Taming the complexities of contaminated sites

This webinar covers:

• Bringing teams and diverse technical expertise together to see the full picture and build a collaborative solution.
• Discovering where the contamination lies – in groundwater or soil – and defining how it affects your project.
• Maintaining a clear, data-driven plume interpretation that will stand up to interrogation.
• Making data work harder and smarter, and use powerful 3D modelling to tell its story clearly and concisely.
• Arriving at a solution that is effective, transparent and defensible; one that can react quickly to the latest updated information, and allow for informed engineering decisions.
• Communicating your decisions in a way that makes client and local communities feel confident in the direction you’re taking.
• Providing reports that will satisfy regulators and help your client meet legal requirements with confidence.

Video Transcript

[00:00:04.563]<v Tim>Good morning, good afternoon, good evening</v>

[00:00:07.430]wherever you’re joining us from today.

[00:00:09.356]This is Taming the Complexities of Contaminated Sites.

[00:00:13.693]This is a webinar brought to you by Seequent

[00:00:16.847]and brought to you by Matrix Solutions.

[00:00:21.974]Today we will be discussing how to build

[00:00:25.179]robust contaminated site models

[00:00:27.722]which can give you a solution that’s safer,

[00:00:30.760]a solution that’s faster, more certain,

[00:00:33.700]and easier to defend.

[00:00:36.545]My name is Tim.

[00:00:37.978]I am a solutions architect at Seequent.

[00:00:41.670]And it’s my pleasure to be hosting this session

[00:00:44.341]along with our two experts today,

[00:00:46.847]and bringing your questions to them.

[00:00:51.329]So let’s get to know these experts a little more.

[00:00:54.282]And it would be remiss of me not to introduce myself first.

[00:00:59.930]So I’ll start there, my name is Tim and I work at Seequent

[00:01:04.556]on New Product Development.

[00:01:08.010]In the past, I led Leapfrog Geo to market in 2013,

[00:01:12.270]and then in 2016 I did the same with Seequent Central.

[00:01:18.260]I’ve got quite a passion

[00:01:19.400]for geospatial and geological products,

[00:01:22.040]but today I’m really excited

[00:01:24.712]to be talking about the challenge

[00:01:26.540]of complex contaminated sites and how we deal with them.

[00:01:32.480]So let’s look at our experts.

[00:01:37.000]First of all, I want to introduce Dr. Amir Niazi,

[00:01:41.625]and he is a hydrogeologist and a groundwater modeler

[00:01:45.900]at Matrix Solutions.

[00:01:47.846]Amir is an experienced hydrogeologist

[00:01:51.111]with a history of working

[00:01:53.390]in the environmental services industry.

[00:01:55.907]And for the past 17 years,

[00:01:57.810]he has applied and expanded his skills

[00:02:00.285]with regard to local and international water issues.

[00:02:04.231]He has developed a range of numerical

[00:02:06.810]and conceptual groundwater models

[00:02:09.074]and for contaminated sites

[00:02:11.037]using his knowledge of hydrogeology, of geostats,

[00:02:14.664]single and multi-phase flow modeling

[00:02:17.640]and porous media, and surface water modeling.

[00:02:21.224]And Amir holds a PhD in the Geological

[00:02:25.180]and Earth Sciences, in the geosciences.

[00:02:28.080]So welcome along, Amir.

[00:02:29.483]It’s great to have you with us today.

[00:02:32.021]You’re joining us from Calgary?

[00:02:34.127]<v Amir>Correct, yeah, thank you for that.</v>

[00:02:35.817]<v Tim>And I understand,</v>

[00:02:37.080]oh, you’re welcome, you’re welcome.

[00:02:38.370]I understand the weather is just about to change

[00:02:40.860]and you’re expecting winter to come

[00:02:44.190]with a thump pretty soon.

[00:02:45.921]<v Amir>Yeah, yeah, I think so.</v>

[00:02:48.550]I think it’s kind of overdue, the winter. (laughs)

[00:02:51.297]<v Tim>(laughs) It’s inevitable, it will come.</v>

[00:02:54.710]And the kids are just going back to school in Calgary now?

[00:02:59.630]<v Amir>Yeah, yeah, just started last week.</v>

[00:03:02.059]And, yeah, I think it’s a process,

[00:03:06.561]kids are going back to school.

[00:03:09.377]I think it’s fully opened today or yesterday,

[00:03:14.180]and all the kids are back to school at least–

[00:03:17.790]<v Tim>So a sense of normality returning there?</v>

[00:03:20.300]<v Amir>Yeah.</v>

[00:03:22.290]<v Tim>Well, it’s great to have you with us</v>

[00:03:23.820]and thanks for your time today.

[00:03:27.729]Next of all, I want to introduce Dr. Tom Krom.

[00:03:31.823]And Tom is a segment director currently

[00:03:35.120]for the Environmental Segment at Seequent.

[00:03:38.097]Tom has been involved in the development

[00:03:40.570]of Leapfrog since 2007.

[00:03:42.800]And now he spearheads the strategic direction for Seequent

[00:03:46.910]on the environmental and sustainability areas.

[00:03:50.501]Tom has more than 30 years experience

[00:03:53.580]in the geological sciences

[00:03:55.088]and he’s gathered that all around the world.

[00:03:58.285]Tom is a skilled groundwater modeler.

[00:04:01.424]He’s involved in a range of consulting

[00:04:03.390]and research projects.

[00:04:04.688]Tom holds a PhD in Civil Engineering

[00:04:07.530]and a master of science in Geological Engineering

[00:04:11.024]with both qualifications involving significant research

[00:04:15.220]into methods which describe

[00:04:17.540]and optimize geology statistically.

[00:04:20.848]Tom, it’s great to have you with us.

[00:04:22.597]You’re joining us from Denmark today.

[00:04:25.413]And
<v ->I am.</v>

[00:04:26.610]<v Tim>How is the weather in Denmark?</v>

[00:04:28.848]<v Tom>The weather in Denmark is not bad actually.</v>

[00:04:32.220]At the moment it’s very dark weather, but, yeah, it’s cool,

[00:04:38.992]10 degrees, not raining at the moment

[00:04:42.100]though it rained most of the day.

[00:04:43.835]So we’ve definitely gone from summer to autumn here.

[00:04:48.869]<v Tim>It’s clicked in here at Claris.</v>

[00:04:51.690]The winter is coming.

[00:04:53.755]Hey, it’s great to have you with us, Tom.

[00:04:55.632]And thank you, thank you so much

[00:04:58.580]for your time in your evening, this evening.

[00:05:01.303]And we are coming to you

[00:05:03.100]from the three corners of the world today.

[00:05:05.260]I’m down in the South Pacific

[00:05:07.339]in New Zealand at Seequent’s headquarters.

[00:05:10.550]So you’ve got three very different times zones today.

[00:05:14.265]And wherever you’re joining us,

[00:05:16.850]it’s great to have you with us and thank you for your time.

[00:05:20.772]So let’s crack into this.

[00:05:22.880]I’ve got basically three, oh, I’ve got five main questions,

[00:05:27.769]which followed how you would approach a contaminated site

[00:05:33.370]starting with the start of a project

[00:05:36.111]right through remediation.

[00:05:38.553]But let’s think about the problem here.

[00:05:42.010]The problem, clients need experts

[00:05:45.540]to help them understand a contaminated site.

[00:05:48.985]There are stressful legal obligations

[00:05:53.420]often associated with these sites.

[00:05:56.089]Clients want to minimize the potential costs.

[00:05:59.585]They only want to spend as much as they should,

[00:06:02.404]and they want to minimize the time

[00:06:05.540]so they can move forward with this site.

[00:06:08.527]And there are other people involved,

[00:06:10.630]lots of people involved,

[00:06:12.079]stakeholders with different levels

[00:06:14.900]of technical understanding.

[00:06:16.060]There are community members, there are contractors,

[00:06:19.290]there are engineers, there are geologists,

[00:06:22.200]often corporate representatives,

[00:06:24.050]legal authorities, regulatory boards.

[00:06:27.894]You need to understand the risks for your client,

[00:06:31.318]for your communities, and for the environment.

[00:06:34.560]But more importantly, you must communicate these risks.

[00:06:39.403]Your recommendations will be based on the data you have

[00:06:42.822]and the cost benefit analysis and how you can make sure

[00:06:47.320]that clients feel confident in the plan that you present

[00:06:50.534]and make good decisions.

[00:06:53.947]Teams need to bring together

[00:06:56.300]their diverse technical experience

[00:06:58.490]and the data to bring the full picture.

[00:07:00.923]And after you make your decisions,

[00:07:03.227]do you feel you can confidently defend them with data?

[00:07:06.681]Can you show why each choice was made?

[00:07:11.163]Technology makes it easier to bring it all together,

[00:07:13.810]all your geoscience data together,

[00:07:15.981]and to tell a complete and clear story.

[00:07:18.082]And with everyone on the same page,

[00:07:20.440]you’ll gain the shared understanding and consensus

[00:07:23.440]you need to minimize time, cost, and risks.

[00:07:26.062]These are tricky sites, so many people involved,

[00:07:30.730]and potentially very legal and litigious.

[00:07:34.872]So let’s start at the start.

[00:07:36.834]And I’ll start with you, Amir.

[00:07:38.850]You’re starting out.

[00:07:39.683]There’s a whole new site.

[00:07:41.026]It’s a new project, it’s a new client.

[00:07:44.269]You might have some preliminary tests

[00:07:48.050]which are pointing to deeper contamination,

[00:07:50.570]but at this early stage of this project,

[00:07:53.540]when so much is unknown

[00:07:55.959]and so many aspects of the site

[00:07:57.870]have the potential to become very large issues,

[00:08:00.732]how do you deal with all the choices,

[00:08:02.972]all this complexity at the start?

[00:08:05.404]How do you deal with all those components?

[00:08:09.369]And what technologies are going to help you avoid the risks

[00:08:12.849]as they cascade further down into your project later on?

[00:08:16.930]Amir, what’s your response to that?

[00:08:18.993]<v Amir>Yeah, As you said,</v>

[00:08:21.084]it’s going to be a really complex problem

[00:08:25.850]that we are going to try to solve that problem.

[00:08:30.330]And starting fresh from a new project,

[00:08:34.320]we know that the problem is complex,

[00:08:36.750]but as you said, usually we don’t have enough information.

[00:08:40.530]And what we are mostly trying to do

[00:08:44.560]is collect as much information as we can

[00:08:47.920]to kind of try to make a model,

[00:08:53.050]a conceptual model in your usually brain too

[00:08:57.570]to kind of try to understand what the problem is,

[00:09:00.914]what is the complexity, what are the moving part,

[00:09:04.880]and what is it that the client wants to achieve.

[00:09:09.595]And then from that, you start to see

[00:09:15.200]that there are some gaps

[00:09:16.732]in the knowledge that you have already,

[00:09:20.440]and you try to fill those gaps with information

[00:09:24.670]that you can collect.

[00:09:27.061]Sometimes you need to go to the field

[00:09:29.440]and collect samples or do drilling to get more information.

[00:09:35.070]And then sometimes you need to talk to your client

[00:09:38.470]and understand the history of the site a little bit better

[00:09:45.070]or the other stakeholder involved.

[00:09:47.771]So then the technologies is, I think,

[00:09:50.860]the role of technologies is kind of critical

[00:09:54.791]as you try to kind of connect the dots

[00:09:58.230]and understand the relationships.

[00:10:01.860]So, as I said, a conceptual model

[00:10:05.810]is going to help you to achieve that

[00:10:09.767]which with technology can help you

[00:10:12.289]easily start from something simple.

[00:10:17.343]And sometimes by just drawing what’s going on can help you.

[00:10:24.108]And then when you collect more information

[00:10:28.070]and then you need to look at the more complex

[00:10:31.841]hydrogeological setting and the site characterization,

[00:10:36.200]then you need to somehow be able to easily

[00:10:41.239]have all of those information in a system

[00:10:45.786]to show your client or kind of collaborate

[00:10:50.610]with your colleagues and show that,

[00:10:52.650]okay, so this is what I think about the site.

[00:10:56.800]So what’s your idea?

[00:11:00.040]<v Tim>It is almost like there’s a mental model building up</v>

[00:11:04.008]in your own mind as you collect data.

[00:11:07.016]And then is that represented as the conceptual model?

[00:11:11.780]Is that the actual tangible thing

[00:11:13.650]you talk about with your client

[00:11:15.590]and other people, other stakeholders?

[00:11:17.660]<v Amir>Yeah, yeah, it start to form in your mind,</v>

[00:11:20.471]but as you move on and then you need to communicate

[00:11:23.674]that picture to your client

[00:11:26.295]and the stakeholders and other peoples involved,

[00:11:29.530]and then you need to move it from your brain

[00:11:32.810]into a piece of paper or a computer model or something,

[00:11:38.629]something like that.

[00:11:40.378]<v Tim>Right, yeah, and Tom, how do you decide</v>

[00:11:43.930]what conceptual model you’re going to need

[00:11:45.740]because there are lots of different conceptual models

[00:11:48.300]and ways to represent it, right?

[00:11:50.295]<v Tom>I think the conceptual model</v>

[00:11:53.580]will vary from site to site and problem to problem.

[00:11:57.207]Generally, you have some experience

[00:12:00.730]or know something about the situation in the area.

[00:12:04.200]You know if it’s in a glacial outwash

[00:12:07.890]or fluvial deposits or whatever.

[00:12:11.760]So you have a starting point

[00:12:15.827]and that can be either confirmed or refuted

[00:12:19.950]by the data that you can get that’s already there.

[00:12:23.959]One of the things I was wondering about is

[00:12:26.670]when you go out to these sites,

[00:12:27.580]you always at some point have to convince the customer

[00:12:33.436]whether to pay or not for additional investigations.

[00:12:38.700]And I’m wondering, Amir,

[00:12:40.660]do you find that technological advances

[00:12:43.580]make it easier to explain the problem to them

[00:12:46.740]and why you need to get more information?

[00:12:50.518]<v Amir>Yeah, I think certainly ’cause most of the time,</v>

[00:12:55.552]for example, in terms of data collection,

[00:13:02.160]for example, I need a borehole in that location.

[00:13:05.330]How you should convince the client

[00:13:08.050]that, okay, I need that borehole.

[00:13:09.820]So I think technology,

[00:13:13.622]doing some stochastic or statistical analysis on the data

[00:13:18.290]or looking at the geospatial distribution

[00:13:20.680]and the relative correlation that you have between the data,

[00:13:29.100]then you can kind of show scientifically

[00:13:32.531]and not just what you feel,

[00:13:35.590]but in terms of data, you’re looking at the data,

[00:13:38.187]and then you say, okay,

[00:13:43.139]based on the information that we have currently,

[00:13:46.230]so this is the separation, the minimum separation

[00:13:51.380]that we need between our data point.

[00:13:53.320]But if you are far beyond that,

[00:13:58.310]so then we need to take into,

[00:14:00.300]and it’s easier to communicate

[00:14:02.380]that to our clients to say, okay,

[00:14:04.002]we need to collect this information from these locations

[00:14:07.920]because science is saying so.

[00:14:13.360]And when you have a model that’s kind of tangible,

[00:14:20.320]the client can see the information and they can understand

[00:14:26.390]what you’re talking about more easily,

[00:14:28.240]and then you can kind of communicate

[00:14:29.630]all the data that you have

[00:14:31.700]in terms of geological information,

[00:14:34.107]in terms of the site itself,

[00:14:40.510]the infrastructure that you have onsite,

[00:14:43.660]and then in terms of the receptors that are close by,

[00:14:46.560]then it’s way easier to communicate that to our clients

[00:14:50.890]and to convince that they need to pay basically

[00:14:55.022]to collect those data.

[00:14:57.505]<v Tom>I wish when I had been doing consulting,</v>

[00:14:59.446]we had some holes that we have today.

[00:15:03.256]It’s being able to give them a picture of everything

[00:15:05.650]and so that they can get context and stuff.

[00:15:07.920]It’s really a game changer, I think.

[00:15:12.780]<v Amir>Yeah, for sure, yeah.</v>

[00:15:15.269]<v Tim>And with so much is unknown</v>

[00:15:17.560]at those early stages of a project

[00:15:19.260]when you’re not sure of the pathways

[00:15:21.300]and just how things are,

[00:15:26.620]how do you talk to your client when you’re uncertain

[00:15:31.900]and you’re trying to figure things out

[00:15:33.980]and they just want a cost

[00:15:35.860]and to know how this project is going to run?

[00:15:40.888]<v Amir>Yeah, I think, so what I believe</v>

[00:15:44.540]is that the uncertainty in every step of the decision-making

[00:15:52.264]needs to be communicated to the client clearly.

[00:15:56.340]And then if we can, so we can kind of characterize

[00:16:02.580]that uncertainty to our client.

[00:16:09.985]I think it’s the way to go to look at the error bar and say,

[00:16:15.150]okay, this is what we think is going on.

[00:16:18.700]This is probably the most probable picture,

[00:16:22.410]but there’s also this error bar

[00:16:25.860]or this uncertainty associated with what we think.

[00:16:29.490]And this also may not be correct

[00:16:31.750]because that might be another scenario

[00:16:34.440]that we are not aware of.

[00:16:35.650]Or based on the data that you currently have,

[00:16:39.770]so this is the most probable,

[00:16:41.870]but at the same time so we can see

[00:16:45.070]that this also might be happening with less probability.

[00:16:49.400]So if we can characterize that probability

[00:16:52.000]and show that to the client,

[00:16:53.170]I think it’s way easier later on down the road

[00:16:59.860]in the project to kind of change the path

[00:17:06.673]and kind of the client is mentally ready

[00:17:11.470]for you to say that, okay,

[00:17:14.150]that’s all of our conceptual model is now changed

[00:17:19.760]because of the one data point that we collected,

[00:17:22.630]and we need to do more investigation

[00:17:24.500]because there is another scenario now on the table

[00:17:28.080]that we need to investigate.

[00:17:30.716]<v Tim>Right, so expressing those scenarios early,</v>

[00:17:34.210]expressing uncertainty early,

[00:17:36.256]and making that really clear with the client

[00:17:39.410]sounds like that goes a long way.

[00:17:41.750]I want to ask the second question here,

[00:17:45.603]second main question as we click down through

[00:17:48.320]from a new project down to for remediation.

[00:17:53.026]So it’s time to make decisions or you want to make decisions

[00:17:58.630]about how to remediate or how to deal with the site,

[00:18:02.476]but we need to accurately delimit the plumes,

[00:18:07.200]let’s say, in the ground.

[00:18:08.620]So it’s time to build a model.

[00:18:10.626]And how are we going to do that?

[00:18:12.360]What sort of techniques are we going to lean on

[00:18:14.440]to be able to make a useful model

[00:18:18.018]of a contaminant plume in the ground?

[00:18:20.300]And I might start with you, Tom.

[00:18:22.332]What are the techniques and technologies

[00:18:26.470]you’re going to rest on for that?

[00:18:31.890]<v Tom>It depends then on the scale of the problem.</v>

[00:18:34.120]Generally I would start by using

[00:18:37.460]some kind of statistical analysis,

[00:18:39.103]do a geostatistical estimate.

[00:18:41.950]And I would also make sure

[00:18:43.870]that I would look at the hydrostratigraphy

[00:18:47.030]because that can really control how the plume develops,

[00:18:51.180]if it’s a wide plume or a long narrow plume.

[00:18:54.288]And ’cause in the beginning, as Amir was saying,

[00:18:57.760]so you don’t always have a lot of data.

[00:19:00.106]You just have really a number of limited points,

[00:19:03.200]and you can bring in other information

[00:19:05.370]like hydraulic gradients and stuff like that

[00:19:08.823]that can also maybe help there and things.

[00:19:13.003]And if you’re going to move on and do

[00:19:16.417]like flow modeling and stuff like that,

[00:19:18.650]it depends a lot again on the risk

[00:19:20.590]and the kind of receptors you have.

[00:19:24.440]But I generally start it

[00:19:25.990]with a geostatistical analysis of the data.

[00:19:29.350]It also tells you a lot about

[00:19:31.256]the weakness of the data and things like that.

[00:19:37.096]<v Tim>And, Amir,</v>

[00:19:39.260]you love a geostatistical approach as well.

[00:19:41.314]Tell us more about how you go about this stage.

[00:19:45.770]<v Amir>Yeah, as Tom mentioned,</v>

[00:19:47.277]it’s just the geostatistical is a necessity.

[00:19:51.693]So I think that delineation of the plume

[00:19:56.109]and founding a boundary

[00:19:59.434]based on the knowledge that you have is critical.

[00:20:08.781]I can’t say that accurately because it’s really hard

[00:20:13.480]in the geological context to be accurate,

[00:20:20.020]but we try our best to come up

[00:20:24.720]with the most probable picture

[00:20:26.900]based on the information that we have.

[00:20:29.154]And then sometimes it’s based

[00:20:31.950]on the complexity of the problem.

[00:20:34.594]I think it’s necessary to go that additional step

[00:20:38.020]of trying to characterize the uncertainty

[00:20:41.077]associated with that estimation,

[00:20:44.439]and then come up with some representation,

[00:20:49.760]and then, okay, this is the volume of the plume

[00:20:54.510]or the extent of the plume that’s most probable,

[00:20:57.800]but also these other scenarios are possible,

[00:21:01.680]so we can be wrong basically by twice the size of the plume.

[00:21:07.200]But I think we need to communicate that to our client

[00:21:14.312]and make them understand the risk

[00:21:16.890]associated with the uncertainty,

[00:21:19.456]and if it’s possible, convince them to reduce that risk

[00:21:26.173]as much as they can.

[00:21:28.120]So by collecting more information,

[00:21:30.507]sometimes that’s possible to reduce that risk.

[00:21:34.880]And show that, okay, this error bar

[00:21:37.547]based on the information that you have

[00:21:39.381]is now it’s that big.

[00:21:43.413]But if you collect these samples,

[00:21:45.568]then it’s going to bring that error bar

[00:21:49.310]a little bit lower or a lot lower,

[00:21:52.330]then that’s going to save you down the road.

[00:21:56.725]And then after that, when the plume is characterized,

[00:22:01.280]then we can move on to the numerical modeling.

[00:22:04.427]As Tom mentioned,

[00:22:05.800]it really depends on the risk to the receptor.

[00:22:08.420]If you want to say, okay, so how long or how many years

[00:22:12.900]it’s going to take for this plume to go

[00:22:16.027]from this current condition

[00:22:18.520]and with what concentration

[00:22:21.130]it’s going to reach to that receptor?

[00:22:23.599]Then those question can be answered

[00:22:26.700]using the numerical models and flow and transport models.

[00:22:32.469]<v Tim>Is it a case of matching up</v>

[00:22:35.331]the tools in your toolbox

[00:22:37.100]to the complexity and the uncertainty

[00:22:38.916]that’s associated with the site

[00:22:40.830]and using the appropriate tool

[00:22:43.000]for the appropriate conditions?

[00:22:45.490]<v Amir>Yeah, certainly.</v>

[00:22:47.908]Sometimes the problem is not that complex

[00:22:53.604]when you’re not having a lot of heterogeneity in the system

[00:22:58.200]or the contamination or the species of the contamination

[00:23:04.630]is not that complex,

[00:23:07.938]then we can say that, okay, use data on it.

[00:23:11.538]So geostatistical tool can also

[00:23:13.991]kind of provide an answer for us.

[00:23:16.728]Looking at the historical data collected,

[00:23:21.153]we can also move toward

[00:23:23.930]from just looking at the spatial geostatistic,

[00:23:26.870]you can use special temporal geostatistic

[00:23:31.690]and look at how the plume has evolved.

[00:23:35.297]If you have information and you do have in some site,

[00:23:38.710]you have some historical information that you can utilize,

[00:23:42.019]and based on that, it can give you an idea, okay,

[00:23:47.550]so this is the information that I have currently

[00:23:52.410]is going to be enough for me to make decision

[00:23:55.486]and move forward with the site

[00:23:58.470]or do I need to collect more samples

[00:24:01.610]and answer some of the key questions

[00:24:05.965]that’s not been answered yet?

[00:24:09.528]<v Tim>Right, what sort of tools</v>

[00:24:12.680]would you have in your toolbox, Amir?

[00:24:14.600]And I’ll ask you as well, Tom.

[00:24:18.570]And what are your go-to approaches for this phase?

[00:24:25.571]<v Amir>Yeah, I think we should always start simple</v>

[00:24:33.190]and then make things complex as needed.

[00:24:36.870]So complexity always is not the solution.

[00:24:42.701]But it can be starting from a simple 2D map

[00:24:47.179]and then some cross sections of the site.

[00:24:51.240]And then if you need it, if you have a lot of information,

[00:24:54.922]a lot of historical data, then you need more,

[00:24:58.829]something more robust.

[00:25:00.630]Then you need to look at all the angles of the problem.

[00:25:03.719]And the 3D tool definitely is going to be helpful.

[00:25:07.476]And then as get things more complex,

[00:25:12.260]than you need to have statistical tools available,

[00:25:15.520]geostatistical tools.

[00:25:16.770]And then for the plume characterization,

[00:25:22.290]sometimes you need to go through that

[00:25:24.165]special temporal geostatistics to answer your question.

[00:25:29.609]Yeah, it’s that level of complexity

[00:25:32.840]really depends on the site and the problem

[00:25:35.368]that you’re trying to solve.

[00:25:38.438]<v Tim>And Tom, what sort of tools</v>

[00:25:41.010]for modeling plumes would you have?

[00:25:44.168]What are your technologies?

[00:25:45.850]<v Tom>In a lot of ways it’s really similar</v>

[00:25:48.210]to what Amir was saying,

[00:25:49.081]though, I mean, I’m slightly biased.

[00:25:51.160]I do work for Seequent so–

[00:25:52.757](Tim laughs)

[00:25:53.840]And I’ve been building 3D tools

[00:25:55.650]even before I started working for Seequent.

[00:25:58.050]So I’ve always liked getting things in 3D.

[00:26:02.368]It’s also maybe ’cause of the way I think.

[00:26:05.010]But I always really like getting

[00:26:07.890]all the data and everything in a 3D where,

[00:26:11.860]it’s like I’ve seen plume starched my sewer lines and stuff

[00:26:16.150]where they poured stuff from the sewer

[00:26:17.740]and there was holes and stuff.

[00:26:19.080]And so by getting everything

[00:26:20.740]that you might have around the site into the model,

[00:26:24.971]I’ve always just found it gives me

[00:26:27.090]a better understanding of the problem

[00:26:29.930]and makes it easier to explain.

[00:26:33.314]But certainly the fewer data you have,

[00:26:37.816]the less robust geostatistical

[00:26:40.866]and other statistical analysis you can do, right?

[00:26:44.063]And you don’t want to over-complicate the problem

[00:26:47.340]in the beginning.

[00:26:48.380]So it’s to weigh the techniques as you move forward.

[00:26:54.190]And certainly time series analysis

[00:26:57.271]if you have that kind of data available

[00:26:59.804]is really valuable ’cause it gives you an idea

[00:27:02.500]of how things are evolving, how fast things are evolving,

[00:27:06.150]and that affects the critical nature of the problem.

[00:27:11.240]If things are moving faster,

[00:27:12.400]you need to maybe be more reactive,

[00:27:14.810]and it could be of a greater risk for your customer.

[00:27:19.291]<v Tim>Yeah, time series analysis,</v>

[00:27:21.630]I guess that’s empirical information

[00:27:23.960]on just what has been happening through time.

[00:27:26.537]And that’s really valuable to understand how.

[00:27:30.350]Hey, let’s move on.
<v ->How things are evolving.</v>

[00:27:32.660]<v Tim>Yeah, how things are evolving.</v>

[00:27:34.220]Let’s go to the next phase.

[00:27:36.260]Let’s say it’s time to design a program

[00:27:38.890]to deal with the site.

[00:27:41.210]So you’ve started, you’ve collected a lot of data,

[00:27:44.050]you’ve made conceptual models,

[00:27:46.270]you’ve dealt with a lot of the complexity

[00:27:49.670]and the uncertainty in these models,

[00:27:51.310]but we want a program to deal with the site.

[00:27:55.097]And so maybe it’s as simple as removing the source

[00:27:58.380]and then just monitoring it,

[00:27:59.450]or maybe you need to dig it,

[00:28:02.040]or you need to pump it or heat treat it,

[00:28:04.226]heat treat it or transform it.

[00:28:07.917]Each of these are going to have

[00:28:09.360]their own impacts, their own costs,

[00:28:13.160]and it’s all going to depend on the type of contaminated site.

[00:28:17.390]What’s your approach?

[00:28:18.610]What’s your general approach to this,

[00:28:20.960]making a decision in this space, Amir?

[00:28:24.530]How do you deal with your clients?

[00:28:26.760]How do you communicate?

[00:28:29.737]<v Amir>Yeah, as you said,</v>

[00:28:30.867]most of those remediation techniques

[00:28:33.490]are contaminate kind of dependent.

[00:28:37.340]So if you were dealing with a NAPL,

[00:28:39.690]so then you have different remediation technique

[00:28:44.390]than when you’re dealing with salts.

[00:28:48.683]So it’s just some of that it’s based on that consideration.

[00:28:55.150]And then the other part is going to be,

[00:28:57.840]so when you have different options,

[00:28:59.290]as you mentioned earlier,

[00:29:01.030]so it’s always a cost benefit analysis.

[00:29:07.420]So we usually come up

[00:29:09.953]with a couple of different remediation scenarios,

[00:29:14.363]and then we can test those remediation scenarios

[00:29:19.300]usually by conducting numerical modeling.

[00:29:22.970]So then you can look at how much of the contamination,

[00:29:28.140]the contaminants that you can remove

[00:29:30.000]and then how much probably is going to remain in place,

[00:29:33.030]and what’s the cost associated with all of those activities,

[00:29:36.260]and at the end, present and look at

[00:29:40.030]all of the costs and the benefits

[00:29:42.530]that that remediation is going to provide,

[00:29:45.090]and then make decision

[00:29:46.960]based on that cost benefit analysis, yeah.

[00:29:53.210]<v ->But there’ll be other factors, weren’t there, Tom?</v>

[00:29:56.360]Not just cost benefit,

[00:29:58.290]but there are the different parties that are involved.

[00:30:01.870]There are different stakeholders here,

[00:30:03.500]there’s client, there’s communities.

[00:30:06.140]How do you communicate that?

[00:30:07.831]How do you get to a decision in the space?

[00:30:12.051]<v Tom>For me, it’s about explaining</v>

[00:30:16.540]the rationale behind the approach, right?

[00:30:20.756]Generally there’s like what Amir was saying,

[00:30:24.290]again there’s receptors.

[00:30:25.790]It depends on the use and where the plume is going

[00:30:28.640]and those kind of things.

[00:30:30.760]So it’s the risk that it poses,

[00:30:33.850]and explain that clearly to all the different stakeholders.

[00:30:39.198]It’s really, really hard, if not impossible,

[00:30:42.990]to completely clean up a site.

[00:30:44.412]There’s always going to be some remedial something,

[00:30:50.257]pollution left behind.

[00:30:52.839]And you have to explain that it’s not possible.

[00:30:57.319]And also how deep the plume is, for example,

[00:31:00.670]also eliminates certain techniques.

[00:31:02.256]If it’s really deep, you’re not going to dig it up, right?

[00:31:06.871]And other techniques that may give you a better remediation

[00:31:10.320]could cause other problems, right,

[00:31:12.951]because they may cause settlement

[00:31:14.870]in surrounding buildings or something like that.

[00:31:18.311]So it’s just basically telling the story.

[00:31:22.430]You have to explain as clearly as possible

[00:31:25.588]why you’ve come to this solution

[00:31:27.220]and that it is the best solution based on cost

[00:31:30.330]and based on everybody’s general interest

[00:31:34.047]around the problem.

[00:31:36.373]In my experience,

[00:31:37.380]most people don’t want to leave an inclusion behind, right?

[00:31:39.902]They want to clean it up.

[00:31:44.021]<v Amir>Yeah, yeah.</v>

[00:31:44.854]<v Tim>Amir, you go ahead, mate, yeah.</v>

[00:31:47.585]<v Amir>No, go ahead.</v>

[00:31:49.150]<v Tim>I was going to ask do your clients struggle</v>

[00:31:51.646]to make a decision when they’re presented with scenarios

[00:31:54.500]and options for remediation?

[00:31:56.290]Is it hard for them because they’re not

[00:31:58.571]necessarily.
<v ->It really depends</v>

[00:31:59.554]on the costs.

[00:32:00.387]So if it costs this.
<v ->Right.</v>

[00:32:02.100](Tim and Tom laughing)

[00:32:05.210]<v ->Let’s say you’ve chosen or the client has chosen</v>

[00:32:09.650]a remediation strategy,

[00:32:11.150]and your remediation is in full swing now,

[00:32:13.540]so maybe you’re pumping or maybe you’re heat treating

[00:32:17.340]or doing something like that,

[00:32:18.650]but you’re operating to remediate.

[00:32:22.109]What are some of the best ways that you can monitor?

[00:32:25.049]Because this is going to take some time typically.

[00:32:27.976]How are you going to monitor and sample

[00:32:31.299]and report on that remediation?

[00:32:34.864]And how are you going to have confidence

[00:32:37.860]or relay your confidence that your technique is working?

[00:32:42.100]How are you going to communicate that with your client?

[00:32:45.288]<v Amir>Yeah, again that mapping</v>

[00:32:47.970]kind of help us to do that.

[00:32:50.180]So when at remediation,

[00:32:55.590]the site is under remediation,

[00:32:57.920]and then you know that some information

[00:33:01.864]kind of get collected regularly at the site,

[00:33:04.570]and then you have that kind of opportunity

[00:33:11.610]to look at those information

[00:33:13.170]and try to relate that to the remediation

[00:33:18.564]that’s ongoing at the site

[00:33:20.690]and try to see if that remediation technique

[00:33:26.022]that’s already in place is effective

[00:33:29.500]or how effective that remediation is going to be.

[00:33:33.010]Is that the rate of, extraction rate

[00:33:36.810]is kind of reaching to a steady state condition

[00:33:41.480]or kind of what’s that rate of change

[00:33:48.210]when you’re dealing with LNAPL also.

[00:33:50.867]So are you extracting, reaching to that point

[00:33:55.590]that your remediation technique is no longer effective

[00:33:59.900]and you’re better off moving

[00:34:02.310]to another remediation technique

[00:34:04.416]or you just simply stay and watch and monitor the situation

[00:34:10.020]rather than doing anything?

[00:34:12.160]So those kind of decisions is hard

[00:34:18.190]and it really depends on the data that’s been collected.

[00:34:22.933]Sometimes because you have the remediation system working,

[00:34:27.450]so you also add another level of complexity to the site,

[00:34:31.209]so that the dynamic and the impact

[00:34:34.904]of the remediation system itself,

[00:34:38.970]so add a level of complexity or noise

[00:34:42.120]to the data that’s been collected.

[00:34:44.582]And you need to understand that noise

[00:34:47.590]and try to remove that noise from the data,

[00:34:51.206]and then make decision based on the actual information,

[00:34:56.530]not based on the noise that’s been kind of

[00:35:00.610]forced into your datasets.

[00:35:02.390]So those are important things that we need to consider.

[00:35:09.510]<v Tim>Right, so it’s a dynamic system</v>

[00:35:12.049]where the actual remediation techniques

[00:35:14.340]are changing the data.

[00:35:16.410]You’re not seeing the system in its natural state anymore.

[00:35:20.209]Very difficult.

[00:35:21.280]And Tom, how are you going to communicate with your client?

[00:35:27.836]I mean that you could email, you could phone call,

[00:35:30.620]but it’s 2020.

[00:35:32.609]What are some of the advanced ways

[00:35:35.926]that you can really make this easy

[00:35:38.160]for the client to understand as the remediation runs?

[00:35:43.290]<v Tom>I don’t know if any of those</v>

[00:35:45.080]are ever easy to understand,

[00:35:47.100]but one thing just to follow up on Amir is saying

[00:35:53.860]is one thing that I’ve done sometimes

[00:35:55.820]is actually do model scenarios

[00:35:58.209]with or without the remediation

[00:36:00.300]so that you maybe you can find how much noise there is,

[00:36:03.707]and then you can use those two different scenarios

[00:36:06.480]to explain to the customer

[00:36:08.153]what it would be if you hadn’t done the remediation,

[00:36:10.853]this is the benefit that you’re getting from a remediation.

[00:36:15.213]Those sites usually require that you had really good data

[00:36:18.700]before you start remediating.

[00:36:22.010]But, like I was saying in the beginning,

[00:36:27.094]a lot of times it’s a good picture, right?

[00:36:31.550]It’s all about being able to show them visually

[00:36:36.214]how things have changed.

[00:36:37.970]And if you can show them where,

[00:36:41.300]say, a pumping well, is for capturing a plume,

[00:36:44.870]and show how the plume has changed

[00:36:46.930]since that has been in operation and things like that,

[00:36:49.792]it just makes it much easier.

[00:36:54.260]Also if you were doing a funnel and gate

[00:36:56.290]and other kinds of remediation strategies,

[00:37:00.608]being able to show it in 3D

[00:37:02.760]and how things are changing around that and context,

[00:37:06.240]’cause I think that’s a lot of times

[00:37:07.470]where they have trouble.

[00:37:08.470]My experience is if you just show them that,

[00:37:10.667]they can’t picture

[00:37:11.820]that it’s a 3D compression of things, right?

[00:37:14.452]And then maybe the plume is constrained

[00:37:17.886]by a specific gravel layer or something like that.

[00:37:21.194]And again that’s where combining things

[00:37:23.903]in a 3D interactive environment can be useful.

[00:37:30.516]<v Amir>Yeah, certainly, yep.</v>

[00:37:33.470]<v Tim>Yeah, and what are some of the creative ways</v>

[00:37:36.740]you’ve reported back during remediation, Amir?

[00:37:40.794]What’s your go-to for that kind of communication?

[00:37:47.466]<v Amir>Yeah, I think that the important part</v>

[00:37:51.020]is we need to acknowledge, as I said before,

[00:37:53.910]so I’m really a fan of characterizing uncertainty

[00:37:58.906]and just considering that impact

[00:38:01.980]on every aspect of your project.

[00:38:06.071]So I think that that’s going to be really be helpful

[00:38:09.269]for the client to understand how things are evolving,

[00:38:13.120]because sometimes when you look at the raw data,

[00:38:16.320]and then even if you map

[00:38:18.280]based on the information that you have,

[00:38:21.244]so you’re not looking at all the variables

[00:38:25.680]that you don’t know.

[00:38:26.513]So the porosity, the bulk density,

[00:38:30.876]all of those variables, the hydraulic conductivity,

[00:38:33.650]all of those variables or even the characterization

[00:38:37.686]of the plume itself is complex,

[00:38:42.758]and it has a lot of unknown or estimated variables in it.

[00:38:48.694]So I think that characterizing that,

[00:38:51.506]it’s going to help the client to understand

[00:38:55.115]that, okay, so we think that things are going down,

[00:39:00.200]so the concentrations,

[00:39:01.510]but there’s also a possibility that we might be wrong

[00:39:04.570]because these factors are involved.

[00:39:09.116]So then it’s going to help to acknowledge

[00:39:13.140]that all of those unknown in the equation,

[00:39:15.890]that we know that that’s also a possibility, so, yeah.

[00:39:22.020]<v Tim>And do you do that using simulation?</v>

[00:39:24.539]I mean, like stochastic simulation?

[00:39:26.757]<v Amir>Yeah, so you can do it</v>

[00:39:29.104]doing a stochastic simulation.

[00:39:32.688]Sometimes that’s the way to go.

[00:39:35.069]Sometimes even looking at a simple Monte Carlo analysis,

[00:39:41.410]just looking at the probability of your input parameters

[00:39:45.360]and then try to sample that probability of space

[00:39:48.310]and kind of map it based on that probability of space,

[00:39:53.110]and then that’s the way to go.

[00:39:55.431]So it really depends on the problem.

[00:39:58.592]Then that’s going to be easier to understand

[00:40:02.870]the complexity for the client

[00:40:05.260]and then kind of acknowledge the unknown in the equation.

[00:40:09.107]Yeah, so and then, okay,

[00:40:10.843]so if we can have a better understanding,

[00:40:13.090]and then we can look at the sensitivity

[00:40:15.250]of all of those parameters,

[00:40:17.000]and say, okay, if we have a better understanding,

[00:40:19.326]for example, about the bulk density or about the porosity,

[00:40:22.990]then that error bar actually

[00:40:25.590]is going to be way smaller than it is right now.

[00:40:28.030]So then we can kind of understand

[00:40:31.190]which direction of sample collection do we need to go

[00:40:34.089]to reduce the uncertainty.

[00:40:42.280]<v Tim>Nice, so some really advanced techniques there.</v>

[00:40:45.020]We’re right at the end of our project now.

[00:40:46.710]You might have finished your remediation,

[00:40:49.380]you’ve monitored it through,

[00:40:50.970]and you have confidence that the job has been done,

[00:40:55.499]but some sites are still problematic.

[00:40:58.806]There’s a residual contamination

[00:41:01.960]or there’s other unforeseen effects

[00:41:05.280]that have happened there.

[00:41:06.270]So in your experience, how do you leave a project

[00:41:12.650]and what’s the strategy

[00:41:15.764]to deal with any of these unforeseen effects?

[00:41:20.329]So, Amir, what sort of problematic effects

[00:41:24.337]do you see in these sorts of sites

[00:41:27.830]and what’s been your experience?

[00:41:32.553]<v Amir>Yeah, there’s always problems</v>

[00:41:36.980]that need to be solved.

[00:41:38.090]So, as Tom mentioned,

[00:41:39.663]I think that we can’t really clean up 100%.

[00:41:44.660]So it’s really hard or if it’s impossible.

[00:41:53.900]I think looking at the problem

[00:41:54.907]and then trying to clearly communicate that to our client

[00:41:59.270]and show that the risk of, for example,

[00:42:04.057]the residual contamination is minimal

[00:42:07.930]or that can be mitigated

[00:42:11.796]by having a monitoring network in place,

[00:42:17.980]and then looking at how that residual

[00:42:20.490]is evolving through time.

[00:42:22.825]And even sometimes it’s just the site is still active,

[00:42:30.491]then even the client has no idea

[00:42:34.330]if the source is contained or not.

[00:42:37.364]So then you were conducting that remediation activity,

[00:42:41.030]then you don’t know if dirty stuff

[00:42:47.900]is leaking in the ground or not.

[00:42:49.610]So then you need to have

[00:42:52.040]some sort of contingency plan in place

[00:42:55.151]to make sure to cover that part for the client, yeah.

[00:43:06.752]<v Tim>Tom, what’s been your experience</v>

[00:43:09.130]with problematic sites?

[00:43:18.010]<v Tom>I can’t think of any problem site</v>

[00:43:20.070]that wasn’t problematic,

[00:43:21.110]but I think that the biggest problem

[00:43:25.840]is explaining to people that what’s left

[00:43:29.459]and the monitoring plan,

[00:43:34.320]to watch it is satisfactory,

[00:43:40.247]because, as Amir’s pointed out,

[00:43:43.090]there’s always uncertainty, right?

[00:43:45.450]So a monitoring well could miss something, right?

[00:43:49.370]But to demonstrate to people that this is good enough,

[00:43:52.699]and you will detect something before there is a problem

[00:43:57.440]that threatens their interest

[00:43:59.490]for the different stakeholders.

[00:44:01.979]And the other problems that I’ve often seen is that

[00:44:05.830]the ideal solution isn’t feasible for some reason,

[00:44:11.532]because of how people want to use the site

[00:44:15.760]or various other stakeholder interests,

[00:44:18.740]that it’s always a compromise,

[00:44:21.290]and getting a compromise that has an acceptable risk

[00:44:24.610]can be, right, at times challenging.

[00:44:32.984]<v Tim>Complex sites, you can’t see everything.</v>

[00:44:35.886]You’ve got to rely on your data,

[00:44:38.280]and that’s the nub of it I think.

[00:44:42.380]You just cannot see what you’re dealing with.

[00:44:44.526]So thank you for answering those questions, Tom, Amir.

[00:44:50.162]I’m just looking at my watch.

[00:44:51.370]We still got 15 minutes and I’m going to turn it over

[00:44:54.070]to our audience questions.

[00:44:56.560]Thank you for that person.

[00:44:57.750]I’ve got one here.

[00:44:59.168]And here we go.

[00:45:00.852]3D plumes are typically interpolated from sample data

[00:45:05.950]and then constrained by geology,

[00:45:07.935]but oftentimes they do not incorporate groundwater flow.

[00:45:12.010]So do you have any thoughts on this?

[00:45:14.350]And maybe we’ll start with your Tom.

[00:45:15.540]Do you have any thoughts?

[00:45:16.799]Would a transport model be a more reliable estimate

[00:45:22.543]since that incorporates eviction

[00:45:25.640]of other transport processes?

[00:45:27.804]What are your thoughts, Tom?

[00:45:28.960]I’ll come to you next, Amir.

[00:45:31.214]<v Tom>Technically yes, but sometimes you know so little</v>

[00:45:35.490]that the added value is questionable.

[00:45:41.892]Obviously a flow and transport model incorporates

[00:45:47.310]a lot of processes that you can actually include things

[00:45:52.090]like the groundwater table gradient in an interpolation

[00:45:56.130]if you want, right?

[00:45:57.508]There are ways to do that.

[00:46:02.400]One thing I would say is that I would actually do

[00:46:08.160]what he suggested, meaning, the other way around

[00:46:10.580]is I would constrain the samples

[00:46:13.810]by the different geological formations

[00:46:17.070]and how they transport water,

[00:46:19.050]say, a sand and gravel are different than, say, a silt,

[00:46:22.392]and then look at the problem,

[00:46:26.193]because otherwise you’re comparing apples and oranges.

[00:46:33.606]<v Tim>Amir, your thoughts.</v>

[00:46:35.408]<v Amir>Yeah, as Tom mentioned,</v>

[00:46:37.640]so when you have information

[00:46:41.370]about the geology constraint and also the water table,

[00:46:45.431]there are ways that you can use

[00:46:48.100]and utilize those information

[00:46:50.872]to kind of help you characterize and delineate

[00:46:55.480]your plume more effectively or more realistically.

[00:47:00.200]So, for example, you can look at, again,

[00:47:04.416]if in some certain conditions that that hydraulic gradient,

[00:47:10.710]it’s going to kind of match up

[00:47:12.870]with where your plume is moving towards,

[00:47:15.880]so then you’re anisotropy factor in interpolation,

[00:47:20.810]it can be aligned with that anisotropy direction

[00:47:24.638]that you have based on the water table,

[00:47:27.584]for example, orientation,

[00:47:29.300]then that’s going to help you in your estimation as well.

[00:47:33.728]And then if you have information

[00:47:36.440]about the geological difference

[00:47:41.770]in the system that you are studying,

[00:47:44.090]the data collected in those geological units

[00:47:49.429]then can help you to kind of have that

[00:47:52.700]instead of having just one model interpretation,

[00:47:57.170]then you have kind of a modular interpretation.

[00:48:00.230]And kind of seeing the effect

[00:48:03.700]of that non-stationarity in the system,

[00:48:05.820]so you have, for example, sandy and a coarser material,

[00:48:11.287]and then you have a finer material,

[00:48:13.860]then you knew that plume is not going to travel

[00:48:17.250]in these two at same rate,

[00:48:19.940]then if the data collected in those two units,

[00:48:26.890]then you can use that in your favor and use those separately

[00:48:30.480]to characterize and use it in different statistical

[00:48:34.510]or geostatistical approach to kind of modulate the main

[00:48:38.720]into two different modules and then re-interpolate it,

[00:48:41.880]and then kind of combine those information

[00:48:45.400]so that there are ways to handle those issues.

[00:48:49.620]And then regarding the numerical model,

[00:48:51.880]so for numerical model,

[00:48:54.830]you need to establish your initial condition,

[00:48:58.584]which the only way that you can establish

[00:49:01.910]is mapping your plume,

[00:49:04.554]which is you need to start with the geological

[00:49:11.070]or interpolation of your plume,

[00:49:13.100]and then move on to a numerical model

[00:49:15.530]to try to assess the risk to your result, yeah.

[00:49:24.417]<v Tom>I think it’s also important</v>

[00:49:25.761]to remember what kind of plume it is too, right?

[00:49:28.240]If it’s a DNAPL,

[00:49:29.510]it’s going to be a completely different problem

[00:49:31.100]than if you’ve got a salt

[00:49:31.969]or something that’s easily dissolvable in water too.

[00:49:36.873]<v Amir>Yeah, yeah.</v>

[00:49:38.227]<v Tom>They’re really important.</v>

[00:49:39.990]<v Amir>Yeah, that’s going to, that’s LNAPL DNAPL.</v>

[00:49:42.507]And then if you’re looking at the aqueous phase,

[00:49:45.450]or you’re looking at the oil phase,

[00:49:50.760]so then it’s completely, or non-aqueous phase,

[00:49:54.910]then it’s going to be a completely different problem

[00:49:57.367]and different tool that you need to utilize

[00:49:59.440]and different considerations for sure.

[00:50:02.029]<v Tim>Yeah, you got to know</v>

[00:50:02.911]what you’re dealing with for sure.

[00:50:05.070]And I guess experience comes to play there.

[00:50:08.063]Hey, I’ve got another question here.

[00:50:09.812]This one’s about geostats and lag spacing.

[00:50:13.424]So unlike exploration geostats,

[00:50:17.498]I guess in the mining sector maybe,

[00:50:20.528]wells often are variably spaced.

[00:50:24.048]So what is your method

[00:50:25.740]for determining a reasonable lag spacing

[00:50:28.160]when your wells are variably spaced

[00:50:31.150]in a contaminated site, Amir?

[00:50:34.930]<v Amir>Yeah, I think it’s a benefit</v>

[00:50:37.616]of having variably spaced.

[00:50:40.790]It’s not a negative thing I think.

[00:50:44.294]Based on my geostatistical experience,

[00:50:47.320]I think that’s kind of a benefit

[00:50:49.200]of having that variably spaced,

[00:50:51.740]so then when you’re looking at your variogram

[00:50:55.250]and doing that variography,

[00:50:57.798]then you have all sorts of variation

[00:51:00.464]in your basically lags operations,

[00:51:04.870]and then you can change the tolerance to kind of cover

[00:51:08.410]and have different points and nodes representing,

[00:51:11.960]and then you can come up with a better understanding

[00:51:14.630]about the covariance function.

[00:51:17.427]So I don’t see that as a negative thing.

[00:51:20.904]So the issue usually is that you have limited data.

[00:51:26.650]So, and then the interpretation

[00:51:30.550]and having that limited information

[00:51:35.476]kind of limit our understanding to characterize the variogram

[00:51:41.939]in a way that it needs to be characterized.

[00:51:48.440]So we need to kind of, when we have that kind of situation,

[00:51:52.390]I think we need to also consider

[00:51:54.710]the uncertainty in the variogram

[00:51:58.757]as one of the other unknown in that equation.

[00:52:03.240]So then you need to try to characterize

[00:52:05.660]that uncertainty in the variogram as well

[00:52:08.414]when you’re interpreting the plume.

[00:52:14.477]<v Tom>But also if you know some characteristics</v>

[00:52:18.410]or general characteristics of the system,

[00:52:20.801]the groundwater table gradient,

[00:52:22.428]the hydraulic conductivity,

[00:52:24.120]and some crude thoughts about the size of the plume,

[00:52:29.649]then there are techniques for designing a lag spacing,

[00:52:34.236]the sample spacing,

[00:52:35.943]so that it’ll better characterize your plume.

[00:52:41.070]But, like Amir was saying,

[00:52:43.174]my experience is that you usually run out of money

[00:52:46.260]before you (laughs) can get a really good standpoint.

[00:52:54.210]But again the variable, just to reinforce what he said,

[00:52:56.636]the variable spacing is a good thing.

[00:52:58.600]It’s not a bad thing.

[00:53:00.327]And if you look at any classical

[00:53:02.990]geostatistical sampling design,

[00:53:05.880]it says to make sure you have variable spacing.

[00:53:10.375]<v Tim>It’s that cost benefit again, isn’t it?</v>

[00:53:12.920]It’s all down to cost.

[00:53:14.175]Hey, you were talking before, Amir, you were talking about

[00:53:18.073]geostatistical simulation techniques and Monte Carlo,

[00:53:21.867]and I’m going to go out there on a limb and say and propose

[00:53:25.580]that maybe not all environmental geoscientists like yourself

[00:53:31.810]use those sorts of techniques.

[00:53:33.160]So if there’s someone in the audience

[00:53:34.560]who is interested in starting,

[00:53:37.583]is there a place that you would advise

[00:53:40.400]that they get going in the space?

[00:53:43.663]Maybe just one more simple technique.

[00:53:46.040]And how could they apply geostat simulation

[00:53:50.020]to one of the problems?

[00:53:52.150]You’ve got any advice where someone might start?

[00:53:56.772]<v Amir>Yeah, I think</v>

[00:53:59.642]the geostatistical simulation in general,

[00:54:02.460]it’s a really complex, I think, science.

[00:54:06.084]It’s been evolving

[00:54:09.123]into then some really complex technique

[00:54:12.120]to characterize heterogeneity.

[00:54:13.423]It’s been developed during the past maybe 20, 30 years,

[00:54:19.557]but at the same time so the technique that usually

[00:54:23.503]there’s just a geostatistical technique that we use.

[00:54:26.654]So for example, Kriging,

[00:54:29.944]it’s a complex equation mathematically,

[00:54:33.530]but it’s not complex to apply.

[00:54:36.366]But at the same time, so then you can,

[00:54:40.867]I think most of the geostatistical softwares,

[00:54:44.598]they have a portion or a part for geostatistical simulation

[00:54:51.200]based on, for example, Kriging.

[00:54:56.021]It’s the Gaussian, for example, simulation.

[00:54:59.370]So it’s based on the Kriging idea.

[00:55:02.742]And then you have that tool,

[00:55:06.140]then it’s just a matter of understanding

[00:55:09.270]some key parameters.

[00:55:11.871]And then from that, it’s not that hard to implement.

[00:55:18.720]I know it’s counterintuitive.

[00:55:23.969]<v Tim>Can it be self-taught do you think</v>

[00:55:26.100]sequential Gaussian simulation?

[00:55:27.504]Can you learn that online

[00:55:29.770]or is there somewhere you’d recommend that people go to–

[00:55:34.460]<v Amir>I think it needs a little bit of formal education.</v>

[00:55:41.040]<v ->Right.</v>
<v ->Or just sitting in a class.</v>

[00:55:43.030]That might help.

[00:55:44.530]I couldn’t find, unfortunately, lots of,

[00:55:48.683]after you pass that barrier to understanding the whole idea,

[00:55:55.191]then you can teach yourself those concept,

[00:55:59.130]but it’s just that the starting point

[00:56:02.060]and understanding the whole idea of geostatistics,

[00:56:06.446]then that’s a little bit,

[00:56:09.628]I think it needs a little bit of formal education, yeah.

[00:56:13.494]<v Tim>No, thank you for that.</v>

[00:56:15.028]Good advice there.

[00:56:16.570]Hey, we’re coming to the top of the hour,

[00:56:18.340]but we’ve just got one more question,

[00:56:20.375]and it reads like this,

[00:56:23.151]Contaminated sites,

[00:56:24.516]often contaminated sites deal with politics,

[00:56:28.186]namely only showing a plume

[00:56:31.790]where it’s been confirmed with wells correctly.

[00:56:35.257]What techniques can be used

[00:56:37.776]to keep the interpolation conservative

[00:56:40.600]and not extrapolate too far?

[00:56:43.618]Ooh, politics and plume boundaries.

[00:56:48.144]Tom, what’s your experience here?

[00:56:51.899]<v Tom>Well, I kind of alluded to that a bit before.</v>

[00:56:56.346]There’s always political constraints

[00:56:58.380]about what you can do and everything.

[00:57:02.682]And this goes back to something

[00:57:05.285]that both Amir and I have been saying a lot through this,

[00:57:08.840]and that’s that wonderful world, geostatistics.

[00:57:15.290]And it takes a lot of, in my experience,

[00:57:18.250]a lot of patience and education

[00:57:22.410]of the customers and of the regulators,

[00:57:24.870]but geostatistics has this, it’s called BLUE.

[00:57:31.290]It’s the Best Linear Unbiased Estimator, right?

[00:57:36.890]And it’s to carefully educate them

[00:57:43.820]so that they come on a journey of self-discovery,

[00:57:46.830]so that they can accept that

[00:57:51.040]these estimates are conservative.

[00:57:56.149]And another way I do that

[00:57:58.863]is also to do these simulations and stuff like that

[00:58:02.910]’cause that will also demonstrate

[00:58:05.670]how the estimate is conservative,

[00:58:09.865]’cause sometimes, I think I’m going to stop there.

[00:58:16.250]I’d like to hear what Amir has to say. (laughs)

[00:58:20.244]<v Amir>Yeah, yeah, even in geostatistics sometimes,</v>

[00:58:28.325]that extrapolation is based on the information

[00:58:32.220]that’s coming from the data.

[00:58:33.870]So it’s just the information suggesting that there is,

[00:58:38.210]for example, this branch in the data.

[00:58:41.210]So there is that’s either 100 meters of correlation,

[00:58:46.344]and then you have a data point

[00:58:48.435]and then that has a concentration.

[00:58:52.146]And then after that, you don’t have any,

[00:58:55.411]basically going back into baseline concentration.

[00:59:00.446]that data point.

[00:59:02.206]So then based on the correlation

[00:59:04.958]that exists within the data set,

[00:59:08.080]then we kind of extrapolate.

[00:59:15.006]Sometimes it seems to be a little bit too far,

[00:59:17.820]but the issue is because we don’t have information,

[00:59:21.890]so that’s why it’s educated guess

[00:59:26.850]based on the information that we collected.

[00:59:30.194]And then if you want to kind of say, okay,

[00:59:33.800]so it’s not going that far,

[00:59:35.970]the extrapolation is not unnecessary,

[00:59:39.350]or it’s not extending, for example, 100 meters,

[00:59:42.469]then you need to collect data.

[00:59:45.797]So sometimes I’ve seen people do hacking,

[00:59:51.080]just adding some control point

[00:59:53.420]based on some kind of information or educated guess

[01:00:00.670]or just some gut feeling that, for example,

[01:00:05.004]I think that that location that the plume

[01:00:09.232]or the concentration is zero.

[01:00:11.160]Then you can do that,

[01:00:14.114]but I don’t think it’s scientifically defensible.

[01:00:18.200]So it is hard to justify those,

[01:00:22.327]other than, for example, if you have real boundaries,

[01:00:25.420]so if you have geological constraints,

[01:00:29.450]so the unit is changing and suddenly you think

[01:00:32.963]that from that point,

[01:00:37.537]the plume is not going to travel into that formation

[01:00:41.430]because it really consist of fine material,

[01:00:45.050]the conductivity is 10 or 100 orders of magnitude smaller.

[01:00:49.870]So then you can do those kind of things.

[01:00:53.120]But other than that, it’s just hard to justify

[01:00:56.523]those adding politics, I think,

[01:00:59.110]into science.
<v ->Politics. (laughs)</v>

[01:01:02.098]<v Tim>So got to go back to the data.</v>

[01:01:04.553]You go, Tom.

[01:01:05.892]<v Tom>I think something else too</v>

[01:01:08.600]where hydrogeology can learn from other areas

[01:01:11.100]where geostatistics is applied

[01:01:13.080]is there’s actually a whole lot of,

[01:01:14.453]there’s a really strong toolbox,

[01:01:18.009]a strong field toolbox of methods

[01:01:21.770]to show how robust your estimation is.

[01:01:24.990]You can do swath plots, you can jack knife.

[01:01:30.100]And a lot of these tools are somewhat readily available

[01:01:36.120]and products that are accessible.

[01:01:40.756]And those would demonstrate

[01:01:45.210]and help educate the stakeholders

[01:01:49.560]on the validity of these estimates that they are robust.

[01:01:53.296]I’m not really sure I like the word conservative,

[01:01:56.010]but robust and sound.

[01:01:59.813]And again this is where our field can learn a lot

[01:02:04.070]from what other people have been doing.

[01:02:06.699]<v Tim>Robust is a good word.</v>

[01:02:08.192]And I think that’s a great place to finish up today.

[01:02:12.351]But I do want to thank our experts.

[01:02:15.870]Thank you, Tom.

[01:02:16.703]It’s your evening in Denmark.

[01:02:19.190]Thank you so much for staying up and dialing in.

[01:02:21.280]And, Amir, thank you so much for your time

[01:02:23.899]and for all your comments.
<v ->Thank you.</v>

[01:02:26.780]<v Tim>It’s been great.</v>

[01:02:27.700]And I hope you and the audience have taken something away.

[01:02:32.560]I hope there’s something that’s come from this session

[01:02:35.820]that you’ve learned and you’re going to apply.

[01:02:39.530]And, yeah, it’s been fantastic having you join us today.

[01:02:44.005]This will be available as a recording.

[01:02:46.415]Thank you, guests.

[01:02:47.850]Thank you, experts, Tom, Amir.

[01:02:50.760]And we’ll sign off from here.

[01:02:53.883]<v Amir>Thank you so much.</v>

[01:02:55.632]<v ->Bye.</v>
<v ->Bye.</v>

[01:02:57.340]<v Tom>Thank you everybody for attending.</v>