive, so covering a large area is a challenge. The tTEM system is aimed at imaging the top ~30m in reasonably quick acquisition times.
The application of interest is land management and understanding nitrate retention. If one could map flow paths subsurface, an assessment could be made on which fields are higher risk for loading nitrates into streams and groundwater.
The tTEM system is towed by an ATV and can operate at ~20 km/h. At this acquisition speed, 1 stacked sounding will be collected every 1m or so. Anders discussed the design of the system and pointed out that a lot of experimentation was necessary in order to decide where the GPS should be, where wires are running, what the distance between the transmitter should be, … The result is that every part of the system has a deliberate design.
To get a sense of the “footprint” of the system, Anders showed the results of a study which compared the footprints of the SkyTEM, tTEM and a loop-loop EM system; the tTEM has a much smaller lateral footprint than the SkyTEM indicating that the lateral spatial resolution is higher.
In the last part of his presentation, Anders demonstrated a case study from Javngyde, Denmark, an agricultural region just outside of Aarhus. They covered an 1000 hectare area in 11 days. In the inverted models from these data, they were able to delineate interfaces between sand and clay layers that might indicate thrusting due to glacial tectonics (lower lying clays appear to be thrust over sand layers). It was also possible to identify individual clay lenses. These are preliminary interpretations and might be followed up with drilling.
We wrapped up the day with an introduction to the suite of open source resources available in GeoSci.xyz and how to contribute. This led into a discussion on the potentially significant role that open-source resources, including software and scientific content, could play in elevating educational levels in geophysics.
“It (EM GeoSci) is also an amazing resource for [where], in many parts of the world, the universities that you go to don’t really present material of this quality to the students. When you see the questions asked on LinkedIn or other forums, you get surprised about the depth of ignorance; and to be able to have a resource like that would be an incredible help for a lot of students in not so well off countries, where educational standards are not as high as here, to learn something the way it was meant to be taught.” — Niels Christensen
Following DISC Lab, we met Anders, Esben, and Gianluca for dinner at Juliette, a French restaurant (where Esben’s daughter works!).
A few adventures
On our first day in Aarhus, we went for a walk through town (with the destination being a Bikram yoga studio!). We came across a raised platform jutting out over the street and decided to check it out. Being five floors above street level, we could see out to the town square and just beyond it to the North Sea.
After working up an appetite at Bikram, we went for dinner at the Street Food market. Inside the market, the walls were lined with stalls, serving everything from duck fat fries to specialty ice cream. We saw the line at the pita stall, and being so popular, we figured it would likely be a good choice (it was!). The pita bread was made fresh on-sight, then loaded with fresh veggies and topped off with some marinated steak. We also had some Jerk Pork from a stand serving Jamaican food, and topped it off with a local beer.
On our final day in Aarhus (and our final day of the DISC tour in Europe!) we did a bit of sight-seeing and went to two of the major museums in Aarhus. We started off in the morning at the Aros Museum. The building is quite iconic in Aarhus; on the top there is a circular walkway enclosed by colored glass. The whole walk takes you around the rainbow.
In the afternoon, we met up with Niels Christensen and his wife Juliette. Niels showed us some of his Time-Domain EM inspired creations and we toured the Moesgaard museum together. Moesgaard is home to the Grauballe Man, a body from the late 3rd century BC that was preserved in a bog in Denmark. We went for a walk on the grass-covered roof before heading back to town to meet up with Esben for dinner.
Esben and his wife invited us to their home for dinner for our last evening in Aarhus. We shared some pasta, wine, and delicious cheeses. It was the perfect end to the European leg of the DISC.
Esben and Anders made a tremendous effort to gather a significant audience, reaching students, academics and industry professionals to make the DISC in Aarhus a highlight of the European trip! We are grateful to them for their enthusiasm and for the time they spent with us in Aarhus.