Video transcript

00:00:06
Good morning all and welcome to this seminar hosted by the office of the Queensland Mine Rehabilitation Commissioner uh on erosion and landform evolution modeling. Before we get started, I would like to acknowledge the traditional owners on the land on which meeting today and pay my respect to the elders past and present. I'd also like to acknowledge that today is remembrance day. Uh a day of national commemoration and on this day at 11:00 a.m. Australians pause for a minute silence to remember those [clears throat]

00:00:38
uh the bravery and sacrifice of all people who lost their lives while serving for Australia and its allies in wars and conflicts and peace operations. So along the lines at 11:00 a.m. we will observe a minute silence. Um, a couple of housekeeping items. We will, today's session will be recorded. We also have the ability for you to lodge questions throughout the session. To do this, you can use the Q&A function in your top toolbar. All your questions will come to our team and then we'll get to as many

00:01:09
of them as we can throughout today's session. So, with that all said, I'm now going to hand over to today's MC, our principal technical advisor, Jason Dunlop. Over to you Jason. >> Thanks Sam and welcome everyone to this session on leading practice erosion modeling to design post mining landforms. We know that this is one of the most challenging parts of mine rehabilitation and building post mining landforms is hard. Uh often we're dealing with materials that are dispersive and prone to high levels of

00:01:46
erosion. uh we're seeing, you know, episodic rainfall events that can cause quite a lot of damage on these uh fragile systems. So, the consequence of failure is high. I think many of us have seen examples where a landform has been built and then over the years hasn't stood the test of time um and and has has failed. When they fail, they can be uh quite expensive, you know, having impacts on site and also offsite um to adjacent landholders. So it's something that we really need to be able to do the

00:02:15
work up front and try and avoid down the track. So we [clears throat] know that the al the objective is to create uh stable land forms. So the three pillars of rehabilitation is to make land safe, stable and non-polluting. So when we're talking about stability in this context, we're talking about u a landform that doesn't have active eroding ress and uh gullies and doesn't have tunnel erosion as well. And we also want to make sure that any hostile materials are buried within the landform and are not exposed

00:02:49
into the long term so that they're covered and remain covered. So erosion modeling is is critical to this process. Um the progressive rehabilitation and closure planning planning uh framework within Queensland now requires modeling to underpin the the planning process and there has been increasing use um of modeling as as a result. Um but it's a great tool. It provides a great way to test and compare different scenarios and it helps you uh look at different options and and and and refine your designs.

00:03:23
It is complex though. It requires specialist skills and some careful consideration needs to go into it. So um this workshop will will talk through some of those issues that do need to be considered. Um we're going to look at uh the current approaches to erosion modeling and landform design. We'll talk through uh what what is leading practice of erosion modeling and then we'll talk through how to achieve leading practices in your own projects out there. So today's workshop will also will step

00:03:52
through that guidance that has been published. Um we have we have a publication on the Queensland Mind Rehabilitation Commissioner's website that talks about this work. Um and I'd encourage you to have a look at that, download it, have a read um if you can after the session. Um we will have a Q&A session at the end of each session. So we won't um be waiting until just the end. We'll have a couple of stop points along the way. So, please um add any Q&A uh questions that you have into the Q&A function uh and

00:04:25
we'll bring those to the presenters along the way and then we'll also have uh a Q&A right at the end to to sum up. But uh before we jump into the session um I I just wanted to give a little bit of background on our office. So many of you out there will be familiar with our office. We were established almost 5 years ago um when there were amendments to the environmental protection act in Queensland um and the establishment of a mine rehabilitation commissioner um came through in that process. So our vision

00:04:57
is to lead Queensland to achieve best best practice in mine land rehabilitation and the commissioner is an independent role. It's not regulatory. Uh we're also not seen as the regulator of the regulator. we actually sit separate from the regulator. Um and our office has an advisory function. So key to that is um bringing together stakeholders and connecting uh people to um to discuss min rehabilitation issues. We also describe leading practice which is the focus of today and uh we also provide

00:05:31
advice directly to the minister for environment and we also report on rehabilitation trends and practices across the sector. We release an annual report on our website and I'd encourage you to go and have a look at that. It talks about, you know, the error of land disturbed versus rehab rehabilitated land um and a whole range of interesting facts and figures in there. So, I encourage you to check that out. So, I wanted to introduce the speakers today that we have. We're we're really

00:06:02
pleased for this project to have engaged landlock to describe leading practices on erosion modeling and landform design. Uh and the the primary author is is Dr. Rob Lach and Rob is one of Australia's leading erosion research and management specialists. Rob has over 50 years of experience in research and management of soil erosion, soil conservation and land rehabilitation. Rob's an emmeritus consultant with Landlock and an adjunct professor with the Center for Agricultural Engineering at the University of Southern

00:06:34
Queensland. Rob's also a certified uh professional soil scientist and holds a registered soul profession professional accredititation in so management and gully erosion. We also have uh Evan Howard with us today and Evan's uh joining us from uh Western Australia. Evan is the managing director and principal consultant at Landlock with he has over 25 years of experience in the management of soils and mining waste and soil erosion testing and modeling. Evan's also a a certified professional soil scientist

00:07:07
and has completed erosion assessments and modeling projects for mining clients across the world and and across commodities including coal, iron ore, gold and other metals uh including borkite and diamond and mineral sands. Um uh Evan has worked across Australia, the US and Africa, including temperate and tropical climates in Queensland. Evan manages the landlock operations in Perth and we um thank him for his contribution. Also wanted to add a shout out uh and for the contribution that 3D data guidance made to this project and in

00:07:46
particular Linda McGrath from Landlock. Uh it's also important to point out that that the work that is prepared by experts um that contribute to leading practice is externally reviewed by um independent experts. And I wanted to thank uh Tim Nash and Evan Thomas in particular for reviewing this work and providing their feedback and input to it. So without further ado, I'll hand back over to Rob and um we can crack on with the presentation. Thank you Jason and um good morning everyone. Now today our focus is on this

00:08:24
recent publication. Um as Jason says said uh I hope everyone's downloaded it and had an at least an initial look at the content. There's actually a lot of detail in that document and we're not going to cover it all today even though today we have got a hell of a lot of information to work through. Today we'll focus on some quite key concepts that were identified and presented in that uh document as well as going through aspects of how to assess and review landformise design studies.

00:09:00
The presentation will be in three sections. First working through key concepts and then applying a check the check sheet which for how to review uh landform design studies which is appendix 4 in the report and there should be time for a general discussion at the end. You'll notice the key concepts which probably should have been obvious to us all but uh some of them certainly rared their head or made themselves more obvious in the process of preparing this report. And the first being the importance of erosion model accuracy in

00:09:37
mindsight applications. I'll talk a little bit about models, the concept of ensuring that modeling is fit for purpose. introduce concepts of landform risk and talk to target erosion rates for modeling. Now, erosion model accuracy. Traditionally, when we've worked with erosion models, we've tended to immediately assume that a model prediction is accurate. So, and it's really important to understand that they're not necessarily all that accurate at all. But in the mining context in particular, accuracy

00:10:17
matters a lot. Which means that we require some rethinking of concerns with respect to accuracy. And in approaching modeling, we need to match particular model accuracy to sight specific requirements. And the reports that are produced need to actually talk to the issues of accuracy and how it's managed. So currently erosion modeling for mine sites and for wasteland form rehabilitation the modeling is used to generate landform designs. Now getting a mind design wrong can be very costly in terms of um

00:11:10
its impact on reshaping. Estimates from 3D data for example suggest that for a 200 hectare wasteland form if we decided our batter gradients were wrong by 2% and had to doze down from 17 to 15% that had cost nearly $800,000. And if in fact the difference was 4% um you're getting well over a million 1.383 million uh 1.383 million. So the point here is that design and um finalization of a land form for rehabilitation does involve extensive and expensive placement and shaping of materials. And I suppose at that point I'd also add

00:11:58
a couple of um uh comments um in my experience in in Western Australia that the there's some examples where the cost of um of dozing down and reshaping are in in the order of 80% of the the total rehab costs for a for a site. And we've seen where the actual shape of the landform is designed or at least built so that it matches the final landform shape. um that that figure can be reduced um to to 30 or 40% of the of the cost. So the use of models um is actually quite quite important and does

00:12:36
have significant cost implications for uh for mining operations. There's also the point that mining is a fairly regulated environment and for example erosion simulations now an essential component of progressive rehab uh and closure planning PRCP developments and the potential model erosion rates can affect project planning and approval. Um delays to projects can be very very costly. Um and so non-compliant erosion simulations as well um apart from the need to get them in the first place, but

00:13:20
if they were found to be non-compliant, they also cost money and time. And I'd add that um getting the design right is important simply because um generally um repeating rehabilitation for a second time is often the most expensive rehabilitation that a site can do. So getting it right the first time is highly desirable. And overall what we do need is we need there to be not just accuracy but confident confidence in simulation accuracy when we work our way through these landform designs. When we look at the history of erosion

00:14:07
modeling we can see why essentially we haven't had that same pressure on accuracy of erosion modeling as we see when we look at mine sites. Initial erosion model development was strictly aimed at agriculture extended to grazing and foresting forestry but typically erosion modeling was used to look at changing practices. Occasionally maybe verifying that farmer Billy Bob Jones was likely to get a some sort of a grant from the US government um to um make it make him more inclined to do the work.

00:14:47
But we're talking an activity agriculture which relative to mining has a very low per hectare value and the cost of the work are quite low in comparison. And so overall, the impacts of low model accuracy weren't of great concern. And there was, I might add, a perception that low accuracy was about all that was achievable. And that that low accuracy was more than sufficient. So there was also >> there was also um an element um particularly in the agricultural setting where um there was the ability for

00:15:27
ongoing um management and maintenance of the land and I think it's one of the key differences in the mining context um that there's often this perception or need for a a set and forget approach or you you do rehab once and you don't have an opportunity really to go back and easily redo or rework work or or continually maintain land. And so the the level of accuracy wasn't really needed because there was that ability to go and do fixups uh over time. Yes. And that um actually in the US that

00:16:03
explains why when they moved from maintaining cropping land to grazing land their um target or acceptable erosion rate was nearly reduced to a third to account for that lack of um fix up. Now in terms of models, one of the concerns with models is that we want to see models used that are well validated that we have some confidence in and this means that they need to be have been validated across a range of relevant materials, surface conditions and mine sites and we will talk I will talk some more about that further on in this

00:16:42
presentation. The most commonly used ones are the revised USLE which is a two-dimensional erosion model empirical statistical and again you'll get more on that web water erosion prediction program um which is process-based two-dimensional and then Siberia which is process-based and statistical to a degree um but importantly it's a three-dimensional landform evolution model. I might add that there will undoubtedly be more models um thought about, trial and mentioned in the future, but for now

00:17:22
those three are the ones that are most dominantly used. Now the revised USLE which is the oldest originally released as the universal sil loss equation and they called it universal they claim the Americans because when they put it together they included finally all of the factors that affected erosion and therefore it had a universal consideration uh of the eroding environment. So it was re released oh gee about the 1960s late 50s early 60s and it was updated to the revised universal s loss equation in

00:18:03
97. The model is statistical in that it was based upon a very large body of um field plot data. As such, it generally fits the data, but for individual soils, its predictions can sometimes be quite poor. It does have limitations, including the very restricted range of soils that were actually tested. And we do find for soils with textures or properties outside that initial data set, studies have reported um errors in estimates of soil erodability by up to a factor of three or four. Now it does have quite simple parameters

00:18:47
and again that reduces its ability to cope with the enormous variety of soils and conditions. Um and it certainly has an issue with its consideration of slope length response which doesn't necessarily pick up interactions between slope length and erosion process. Interestingly, the revised USLE originally was going to be released with three separate slope length response uh curves to uh try to address that um smearing shall we say of a range of responses. But eventually for simplicity they just went with one and the model uh

00:19:28
structure does provide other uh limitations as well. Nonetheless, it's been the workhorse for many many years and um it will continue to be used a lot. Now, the water erosion prediction program um it came out in the 1990s. Um there was a significant program of field parameter collection. Um I'm there was essentially a circus of um scientists and probably graduate students that went around the US carrying out experiments and gathering data and probably improving the turnover of the local um

00:20:08
bars somewhat. But it has got underneath it quite a significant body of field data. Now it's a much more powerful and more detailed model. It's process-based. they it's described as single event but it actually runs on a daily time step because the Americans assume that you get one rainfall event a day at least for modeling purposes. Now um the detailed erod roadability parameters that are there are directly measurable though of course given the data set that web's based on it's got

00:20:45
algorithms to predict the parameters as well. the um there was some dumbing down of we in terms of its uh input files which was done to meet requirements way back in the 1990s that the model was expected to run for a like a 100red years of data on an XT computer for I think it had to run in 90 seconds with the computer sitting on the farmer's kitchen table. So a lot of stuff was chopped out. uh very few amongst amongst you will remember an XT computer but trust me they were incredibly slow. So

00:21:24
nowadays we'd wonder why all that those functions were were lost but nonetheless web gives very detailed output and you can even get more accuracy and detail if you can get hold of the more detailed parameter files that were were not originally released. They picked something very simple for use by their extension staff. Now, Siberia and now we're into a landform evolution model. And just for the record, Evan, landform evolution model. Yep. Um, well, it's a it's a it's a model that uh that has an input uh

00:22:08
elevation or DEM file. And uh the key difference I suppose between it and the other models is that um that elevation is then changed through time. So um uh what Siberia does is actually modify a surface through time in response to erosion. We or Russell um just can assume the same slope uh each and every time. >> Yes. And although there are versions of um Russell and also of WE that can be run on a three-dimensional basis, those versions have no ability to update a DT a digital elevation model input. So this

00:22:53
makes Siberia very valuable for looking at impacts of flow concentration um gully development depths of incision um it's a few years it was released a few years earlier than we but um the interesting bit with well there are much there's much interesting about Siberia um it can be the maths under it can be quite complex and Gary Wilgus who produced it actually won an EML Truog award for the best PhD in the US in the year that he um presented his thesis and um describing Siberia. Now it's conceptually processbased so

00:23:37
there's a lot of data used and some fairly heavy stats to get parameters. Um if it is being fitted to field data um you do need to consider the length of record. A lot of studies have used input parameters taken from existing databases which is um great if you say it quickly but there is and should be a concern whether those database parameters closely match the materials in the climate. >> Now Siberia Siberia doesn't actually take in rainfall data. That's one of the one of the things that we find is a

00:24:14
misconception of the Siberian model. Um it's actually fitted to time series data of uh runoff and erosion. Um rather and doesn't actually have a rainfall file in it. Uh so it makes it um uh actually the parameters uh linked to both material and the climate um rather than just materialbased parameters. So it's it's uh it's important that you closely yeah definitely closely match the materials and the climate that's um that's being considered. M and because there's so

00:24:46
many different ways to parameterize Siberia in general both in this presentation and the report we've um essentially viewed um Siberia's accuracy as having a lot of potential variation been a bit reluctant shall we say to uh categorize it. However, overall if we look at model accuracy um and there's much detail in the report especially for the revised USLE uh and particularly for it although we've suggested using plus or minus 50% people do need to be aware that larger errors are possible. In this case, we've used

00:25:31
that rating simply to flag that the accuracy is moderate at best and then you do need to apply more stringent modeling target targets. we see we with estimated parameters moderate about plus - 50% and uh that was my view and I was pleased to see Evan had a chatter with one of the the web team and um that guy who'd done some extensive review of their uh parameter sets was again talking plus minus 50%. We with measured parameters we'd say has high accuracy and we'll talk to that in a minute and yes Siberia we've got down

00:26:10
as highly variable accuracy depending on its parameter source. So we with measured parameters um now um I'll ask Evan to talk this because this is his data. Um, remind me Evan, that very highest erosion point that comes from a, uh, nickel mine I worked with. >> Yeah. So, um, it comes from from a nickel mine. Yeah. So, >> I'll I'll come back to that one. It, um, strangely enough, it it, um, I get some pleasure out of it. >> Yeah. Yeah, I suppose we've just talked

00:26:49
about there being um being um uh algorithms within web that allow you to estimate parameters and then um the ability within web to actually put in measured parameters from from lab or field studies. And so this this diagram or graph here shows um times where we've measured annual erosion within the field. So using a range of techniques and that might be erosion transcts or measurements from LAR pickups those sorts of things and we've also got erosion predictions using wet um calculated for the for the same period.

00:27:29
Uh so the data set here is over 37 locations over a range of different climates includes um arid zone and temperate range of different materials from um quite sandy or lomy soils all the way through to quite rocky materials um and a range of slope shapes. So this data set includes uniform gradient um slopes and also some concave um profiles as well. Uh so the line the lines there show the 95% confidence interval. Um and really it demonstrates demonstrated for us that that there is the ability through appropriate measurement to to um

00:28:10
be able to uh accurately assess um or accurately predict erosion um using the uh the web model. Um the the next slide, Rob, if you could move on, is is the same data um but instead of using the the the calibrated or the measured parameter set, we use the unccalibrated or the estimated parameter set. And I suppose the the message or the lesson learned here is that the use of the um internally derived parameters really provides the uh a a greater level of inaccuracy uh when looking at the predictions uh in in the modeling. Uh so

00:28:50
that was certainly a lesson when we when we did this exercise to to demonstrate how how wet and its accuracy can be impacted on on the parameters that you feed into it. Now, I occasionally used to scratch my head at why people were so intent on having predictive algorithms, but um you've got the issue that particularly for roadability, there's a view that measuring a roadability is too costly and slow. And because models were often used as extension tools, they might need to be able to walk onto a farmer's farm

00:29:29
and virtually carry out a calculation on the spot. So algorithms, however, have not been a raging success. And that's essentially because whatever algorithm is fitted can't consider all of the soil factors that affect the roadability at a given site. Um you find with the USLE that they make an assumption that erodability is uh driven by dispersed so particle size and strangely enough when they tried applying the algorithms to aggregated soils it all fell apart. So the problem you've got is that no algorithm is going

00:30:12
to consider everything that matters and so it can't work well and but that's makes experimentation a much more attractive option simply because you can go to the soil in question apply rainfall and runoff and find out exactly how it responds. There's also for the USLE a very restricted range of um textures. Um in general for erosion modeling and erosion model development, there's a real problem in terms of the shortage of baseline erosion measured erosion numbers that you can compare against to

00:30:52
prove whatever relationship or model you're trying to to work on. And you've got to bear in mind that with um particularly the USLE but also with some of the others the errors aren't just in terms of predicting erodability. There can be other factors and particularly for the models the variations in erosion process and slope length responses associated with erosion process are sources of error. So, um I understand why people want to use algorithms and in some cases have to, but um it really does guarantee that you

00:31:30
won't be all that accurate. And now we're to the first point to pull up and uh Jason, >> thanks Rob. Well, I think we've really covered a lot of ground there um on the background of the models, the different models that are available and their utility. Uh and then some of the levels of accuracy as well. So, some really good um content. We do have a a question from Emma Ryan Reed and thanks Emma for this question. It's a good one. Uh how how can Siberia address climate change

00:32:02
projections if it doesn't have a climate input? >> [sighs and gasps] >> Uh yes, you want me to take I I'll have a go first, Rob. Um >> okay, in that way that Emma might get two point two slightly different points of view and between the two we might get a good answer. >> Yeah. So I think um I think there there are ways for it to address climate change. the the the model uh requires um the the input or the the use of a data set of runoff and and um and erosion. Now, that typically doesn't exist really

00:32:40
anywhere um particularly if you've got a a mind landform that doesn't yet exist. Um so one of the one of the approaches that has been taken is to use other models such as web to generate uh that time series data of runoff and erosion and and by doing that um then fit them fit the Siberia model to that. So to address climate change, it is possible uh to use other models such as WEP to uh to do erosion predictions for a a changing climate scenario and then use the runoff and erosion outputs that it

00:33:16
produces to create parameters that represent that changed climate. And in doing so, you fit you're fitting that model to uh to that change climate scenario and then being able to to use that within Siberia to model evolution over time. >> Um Evan, you stole my question, my answer. >> That's great minds. Great minds think alike or fools never differ, they say. >> Oh, we differ a lot. So, we're safe. >> Thank you. Um, another question we can jump into is um, how much modeling

00:33:51
detail is enough? Uh, and when is it best to use a 2D model and is it necessary to use a 3D landform evolution model all the time? Um, okay. Typically um a 3D well often a 2D model is used as the initial working tool um to look at slopes and um to consider options. They're easier to to parameterize. They're easier to run. um often as a final look at a landform once you've got to the stage that you've got a um some sort of uh [sighs] um GIS person who's taken some vague statements about contours and gradients

00:34:46
and put it into some great monster of a land form. um that's at the point at which you would run Siberia in part to look at the issue of flow concentration, flow pathways and this is where Siberia is used to make sure that those things don't happen. Equally, if you're really worried about incision of a cover layer exposing encapsulated materials, well, you would definitely run Siberia. If you had a rel relatively lowrisk land form which you know might have been 10 or 15 or 20 meters high made of

00:35:27
something absolutely benign growing in an environment where if you stood still for too long you had grass growing on your boot you probably wouldn't worry with running Siberia at all. So you would consider risk of the landform and you'd list consider what material um or where did I lose that? Um you would have to consider what sort of uh concerns you had whether you were worried about incision um or any flow concentration. >> I think um there's some slides coming up Jason that um

00:36:03
>> that will um go through some of those things in a little more detail as well. >> Yeah. All right, excellent. Um, we we have one more from Melissa Salt and the question is, do you consider an unccalibrated model is still acceptable to compare options in a relative manner? So, I not accurate but precise. Thanks Melissa for that question as well. Cheers. Thanks, Mel. Um, and good morning to you. Um the unccalibrated models um have a have a role um at different stages of um project delivery. Um and they certainly

00:36:44
can uh give options for comparing things and in particular with for example the USLE you may just pull out cover factors and use in comparing things. that um there is a use there is a use and a place for unccalibrated models very definitely. >> Okay. Well, that's a good good point to um continue on. I think Rob we can keep moving through the slides. >> Now we come to the answer start talking about the question. So um basically when you consider the stage a project is at there can be situations particularly

00:37:31
early where a model of low to moderate accuracy is quite appropriate. Um for example in project planning you may not have any um capacity to go out in the site and dig up too much in the way of materials um or for that matter any sort of samples. Um it might also be where the landforms are very low risk and possibly simple. So there are situations where you might be perfectly happy with a model of relatively low accuracy such as an unccalibrated model. Now if we take that a little bit further and put it into the

00:38:12
context we would think that for planning the revised USLE we internally predicted parameters and if you must uh Siberia with database parameters would seem like an appropriate level of modeling to go into at that stage. Let's remember that if the project is still in the planning stage, it may not even happen. And so there is a a practical length in terms of time, e effort and cost that is worth expending at that stage. When you move to the point that a mine pro mine is actually seriously starting

00:38:49
to move thousands or more of tons of of cubic meters of material and starting to construct a landform, that's at the point where you would like to see greater accuracy because, for example, and Evan can talk to this one better than I can, it it's a point at which if you know what the the contours of your final landform are going to be, if you build to it, you can save an enormous amount out of earthworks. Evan did touch on that earlier. And similarly, you might use Siberia with calibrated

00:39:20
parameters, but in the first instance, you would certainly run web. And there was one exercise. Yeah. One exercise that we did a few years ago showing that um uh that there was a six-fold um difference uh in in the actual um cost of cost of moving things around uh when you went from the a poor poorly poorly detailed designing to using the actual detailed shapes. Um so it can actually be u quite a considerable amount of material and a cost uh that's involved in in detail construction. So that

00:40:00
detailed design is also is also needed at that point. >> Yeah. And then when finally we think about closure planning, there might be some uh review of um weapos or barrier modeling where you might look at parameter updates and maybe field validation of parameters if it's felt to be necessary. So the model that's appropriate obviously changes through the the life of the project. And as just a comment, I asked about this one because my thought was, well, having talked about project stage for

00:40:39
PRCPS and the like, what sort of project stages in Queensland, um mainly the um the subject of uh landform design using erosion modeling and most of the landform design studies being submitted are associated with PRCPS mostly from established sites and would probably largely be classed as dealing with um detailed design for construction. Um it might also for example be um sites which are well established but might be thinking about starting a new pit or a new land form that uh there's a significant number

00:41:22
that talk with to closure planning and some associated with new projects. >> [snorts] >> The point of interest for me was that in general, the modeling and landform design work that's going on is dealing with sites that should that have been established long enough that they should have uh reasonably uh significant background information. So lack of information on the site, it should not generally be a major issue. Uh and now we move to uh [sighs] one of the interesting bits. We talk about assessment of risk.

00:42:04
From the point of view of a modeler, risk is an important consideration. It affects what model the model is thought to be necessary because it talks to accuracy requirements and it does talk to target rates of erosion. So this is something that the modeler needs needs to know. Uh I'll talk through some of the target rates and you'll see more on that later. But assessment of landform risk particularly as detailed in the recent document um has potential to be a a major outcome and I think a major basis

00:42:46
for communication about landforms between mindsightes consultants and regulators. Now risk assessment's pretty obvious pretty clearly done on just about everything that moves and a fair bit that doesn't on minds sites. So having a compre comprehensive template for assessing landform risk isn't um a great change or [snorts] shall we say a totally novel idea. Now the uh people will undoubtedly try to um work out the lowest risk assessment that they uh can come up with. But um I think the method here is

00:43:27
re reasonable and should give sensible ratings in uh when used appropriately. And I'm sort of I'd like to point out the little kicker in terms of target rates that if you've got a high-risk situation um then the uh target rates are reduced and again that shouldn't be too surprising. The simple analogy to that is the speed slow speed limit past school. Just to give you an idea, this is covering off on what is in the report and talking about all of the issues to be considered. So, it's a comprehensive

00:44:08
list and some of those factors are quantitative and will require some reasonable data if you're going to assess them properly. Others are more qualitative, so there could be some debate or disagreement on the levels etc. there. But nonetheless, this list puts out a template, if you like, for sites to carry out a risk assessment and to collect sufficient and suitable data to address or to assess some of the issues that might be associated with their planned waste landfall. So in terms of integrating the uh the

00:44:49
issues, you'd say three or more high ratings indicates a high risk land form. But if you're three or more highest, we're all in um medium. Well, that's what you'd have. Now, a lot of emphasis on thorough assessment and that is definitely going to be part of deciding whether or not the risk assessment is acceptable. But the other thing is that this risk assessment process should identify key issues of concern and should focus mind planners minds on the management of those risks.

00:45:27
And I would even add that if um having recognized the risks and discussing the consequences of them which might be greater margins of safety in erosion targets or maybe a higher level of inputs required to manage risk then uh the M mind planner may well be encouraged to um look for options that might be more cost effective and reduce the perceived level of risk. Now we come to the next checkpoint. >> Thanks Rob and Evan. That's great. Um we've we're covering off on um I guess the need to consider the different level

00:46:09
of modeling you might need at a different stage of mine life. um how to consider the elements of risk and you've stepped through the factors to consider there and how to actually bring and describe how to bring th those risk factors together and integrate them into an assessment. Um so I'd encourage you to uh publish your questions on the Q&A. Uh we do have one here that says um how do we avoid confirmation bias where the models are used to show that a design works rather than using models to tell

00:46:42
us what the best possible landform design is? [sighs] >> Um um >> be a tricky one. More of a general. Yeah, I it's a bit more general because um I'm quite sure that there are many situations where a mine has a land form in mind and they I'd like a modeler to prove that the land landform they have in mind um is [snorts] going to work. Um if the modeler is worth their salt and um the data are accurate, if the landform isn't going to work, well, the modeling should show it. Um

00:47:22
probably you will recall um folk that I pointed out a high erosion point in the calibrated um web output. That point um related to a mine site where the uh rehabilitation manager had his preferred landform design and it did not want to accept the landform design I suggested would uh be more stable. he went ahead and built his design um and it eroded actually at exactly the rate I predicted which gave us a wonderful point to improve the regression on that um particular graph. So um I sort of a little uh concerned if

00:48:07
a model simply sets out to fudge everything to um prove that an inappropriate landform is more stable than it really is. um that's the fault of the modeler not the modeling. I suppose it also comes back to the question around um the parameters that you use and um and we do later on in the presentation talk about confidence and and um when it gets to the review process um having having sufficient information in the reports that we write uh that give the reader which would mainly be the regulator um confidence

00:48:44
that the the modeling is worth its salt. Um, so I think I think that um through through good reporting of what has actually being modeled um that confirmation bias could be could be highlighted um and could then be questioned as part of that review process. Um and I think yeah we'll uh as we touch on that maybe that'll uh be be a place where we can talk further about uh addressing things like confirmation bias. >> Yeah, thanks. There's a really interesting point and yeah, great

00:49:17
comment. Thanks, Emtt. Evan, uh, we have another uh question or perhaps a comment. Um, so I'll just read through it. So, bear bear with me while I run through it. But a mine site that is established will need to also consider the life of of the asset and whether the design is at detailed stage or concept stage. For instance, a longer life of asset most likely would be at concept level design phase whereas closure within a few years would require uh detail designs. >> Yeah. So I suppose the question there is

00:49:51
how does that then impact the models that get used would be my assumption. Yeah. >> Yeah. I think I think I suppose the the way that we're talking it's um you know the the the difference between planning and and construction really is that the the period you have until you are actually developing those those designs. I think um if you have a longer mine life then there would be some justification for using planning based models in that longer time period where you're still really planning and looking

00:50:23
at options for um you know even feasibility studies. those sorts of things using some less detailed models. But as the as the window window closes or the or I suppose the scope for adjusting things closes then more detailed design u may be required. I know there's some situations where uh I've I've seen uh rehabilitation might be a long way off but it's or or that it's got a long mine life but because the way the mine is developing getting access to that bit of land again is very

00:50:55
unlikely. So you've so now you're you're moving forward that period where you've got to actually have a detailed design. So I think that'd be um just consideration of those sorts of things as well that in terms of the the model that you want to pick in really are you at that point where you really need to have some some fairly high certainty uh in what you are going to about to to implement. >> Yeah. Yeah. That's a great point. And um we have about 9 minutes till 11 when we'll

00:51:27
pause. Uh so we might crack on through the next section, Rob, and see if we can cover those. That's my plan. So setting target rates for erosion modeling. Now that seemed to generate a lot of discussion and concern in preparing the report which puzzled me enormously because as you see on this slide um target erosion rates tolerable or acceptable soil losses or T values even uh was being talked about and they were being used as way back as far back for crop land modeling as 1947. Um setting target rates has had a lot of

00:52:10
discussion and there's been all sorts of approaches that have been suggested trial and with varying degrees of success. Now early US focus was on maintaining crop land productivity. A target a rate of 12.6 tons per hectare which conveniently is 5 tons per acre was adopted for deep fertile soils. They were never going to be fertile forever at that rate, but um they wouldn't hit the wall until long after everyone involved was uh had shuffled off the mortal coil and was unlikely to be um called to account for it. But that rate

00:52:47
was reduced to about 4 12 tons per hectare for shallow rangeand soils because again of the inability to plow out rules etc. Now the problem or the challenge for what we were doing was to look at finding an approach that would for the mind site situation that could be applied fairly clearly and consistently across a wide range of mind sites, climates and soils and whatever. Um so things like for soil formation rate um no too difficult and u impossible to get a measure of something that's slow and

00:53:26
in some areas the amount of deposition of dust was a complication. Um we really didn't think about um natural soil erosion rates because they're highly variable and um they don't have a lot of relevance really to the modeler who's consist considering creating a stable land form out of a large mound of um less than lovely unconsolidated fill which um even if it's eventually going to be stable is never going to be a natural landform anyway. Okay. So, um we really had to look at what else we could find or what

00:54:06
what approach worked. So, we specifically focused on preventing accelerated erosion due to development of rules and gullies and that was selected specifically to contain on-site erosion to the relatively slow interr. We were very aware that ifs or gullies form the change in eros erosion process gives you a quite dramatic increase in the rate of erosion. So that approach is quite consistent with other issues in the central in central Queensland area in particular as for example gully erosion's been clearly

00:54:47
identified as a major source of elevated sediment concentrations and runoff from the Great Barrier Reef catchment. So um equally for the mine sites we were addressing the erosion process of greatest concern for that catchment. Now interestingly it then also qualitatively uh basically exactly align with the current Queensland closure intent of no no or limited reals or gullies. And um in case you're pondering why we're so excited about real and guttery erosion, um this is slide shows old

00:55:28
research of mine published back in 1983 looking at erosion of um black cracken clay under a very large rainfall simulator. And uh so we applied simulated rain to a range of plot lengths. And as plot length and discharge increased uh at about 9 m plot length, we saw the rules development develop and sediment concentrations in runoff increased by about a factor of three or four. And that sort of rate of increase or magnitude of increase in erosion with riddle development has been reported quite widely. And certainly I measured

00:56:08
it for quite a different r quite a range of soils. So hence why we want to stay back at the nice low interrel level. Having thought through all that and with the information that we had about um the rates of erosion at which uh gullies developed. We've put together recommended target rates for modeling. And you'll see here firstly that there's provision for the targets to increase or decrease depending on the level of landform risk. So in other words, you could the target is a higher rate at a

00:56:51
low risk and it reduces to something quite low if the risk is high. The second point to be interested in is that we've included not just we've put in not just a slope average erosion rate, but we've also included a maximum peak erosion rate on at any point on a slope. And the reason for that is if we're concerned about initiation of rules and gullies, average erosion rates alone really don't tell a useful story. And the inclusion of peak rates at a any point on a slope is really quite

00:57:26
important. And note here also damn come back. If using a model or parameters such that accuracy is low or moderate then all of the above values should be h haveved. So we are think taking into account both risk and model accuracy. Now back to the slide that you saw fleetingly. Um, gee, we're gonna make it, Jason. So, averages and peaks. Averages should be reported for distinct landform segments. They shouldn't be averaged across landform segments that are distinctly different. And although

00:58:08
the revised USLE only gives averages for a slope, I'll just flag that it is quite possible if someone thinks hard to take the USLE data and calculate peak erosion rates at slopes from the data that you you obtain from the model. So the model target rates help the modeler set a design. They're essential for modeling. A design developed to to a specific maximum rate that does that is uh consistent with what's suggested has got a fairly good chance but is not guaranteed to achieve what the regulator

00:58:46
would like because what's built is not going to be the same as what's modeled. And however, if the initial modeling uses a target erosion rate much higher or higher than what we'd consider stable, then the chances of the final land form being successful and free of rules and gullies is basically zero. They won't achieve the regulatory uh requirement. >> All right, great timing there, Rob. So, as I mentioned at the start, uh the clock is just about to tick over to 11:00 a.m. So, we will observe a minute

00:59:22
silence to reflect and to silently remember all of those who served and died in war. So, I will just turn off all our cameras. Uh we'll have that minute silence and then we'll continue with our presentation. Thanks everyone for that for reserving that moments of silence. I'll now hand back to Rob and Evan to continue with the presentation. Okay, back we go. Now we come to um h the interesting bit. Was there a pause for questions there? Jason, did you want to pause? >> Oh, yes, actually. Yes, please.

01:01:03
>> Uh, yes, sorry, if we can go back, that would be great. Um, but just to reflect on the the target rates uh and the use of uh the consideration of risk when identifying an appropriate target rates, I think that's a really good way to uh have a basis to compare between your modeling scenarios. It's not really intended to be a a regulatory tool as such or a criteria. It's really uh about having something to be able to compare and contrast your model scenarios. I think that that that is a great addition

01:01:33
and it's something that we haven't seen before. So, it's it's it's an excellent component of this leading practice. Uh in terms of questions, um we do have one that is a short one. Is there such a thing as a stable galley? It's a tricky question. Thanks for that. Uh possibly. Um back in the midst of time certainly um I put together a um methodology for assessing gully activity and um identifying whether the gullies were stable or not and um [snorts] the degree how active they were. But um

01:02:21
delving through the deep recesses of my memory, I seem to recall that um some catchment studies found to their horror that gullies went through periods of nasty activity and then periods of quiesence. So um [snorts] quite often if you've got a gully that is um pretending to be stable, don't necessarily assume that it's going to stay that way. Um I think it was New South Wales saw con service set up a uh catchment study and their so-called stable catchment went feral on them about one year into the

01:02:57
trial with the gully just deciding to let rip. So um under certain circumstances maybe yes but um it would be relatively unusual. I think maybe in the mining context one possible situation might be is if the the the base of the gully starts um in sizing into much less erodable material. So if you've got like really blocky waste at the base, you might have a a feature that forms rapidly, but then over time because it's in sizing into less erodable material, the the development of that gully is kind of

01:03:38
ceased or stopped. And that's not really what you get in a natural environment. You often get like deep soils that that cut through deep deep profiles of the same material. Um but that's probably the only only place that I I've seen gullies like gullies rather than reals um stabilize. So >> yeah. No, that's excellent. Uh we have another one uh that is shouldn't the target rates of erosion be also subject to local or undisturbed natural erosion loss rates in the region?

01:04:12
Hang on. Sorry. Could I have that one? >> I'll just read it again. >> Sure. Shouldn't the target rates of erosion be also subject to local or undisturbed natural erosion loss rates in the region? >> So, I think it speaks to that, you know, consideration of >> the natural erosion rates across the landscape. You've got a whole host of problems and we get bit people have been brainwashed a bit that if we if say natural enough times everyone starts to feel good the the natural rates of erosion

01:04:50
occurring on a natural landscapecape. Goodness swear I didn't touch that. Um. Oh. Um, >> that's okay. Can leave it there and we'll >> No, no, no, no. Now we're fine. I was just worried that the I was freaked, Jason, because the um >> the what was on my computer wasn't coming through in the screen. Just leave it. Um, but the point here is that what we're building when we construct a landform on a mine site is not a natural landform. It's an unnatural land form

01:05:30
with very very different properties and therefore um it's not strictly or even closely comparable to what we see in a natural landscape. Typically the gradients are different, the materials are different. very often the amount of uh when you go down through the layers um what part on a on a mindsite landform can be awfully different to what's out there in a natural landscape. So uh motion on what the um uh finally m system was freezing. We might have information on what is happening out on a

01:06:23
natural landfall, but um we um can't actually use that in designing um what the um what you know what what we're going to do with a a waste landfall. Yeah, I think um Rob, you might have some lag issues there. Um with some terrible lag issues. >> Yeah. Um but I think yeah, with natural erosion rates, I think Rob Rob's right. Um particularly in really flat low gradient areas. I'm thinking like in coal mining, it's the natural erosion rates there are very very low. Um and so

01:07:04
you're having to then if anything, if you're going to go that approach, then start to find generally steeper ground that that might be like an analog. um uh so so then you're into different materials and and and the like. So hence the approach taken was really uh to really keep it quite simple in that there's certain rates of erosion uh that are known to lead to really and gullying um and and if we can model to that then uh uh then uh the design will have a tendency to then erode at at rates that

01:07:38
are that are not prone to reallying and gulling. Yeah, absolutely. Okay, so we've covered a lot of ground in terms of the background um and theory and now we're going to jump into um how to apply the check sheet and and assess an erosion modeling report. So over to you Rob. Okay. So, the re review repo approach and my approach to reviewing a lot of stuff comes from the time when I was getting an awful lot of um journal papers, scientific journal papers to review for about seven or eight separate international journals.

01:08:17
So, it was a hell of an imposter on my time. And the best outcome for me as a reviewer was to get something off my desk really, really quick. And the easiest way of that was to find what I would call a fatal flaw. A fatal flaw is something that was a failing in the work to the degree that um no amount of um editing or rewriting was going to to um solve the problem. Essentially, generally, it was a problem in the data. And so you had a fatal flaw which you could reject and say no you know don't

01:08:55
even edit this just give up go away and leave me in peace. Um so that is something that to be looked for with these with any report that comes in. Then there might be critical information gaps and you can put in queries about that and eventually you if you're working through that after various queries and you might come to a response as a reviewer where it's um a tick or a cross or a query and sometimes what looks like a fatal flaw may be able to be addressed with a limited amount of work but that wouldn't be certain.

01:09:32
But um the main thing here is that you're looking and you're thinking about what you're getting. Okay. So a broad strategy the first thing is ask yourself does this report address all the key issues and this means go through tick them off are they there? So the first question is are they there and then the second question is are they really there in other words have the points been addressed adequately and now this is where the review becomes a lot more challenging but I'd suggest thinking of

01:10:12
the analogy of looking at a tapestry and then looking for a loose thread to pull. There shouldn't be loose threads in a competent report. So you're saying are the references relevant? Are the validations credible for models or whatever? Um are the are all the decisions that were made justified by data? Is a decision process explained? Our parameters consistent with what was um discussed in the uh doc document associated with this presentation. Is the modeling output credible? Now we'll keep coming back to this point

01:10:53
that the report should present sufficient information in a form such that a reviewer can form a confident opinion that the report is credible. Now it's not the reviewer's responsibility to um get all that information. That's the responsibility of the person writing the report. So modeling strategy what should we see there in a report we should see project stage identified risks identified and the level of risk identified together with the level of accuracy that's required and that should

01:11:33
be clearly stated. So the risk assessment should be thorough and cover all the listed issues. Um the models used should be stated with together with the way in which they're applied you know and is the model selection appro appropriate and credible. If for example somebody decides to use we as a landform evolution model you're uh you're home and hosed and you've got a report off your desk straight away. But um these are all questions that you need to ask and to see what has or has not been

01:12:07
done. Next one please mate. So I'll give you some examples now. Um you might have a report that says we used a model some strange model from somewhere and it's been thoroughly validated references. So the first point there is that um if somebody wants to um check the model um it's the author should supply the supporting references. It's not the job of the reviewer to thrash their way around the internet trying to get them but the reviewer should then consider whether or not the um those particular references are

01:12:51
credible. So you're talking about are they independent? Have they been is the work carried out for Australian conditions, mine sites? And does it cover a sufficient range of soils and conditions to be um credible? A model applied and working really well for one soil in one condition um is all very lovely, but um it's giving you no con confidence that it could be applied across a wider range of soils. And then again, there should be information on accuracy of the model. And if it's no

01:13:27
information on accuracy, well, it's really not suitable for the mindsight application that somebody wants to use it for. And if we move on to risks, the next question, are they identified? It can a risk assessment may well become a somewhat um uh complicated and less than a small task but if it's done well the risk assessment should cover the range of points. So have all the risks been identified, adequately investigated? Uh is there analytical data? And then does the design process really address the risks? Um can it it's

01:14:14
can or does the report verify and demonstrate that the modeling approach will address those risks? Now I make the point here I'll talk about the um the fact that a a comprehensive landform design should um consider quite a range of issues and risks. And the next slide is a little bit of an old favorite and I apologize to those who've seen it too many times but this was a trial where the opinion of an ecologist was that the rehab was a raging success because it was growing trees. Now, obviously for those who wanted to see a

01:14:55
stable long-term um outcome, the amount of erosion and the lack of any other vegetation growing clearly was not consistent with a good outcome. So, having a simplistic goal or adopting simplistic goals for landforms and landform designs actually adds to the risk. It doesn't address them. Next one, please man. >> There we go. Thanks, Sen. >> That's great. Uh we're we we're a bit of a way into the checklist covering the fatal floor approach and the focus on risks and discussion on model selection.

01:15:38
And I think it's really key that uh we have that description of the modeling strategy and the rationale and the logic that sits behind that described in the report and we look for that that's there. But I think at this point Rob, we might just keep moving through the slide deck. >> Yes, please. >> Thanks. >> Okay, model inputs. Now we come to the challenging bit. I keep saying that it just gets harder. So this is where it's easy to be swamped by detail and to miss that your absolute

01:16:11
act the reader is being blindsided. Now this doesn't mean that somebody is deliberately set out to deceive at all. You might be effectively conned by the text but to a degree the person who wrote it probably equally was equally so. Um it might often it's merely careless text and so on. So you really need to work methodically through the points and see if that in this slide are they present. First question are the relevant materials considered? You'd be amazed how often it's that's not right.

01:16:46
Parameter sources, erodability sources, time periods for erodability parameters. Uh how was if there was experimental erodability measurements, how were they done? if it was database materials, were they demonstrated to be similar and climate file records for the modeling? What what climate file record was used relevant to the site or not? So, I'll give you some examples. Um, now what we're going to try to do here is to demonstrate some of the things that people can look for. They may not be

01:17:25
experienced at this but um try to ill illustrate the critical mindset uh that is applied. So first one couple of top soil stamp samples taken and a bulk sample was collected for physical characterization and erodability measurement. Um the question there how do you how do you go with that? Um all sound fairly straightforward. Well, um actually personally um I would have given this a fatal fail. My first response was stockpiles, which stockpiles? Um are the the particular stockpiles mentioned relevant? Were there more

01:18:13
stock piles that weren't weren't mentioned? Um so you would want to have information provided on stockpile volumes to give you some idea of what's required to cover the land form. You would want to know their locations. You would want to know um sample locations and you got the question if you've got two stock piles but you only took a bulk sample from one are you really sure that the two stock piles were identical? Um and could you be sure? Say for example, you had stock piles that might have been 60 or 80,000 uh

01:18:48
cubic meters. If we took one sample from each uh would we be inclined to believe that that sampling intensity was in any way meaningful or credible? Definitely not. >> And and even the sampling even the sampling method is important. Did we was the sample taken just from the surface like as a surface scraping from one spot or was it taken through the pro through the profile of the stock pile at at different depths? Those sorts of things would be useful information to know. I I can tell you that Landlock has

01:19:23
received samples from a mine site where someone went out and sampled a whole heap of stockpiles and despite instructions to the contrary at about uh 10 to 20 cm depth, they took a sample from each stockpile which was about 3 m deep in fact and uh so the samples got to landlock and were thrown in the bin and uh we had somebody go back and sample properly. So we first up you really need to be sure that um the sampling is adequate and there has to be genuine proof that the materials considered are the materials that

01:20:01
matter. So we go to the next one please and here we go again. This is um difference all surface sampling four locations across the site. Um so we're looking there um soils broadly described as gravity lonesomes gravel content 0 to 37% and they ran the USL nomograph. So, um, we're then thinking about are we okay with that? And the answer is no. Basically, a fatal fail again. If we're considering a 1us to 25,000 survey intensity, which would be consistent with moderately intensive use

01:20:48
of the soil at field level and detailed project planning, we would be looking at uh a sampling intensity there of somewhere between 5 and 25 hectares per sample. And if the the material was variable, you'd be going closer to five than to 25. So, we'd want to see a lot more samples. Equally, um, we really don't know what sort of sample they take. When they say surface soil, um, I [snorts] can think of times we've received something was barely scraped off the surface, let alone dug

01:21:23
in. Um, and then we got the the other question. Is there any way that we can be sure that um the um surface soil samples had any relevance to what would be stripped for vegetation. So you you would then actually just for fun ask them how the see if they had an omograph for grally soils because the rest of us don't. Okay, questions. >> Okay, thanks for that. Um, do we have any questions from the audience there? If not, we probably can just keep going. Um, it's there some great content there.

01:22:12
It's um it's excellent to get these um discussion on this. Hang on. We have >> someone >> we do have one there. Oh >> uh from Thomas Banggato. Thanks Thomas. Uh not every client for erosion assessment or reader of an erosion report is an expert. How can a certain level of quality be achieved in reports or erosion assessment work? So [laughter] yeah, I guess thanks Thomas. I think that really um speaks to like the idea of do you need to be an expert as a reviewer or can you go and

01:22:44
apply this checklist um as best you can? Well, the checklist gives you a starting point. Uh the stuff we've talked about today gives you a starting point and the text in the report does go through a whole heap of things that you could look for. Um just [snorts] based on um what's in the the document and what's covered today, um I can certainly tell you that I could see a a number of reports that I've seen would suddenly find themselves in the bin. Um but it what we're going to we're

01:23:17
working through and we of trying to give suggestions on how to think about stuff. Um we've totally agree that the issue of um uh background knowledge is a problem for the reviewer. But um short of providing advice and hoping that they get um um take a lot out of today. Um given the detail we've provided, we have to hope that we've at least given them a hell of a leg up compared to where they were. Yeah, that is it is a tricky part of the checklist and I think it is a starting point I think and that it's steps out

01:23:59
the questions that do need to be asked and um and provides a basis for uh querying those as well. >> Suppose it's also not it's not something that's unique to erosion modeling. It's in it's the same for any modeling, surface water modeling, groundwater modeling. the the review process um is is needing at least a level of a level of competence to to pick the the obvious the obvious areas errors to start with. I think one of the things that's being highlighted here by Rob is that some of

01:24:31
the some of the issues that are that are seen in the current um erosion modeling is quite obvious um or fairly fairly um routine or obvious errors that can um be picked by um by even the non-expert in the in the room as well. So >> yeah, for sure. And I think if we do that then we're lifting the standard dramatically. So that's that's a great thing. All right, we might move on. Rob. >> Okay, Sam. >> Beauty. And uh then we get to the other one. Uh the next bit and again I say

01:25:08
things get more difficult again. But here are some clues. The strategy in this pres presentation is very much outlined in this for providing help. Here is outlined in this slide. um we accept the the absence of hands-on experience and um there are there is the difficulty which Thomas was quite aware of. So we're trying to identify key or essential components of the of um experimentation. Try to at least identify the ex essential information that should be provided um and demonstrate key some key

01:25:48
issues and even um give a cheat sheet of um some ballpark data that could be used as well. So for next one, same place. Oh yeah, sorry you're ahead of me. Um with experimentally um [sighs] uh experimental uh experimentally measured parameters. So clear statement of methods, photographs of test plots should be available if required. Experimental data should be shown. Um the one thing I would hate to see would be where somebody said we done an experiment. This is the measurements we made and here's the model output. And

01:26:41
you think, oh h I think I didn't see something in there. Um so there should be experimental data shown enough to verify the work calculation should be stated parameter values reported and all methods that have been outlined in the very first dot point should be consistent with good practice and I know good practice is in the eye of the beholder but um nonetheless um it is important now I'll give you an example here um an experiment to get wet parameters. So interal erosion measured on the

01:27:23
erosion plot set to a gradient and rainfall applied for so long time runoff samples were collected and the effect of hydraulic conductivity the soil is estimated to be 95 ms an hour. Next slide. I'm going to I would normally have given a crowd a little bit of time to think about this, but let's move right on. So, what are the things that jumps out that jump out at me from that example? The first is the reference to the erosion plot. So, there's a statement there that there was no replication.

01:28:01
So, that's not good practice and not acceptable. So, that's a fatal flail fail basing um erodable erosion model parameters and design of a waste landform on a single unreplicated plot um is not good practice. Um we think look at the um rainfall rate um of 95 ms an hour which is not unusual but we see that runoff happened because timed runoff samples were collected which meant that um the infiltration rate uh was less than the rainfall rate. Can we go back please mate? Thanks. Well done. Good.

01:28:46
So we're in this situation that um we've got infiltration of some sort and we've got runoff and u yet the hydraulic conductivity is 95 ms an hour the same as the rainfall. Now if you use the Philip infiltration equation, a saturated hydraulic conductivity could become could be or would be roughly the same as the long-term rate of infiltration. But if we know that our long-term rate of infiltration wasn't 95 ms an hour, well then obviously that estimate of hydraulic conductivity is wrong. H which

01:29:27
is also a concern. But then if we think about it um we get another couple of concerns. First up just back again please. Thanks S. Uh first up we um we're conscious that we doesn't use the Philip infiltration equation. the one that it does use um and I won't go through it um but will typically have an effect of hydraulic conductivity somewhere between a third and a half of infiltration rate depending on initial water content and um possibly also um particle size distribution. So when we see evidence that the person

01:30:08
is trying to use an infiltration equation that isn't even the one that we works, we feel we have to conclude that they haven't read the web manual. Uh which is again of concern. And then we move to a final point that the infiltration rate of apparently 95 ms an hour into a soil and that is mentioned it's a soil is extremely unlikely making the data even more improbable. Um and if if infiltration rates appear unusual certainly an experienced person would insist on seeing photographs of

01:30:45
the plots under rain. though uh in this case uh given the magnitude of other problems uh pursuit of that information probably wouldn't be necessary. It's a matter of how many times or for how many reasons you really want for rejecting a report. Okay, next slide now Sam. So now we look at one and here we go. Measuring real erosion assessed by applying varying flows to the overland flumes which are set to a gradient. Runoff samples calculated real erosion rates were plotted against shear stress values. Yes, that's common. The

01:31:25
critical shear stress is 62 pascals line fitted to three data points. So again we ask well um how do you feel about this one? And the first thing again we look at is to say well um hm the highlighted words that jumped out at me again no replication. So not good practice not acceptable fatal fail. Um typically for infiltrations and measurements and flumes you would expect to see three replicate plots um not a single one. My second concern would be having an estimate of critical shear stress based

01:32:10
on a a fitted line that only went through three points. Um really not acceptable again. So um another strike and then this one also has evidence that the modeler hasn't read the rep manual wet manual as the sheer stress value is simply not realistic. And we can check that if we read the manual which will show on the next slide. See yeah here we have the cheap table. So this is in the user manual. Now if we look across texture soils KI KR and then TAC CO is short for critical flow shear stress for the

01:32:54
initiation of rhythm. And the values in pascals there actually go from about 2.1 to 5 pascals which is a long way short of 62. Um you will get higher values if soils are consolidated by wetting and drying cycles but um the value quoted in the previous slide is basically way off the scale. And this table does allow you to look at the values because there is KI is interreal roadability. KR is realer roadability. So this is an opportunity for anyone to check whether the parameter values that have been derived

01:33:34
are at least within the ballpark. Remember these are values that you get out of um the web internal algorithm. So you might view them as plus - 50%. The next slide please mate. Right. And again checking adequacy and so we don't just pick on people using web Siberia parameters and parameters developed previously for a similar material were adopted for this site sounds fantastic if you read it quickly and but if a practice nitpicker gets hold of it your first question is well parameters what parameters how'd you get

01:34:15
them um you would then say and if you got them from field data Tell us about the duration of measurement, rainfall during that period. Was it consistent with the long-term average? How did you determine similarity? What properties of the material were considered? Um, you're then talking about if there was a difference in annual rain amount between the two sites, how was that accounted for? Um, there are various ways one could do it. uh but you would be looking for to see someone having at least made an effort. So

01:34:55
questions. >> Okay, that's great. That's some really good tips and tricks from your experience, uh Rob, and it's it's interesting to see those. And these are real examples of uh reports that have been um circulating and and are out there in the public sphere. So >> um not verbatim. No, they have been there are elements of work that was done in those reports that has been uh shall we say rewritten hypothetically put in and maybe collected even collecting a few little um fails under one heading

01:35:33
where they might have been separate. So, uh um generally generally you wouldn't find exactly out out in the u multiverse precisely those words that I've been showing. >> Yeah. No, that's okay. Uh we do have a question there that is um is there a guideline on sampling regimes and best practices that can be can be implemented on site? Um the document prepared for the mine rehab uh commissioner that's the basis of today does have uh advice in there on um sampling intensity for stock piles

01:36:14
and um there's various um publications that for so survey give um guidance on sampling for if you're just doing a so survey for different purposes So there is information. Yes. >> Okay, that's great. Um I think we have one more coming through. Um I think that number you referring to Rob was uh is that where the paper talks about a minimum of one sample per 10 to 20,000 cubic meters? >> Yeah. >> Okay. So that's a that's a good rule of thumb I guess to use. Is is that

01:36:56
something that would need to be varied on a site by sight basis? >> Um >> or it's more of a blanket kind of just a general rule that's suitable. >> Um I would be interested to see the argument that would be mounted to um to change it. um you would be more likely to argue that you would um increase the intensity of sampling above that if uh you had reasons to believe that the material in the stock piles was highly variable. Okay. For example, um >> in terms of reducing it, um

01:37:42
the argument that we've sampled three stock piles and they were so uniform, so we'll stop, you know, we'll reduce the sampling just means that the very next stockpile won't be uniform. But um yeah, I I don't as I say, I can see instances where you would sample more rather than instances where you would sample less. >> Yeah. Okay. And then depending on that variability >> that's that's there and >> okay we have a a question from Peter Berghoffer. Thanks Peter. Uh the

01:38:13
question is what are the advantages of calibrating models with field erosion plots over lab erosion plots. >> Um >> might be one for both of you guys. Evan as well. um in the field uh you get to um get a longer term look at um the processes. You get to track the materials through a range of conditions and if there are any uh short-term change changes um due to wetting and drying consolidation um they're easy attracted in the field. The only thing you need to worry about, apart from

01:39:00
possibly uh things as arcane as having damned dingoes dig up the wiring or something, um is that the plots essentially can due to if you have high levels of erosion, the plots become exhausted of sediment, which means that their erodability actually changes. So if you can't if if plots are subject to high rates of erosion, you can't run them for too many years. >> Okay. Yeah. >> I think it [clears throat] also picks up some of the practicalities of constructing landforms. So

01:39:34
>> when you're building a large field plot, um you can you can um >> sort of see the say the impact of of ripping um different ripping techniques, those sorts of things. I think um even the the the role of vegetation, which I believe we'll get to next, but um you can even look at at quantifying with with the vegetation on site what that impact is on on infiltration and therefore runoff and erosion potential that they're they're easier or or more demonstrable within within field plots.

01:40:08
Um the the the lab plots really are are great because they you allow for um quite controlled and and um simpler ways of getting the the parameters that are that are needed are way more cost effective in terms of um getting the data quickly and cheaply. So that's why I think I think both of them have a role. The other thing that erosion plots can be used is that validation process. So, we talked about there being that closure planning phase and and um erosion plots if if set up properly uh should be able to not only give you like

01:40:42
an erosion rate, but you can actually get runoff rates and and even rainfall rates if coupled with the right sensors. So, that gives you a data set you can then fit models to and validate the the setup as well. >> Yeah. Yep. The one last thing that Evan triggered for me is that if you've got large field plots, you can apply treatments to those field plots that you cannot apply on a lab plot. >> Yeah. So, they're just not comparable necessarily. >> Oh, lab plots are too too small

01:41:12
sometimes. Yeah. >> Yeah. The scale issue. >> Yep. >> All right. Um, we we've had a couple of questions come in, but I might hold those off because we're plowing through the uh checklist and um beyond halfway, so we'll hold those until the end. Thanks, Rob. >> Okay. Now, I've put this in just to explain for those who might read reports that there uh in terms of considering the impacts of vegetation on erosion and on a landform design, the modeler or modelers typically would

01:41:43
be left with two approaches. One was where basically they've got next to no guidance to work off. So in desperation they simply um identify what vegetation level is needed to achieve stability for a particular design and you put responsibility for achieving that co level of cover back to the site. Um which is fine. In fact, they may even give the site different land form and cover combinations and the mind can decide which land form uh it thinks it can um successfully uh stabilize giving how

01:42:22
much beds they think they can grow. Um so there is a bit of a question there Mark there and um in that case a regulator looking at the design may well put a condition onto some approval that suggests that the mine get out and find out how much vegetation they can reliably go go so that that can feed back to the um uh landform design. Now, alternatively, a design report can or may focus on a particular ve vegetation cover level if there's monitoring to verify that that uh cover level is going to be reliably

01:43:00
achieved. So, yeah, the modeler has choices um but much of that is um based upon um whatever whatever information the modeler can get. Okay, next one. So we have a look at this one and again if you say it quickly this sounds quite um credible. Cover establishment of 70% including lead flitter and canopy is considered to represent a well-established vegetation community within the region. Right? But um [sighs] does someone old, cynical and grumpy believe that? Well, um, first response would be, well, please show me show me

01:43:45
the evidence. Um, it's really first up, please give me some verification, not just an assertion. And it is considered is that one is a red flag to anyone who's read too many reports. Um, that immediately fires up the desire to put red pen somewhere. Um then you've got the questions of leaf litter and canopy cover. H wonderful and is there distinction between canopy and contact because canopy cover can be very ineffective um on a steep slope. All the actions happening down where a flow is

01:44:25
roaring across the soil surface and uh leaves waving in the breeze 3 m above that don't seem to do much to help. So uh you'd want to know a little bit about that. And then of course you've got the the other key question. Is there any evidence that mine rehabilitation um is going to achieve the cover levels that are consistent with the region? Um again can you prove this is all really about um asking that the report lifted standards and avoid slipping in all of these unverified assertions. They need

01:45:05
to be need to be justified. Ah, oh, this is a goodie. I can think of one regulator that's going to be happy. This is in. So, simulations used wet parameters used for rocky waste and top soil. Surface contact cover veg and rock was of 90% has been measured. Um, again, well, it sounds like they've got some data. Sounds as if they got lots of cover. Must be great. But the question and the concern right here is that, uh, basically somebody's just been caught double dipping. Um, the effects of rock and erosion are

01:45:50
accounted for in the web input parameters. So not fair to go and then want to add them in again as part of a cover factor that you might use to reduce your um model output. So um this means really that um the person who did the modeling will have to go back to their simulations, change whatever cover factor they used and um with an altered cover factor not no longer considering rock, the mal model output may not look quite so rosy. But anyway, thus it goes. Next one. Ah, yes. We're turning now to climate.

01:46:29
And uh here we have a 9-year duration climate c sequence based on daily recorded rainfalls from the site. Um which rings some good positive bells, but is it good enough? And there would immediately be the questions um how do the rainfall amounts and erosivity during that 9-year period compare with the long-term average if they didn't well a how do you how do you demonstrate that they do they are comparable and how if there is a difference do you account for it? So um these things are important. Just go to

01:47:16
the next slide to demonstrate why. Now this is data from a longunning experiment in southern Queensland. Black Kraken clays southern eastern Darling Downs. Um what the graph shows is annual it's called soil movement but it means erosion. It's how much soil deposited in the contour bank. And you can see enormous variation from year to year. In fact, throughout that particular period of 14 years, there were only six storms that contributed 70% of the measured erosion. So, bearing in mind the dominance of large storms on erosion

01:47:55
amount, um you start to see why we can get enormous variation from year to year. So the the period that used a short period used um to look at um to or to use rainfall data does have quite a risk of being very far from the the long-term average. And you just to uh raise a few warnings um there does need to be a review of um how meaningful the 9-year period was, but there was very poor correlation in this particular experiment between [clears throat] annual rainfall amount and erosion. So, in other words, the model has to try

01:48:43
a bit harder if they're to consider. Um, so there we go. Um, >> that's that's great. >> Questions whether I like it or not. [laughter] >> We're nearly all the way through. Um, >> yeah, >> but it's it's good to talk about climate variability. think that is something that's it's a perennial issue that keeps coming up and it's difficult to find ways to the right way to manage that. Um and the discussion around vegetation effects and not being too reliant too

01:49:16
heavily reliant on that and being able to justify the numbers that are being put in I think they're they're really important points. Evan, did you want to expand on any of those in this rake? Uh yeah, look the the only thing that came to mind I suppose just around that climate variability is um is that it can even be more variable in arid in arid zones. So I know in situations say the the Western Australian golf fields there's there's periods of you know four or five years where

01:49:45
>> where essentially no runoff no runoff occurs um because the the rainfall the rainfall though it might occur it just doesn't doesn't run off and therefore erosion can be very sporadic. I've I've seen several cases where there's you know de a decade or more between between um erosion events. So it does it does sort of beg the question that uh that the length of that record does need to be uh does need to be considered. Um and more so not so much just rainfall volume but the actual erosivity or the the

01:50:17
storm energy or the energy of that rainfall is really really important um u because it's really what is more closely linked to to the the runoff and erosion uh is just the intensity. So you might have a have a storm at, you know, 20 mm storm that occurs over 5 hours versus 24 hours. You they're going to have different amounts of energy and they'll translate into different amounts of erosion. So these things are important to see in the reports that the modeler has thought that through and has has

01:50:49
addressed that. >> Yeah, absolutely. And thinking about where the site is and uh what the climate is there and talking to that in the report. Yeah, >> we we have a a question from the audience on uh climate um that's that's worth asking here at this point. Um so uh they ask what is the recommended climate duration to consider in the model in terms of duration I guess how long is long enough um for climate or how however long is short enough might be the uh the other way of looking

01:51:20
at that as well. Um I think um the the the meteorologists would say in the order of 20 years um as a minimum length um as as a minimum length of record. Um I know the Bureau of Meteorology looks at at defining a suitable period of about 30 years. Uh so I think it's in that sort of range minimum. So that's where your 9 years would be quite low. Um uh we we at landlord would typically adopt a adopt a hundred years and often that's because people like to um to get a feel for what say a one in a a one in a 100 or at

01:51:58
least some of those more rarer rarer events that are going to occur and what they do as well. Um so I think that's it's in that sort of ball in that sort of ballpark um uh that we we would consider. Um, I suppose that because also the uh the modeling is is targeted towards a long-term average annual rate. As you go longer and longer with your rainfall record, the average annual erosion rate starts to settle and doesn't change as much. So there's potentially not much benefit in going

01:52:31
sort of for really really long climate sequences because the that average long-term annual rate doesn't then actually change as you just add more more rainfall events into the sequence. >> Yeah. Yep. That's a great answer. Um we have another slide on model outputs, Rob, and then we might then pull up for a broader Q&A. Okay, this one's a little bit of an orphan here on its own, but finally we get to the output of a modeling and uh see what on earth the modeler is made of at all. So you're going through and

01:53:07
again you're checking that because these will be should be discussed. um were the targets adopted appropriate to the risks and the accuracy and then think do the outputs clearly identify the issues as being important. Um and at this point I would remind everyone of the advice from the guy who was the leading US erosion modeler for many many years and he said the first question that you ask when you look at model output is simply does it make sense and in this question looking at addressing issues and coming up with

01:53:48
conclusions the obvious point there is simply is the output and the way being used and applied making sense. So that that is something that you really need to bear in mind. Then you can think if a modeler can find on-site observations or data to pro validate or um justify the model output um then those are great. You wouldn't expect them. Certainly, I know if I go to a mine site, I'm always very interested to see eroding areas and seeing different materials eroding and the rates to try to get some feel for what happens at the

01:54:28
site, even to get a feel for the things that uh are basically the cornerstone of stability or or not and whether or not the model output agrees with that. Some sites are stabilized by vegetation, some are stabilized by rock, for example. Um then you look at if the model output is thoroughly interpreted, if it's been applied, nothing more frustrating to them than to see that a report that says we run the ran the model, here's the data and uh no further comment. You're thinking, yeah,

01:55:06
what do you make of this? You expect to see something. So you expect to see some interpretation. Um and you would even expect to see some discussion of you know perceptions of limitations to the data identified and are those limitations addressed. It might be that you know we've presented data that we're very confident for with in terms of the erodability of the soil but um information on for example vegetation growth is a bit sketchy and that might be a matter of concern or it might be

01:55:45
questions about the depth of soil. So um there are all of these questions that you can uh ask and try to think about. Am I happy with this or do does for some reason it cause me a worry and on to the last slide. There we go. >> Okay, that's great. Um, it's a good point to stop and um, I guess reflect back on the work that's there. I I really like the checklist. I think it's a fantastic way to um, take what's in that process that's done through the modeling um, and then turn that into

01:56:28
something that people can apply um, from different sides of the industry that are out there and being able to, you know, step through what what what needs to be presented in that. So it gives a consistent approach and methodology that can be applied by uh miners, by cult consultants and by regulators. It gives us some clarity in terms of the expectations that for a report what we might want to see. Um and it it should lead to some uh efficiencies in in making sure that the information is there and it can go through the process

01:57:00
more easily. um and ultimately better communication within land form design and um and more effective engagement in that in that process as well. So we have uh some questions from the audience and there's one here from Christopher right uh would models be able to accommodate change in weather patterns specifically extreme weather patterns. So continuing on that theme from the climate variability. >> Thanks Christopher that's a that's a good one. >> Um Evan you first maybe. Sure. Oh, I

01:57:40
love love love to um so yeah the the different different models are able to um so for example with with WE within the actual we sequence for the for the climate it um it actually has has every day with a different uh a different amount of rainfall. most often is zero. But when it's when there's rainfall that happens on a particular day, uh the model is able to uh then um give parameters that then describe the the intensity of the storm uh and the amount of the storm uh and the duration of the

01:58:17
storm. Uh so in that way the sequence then can contain a wide range of storm events and that would include extreme events and rare events uh that then u that can be modeled. So, so that definitely has has that ability. Often we um we look to model at a long-term average, but uh you can pick apart in more detail some of these outputs to then look at how how it would perform under particular uh particular range of events or extreme events um or what have you. Yeah, I'd add that with Siberia which

01:58:54
runs on an annual time step, generally there's not a lot of um facility to run a single um extreme event. It's possible to run Siberia for a shorter time step and um but you would basically have to model the event um with another model to know how you might alter the Siberia parameters. Uh so there I know there's been studies for example using Siberia that have looked at running the model for simulating the first year or two of a rehabilitation developing um a channelized landform and

01:59:36
then taking that particular um channelized landform and then running it for a different set of uh surface conditions just to see whether or not for example um if you get a lot of rules develop early whether they persist. So there's a range of things can be done. Um some of it is just how um energetic and uh inventive someone chooses to be. >> Yeah. No, that's great. Uh I'll move on to the next question from Emma Ryan Reed and that's about target rates for erosion. Um question is should target

02:00:17
rates be based on reference or analog within the local environment um to be justifiable? So yeah, is it justifiable to have target rates from your local uh environment? >> I suppose it's similar to the natural erosion rate um sort of conversation that we've had previously. Um, so I suppose the the response there is that um it all it all depends on on the natural erosion rate that you're you're wanting to pick. Um, so uh so for example in in coal where it's really really flat, the natural erosion rate

02:00:54
might be really low um where and the hills might be really um steep but um controlled by rock and would be would be very different. So, so I think pegging it to different uh natural erosion rates just has to be done with great caution um because um it may actually not reflect what you what you have on your on your land form. I know uh Emma probably more more um interested in the Pilbur region knowing knowing Emma. Um and we did a study that was um uh was reported at the mine closure conference at some stage I want to say 20 something

02:01:34
or rather 18 or 19 where we looked at acceptable rates through the Pilra and found that um that it was interesting in that case that the the natural erosion rates were actually somewhat similar to the the rates that generate um low lowing or low tendency for gullying. So um so we did do an assessment in there and some discussion in that paper might be useful to have a read for those who are interested. >> Yeah. In general the the issue is very much that uh the mindsite landform that's being modeled and will be

02:02:09
rehabilitated is not a natural landfall. >> Yeah. So taking um an erosion rate from a natural landform and setting as its target. Um I would question why you would do it. Um and why you would not simply set an erosion rate such that your unnatural landform did not erode at an unacceptable rate. Anyway, different point of view. >> Yeah, we also talked about uh risk in terms of erosion rates. Uh there's Thomas a question from Thomas Bamgartel uh which says which of the risk factors

02:02:46
is commonly the most sensitive from your experience? >> Oh thank you Thomas. [laughter] Um I don't think I could pick a favorite. Um I really don't. Um >> maybe sight specific. Would that be true? >> Very much so. Um, you can get very ugly materials. You can get people in central Queensland will blanch at the thought of having a tertiary waste sitting on or near the surface. Um, in some instances, in numerous instances, um, slope steepness and length um can be major concerns. Um,

02:03:36
you know, it I don't think it is very sight specific and the intent of the risk table is just to make sure that somebody goes to a site and are thorough in considering all the things that might go wrong. And I think it's worth considering it. I think um I think given that there's a there's a tension I think between what mining wants often and what what rehab and closure wants in that um mining often has a or operations I should say really has a desire to want to build things that are short hall and

02:04:13
and small that's so they want to they want to they want to go steeper and and longer and those sorts of things and I think there's often a tension that I see there so of for me often it often things like gradient and slope length are are issues and ultimately the overall height of these landforms um is a is a common one that um I come across that uh that is often u putting these landforms into the high-risisk category. >> Yeah. Yeah. We also have a question from Peter Bhoffer. Um he says uh hi Rob and Evan

02:04:51
in reference to question uh on table two regarding the depth of growth medium. So with knowledge that the bowen basin is running out of top soil it is specified as low risk with a depth of less than 30 cm and high risk uh oh sorry greater than 30 cm and high risk of less than 30 15 cm. If the material if the material uh is of replaced top soil is highly erosive and more so than the underlying material would having a deeper layer of top soil not be a liability. >> Um probably so uh one of the things here

02:05:28
is I would hope that that risk assessment should be applied with a certain amount of thought applied. um you know in that whilst it's suggesting something if a site has got data that says that putting a shallow layer of top soil gives us huge rates of runoff and erosion um well that should be in the the risk assessment. Um, you know, there is nothing about the risk assessment that says someone shouldn't include their knowledge and their of the site. And if they've got extra information, it it would be

02:06:13
totally shorting or ruining the process if you weren't to add that in. And uh I'm well aware in the bow and basin the uh the discussions of um top saw usage and the shortage thereof um are getting um shall we say [sighs and gasps] more and more pointy and uh more and more important. So I fully appreciate the reasons why Pete asked the question. >> Sure. Um yeah, Evan, did you have a thought on that or >> um No, I think Rob Rob covered that up. >> Yeah. No, that's great. I I think

02:06:51
there's yeah, some flexibility in there around uh those sight specific considerations and being realistic there is important. Uh so thanks Peter. That's a really good question. We have another question from Melissa Salt. Um you've mentioned average a number of times. The Bureau of Meteorology recommends the median is average for climate data whereas many consultants would use mean without thinking about it. Do you have an opinion as to whether mean or median should be used for average when it comes

02:07:22
to erosion? Um the point in part is that the the models have an opinion >> on what they need to accept >> in terms of the output they give you. >> Right. >> So um the universal solosid equation gives you a long-term average. >> Right. Yeah. Yeah, thank you Walt Wishmire. But that um and in terms of WE, one of the things it will give you as a long-term average. I might add that we gives you um one option is to get annual data and you can look at that annual data and

02:08:06
calculate a median if you choose. Um most people tend to pull up at the uh um having grabbed an average and run. Um now with Siberia you get accumulative erosion and that's you don't get an amount of erosion per year. If you were obsessive about it, you could run the model for one year, two years, three years, four years out to 200 years if you were seriously obsessive and you could work it out. But it's generally people will simply look at the cumulative um soil movement and um they will um

02:08:55
take it from there. And because Siberia is actually applying the same amount of runoff and erosion each year, there doesn't seem to me to be the same justification for using a median to encapsulate uh variance in the data. So u anyway, that covers the options. Um for for the word on climate, however, I'll pass it over to Evan. Thank. Yeah, thanks. I I actually always thought Bureau used mean actually um mean average rainfall is a very common statistic that it that it pumps out. Um I think um I think I think for for me so

02:09:38
long as you understand what it is you're trying to measure and um you're then using a target that that has is the same measure, then I I would be I would be happy enough happy enough with that. >> Okay. Okay. That's I think that covers that one off. That's good. Uh we also have a question from Michael Frankham. He asks, "Is it possible to apply WE and Siberia Siberia process to geomorphic landforms to achieve tolerable rates of erosion given concentration of flow and long steep slopes?

02:10:14
Thanks, Michael. That's a that's a good one. >> I'll I'll say yes and and leave it at that. How does that sound?" No, [laughter] >> please explain. >> Yeah, want some explanation. So, no, so certain geomorphic land forms um the the are are growing in their um in their uh desire or their desire to have geomorphic. I think one thing the industry hasn't yet done is to define what they actually mean by geomorphic, but it tends to mean more complex um shapes. So it move movement away from

02:10:48
simple linear um shapes and um and and to more threedimensionally complex shapes. Uh so yeah, there's tools that allow allow for those shapes to be created. Um but they don't actually have an erosion component to them. They often the ones that are commonly used are are trying to replicate the hydraulic um the hydrarology of the surface which may may be very different from the a natural u bit of ground because you're now dealing with a pile of unconsolidated waste rather than a rather than a hill that

02:11:21
might be controlled by a rock or what have you. Um, but there's nothing to stop you from taking a more complex shape um that's generated with a with another tool and then using um web or or Siberia to then assess the suitability or the erosion potential of of those complex shapes. And I think that would be a really useful thing to do as we move to into more complex shapes to have that have that check that what's what is happening isn't um isn't then leading to increased erosion cuz what often the

02:11:53
other thing that I'd say is that geomorphic shapes uh ultimately tend to concentrate flow over the over the landscape and we've seen in history that by doing that in mining contexts that's where gullies will always form. So if we're going to go that way, then we're not wanting to create landforms that um concentrate flows to such a point that the that the critical shears are are exceeded and um gullies and reals start to form that would that would end up being a failure of a landform.

02:12:25
>> Yeah. Yeah. Now certainly it is possible rather than use software to mimic a landform which may or may not be relevant in any case. Um it's possible to use web simulations to come up with slope rules where you might come up with a rule for a essentially concave slope being careful to make the rule looking at um not slope length but discharge versus gradient. And um there are people around like 3D data who can take that rule and they can make that into a complex landform and then having used that sort of um

02:13:06
software analysis and uh 3D skill to create a landform perfectly possible to then use Siberia to assess the stability of it. So yep the two models have a lot of possibilities and potential. >> Yeah. Okay, that's great. It's really good um questions and interesting discussion. So, we've have another question here um from the floor. How would you recommend addressing vegetation cover in modeling outputs with programs such as WE which have a vegetation growth modeling component? And a second part to that

02:13:42
question, do these veg uh growth components impact the validity of the results for these models? Um well certainly from discussion with some of the American scientists who put web together um our view was that we don't actually try to use the growth component. First up, >> yeah, >> it's it is possible to use the presence of vegetation and u living vegetation and with relationships to account for the presence of the vegetation in terms of uh infiltration and runoff. And it is then possible to use info on

02:14:26
the levels of cover uh that the vegetation produces at the surface to apply a cover factor to factor down your predicted erosion to account for the presence of that cover which was very much the u the response of to me years ago from the American experts. Now we have there is one site where in the last year or two we did field rainfall simulator studies and we measured a lot of properties erosion properties of vegetated slopes and we just ran the model for that and we alternatively ran the model considering that we're looking

02:15:04
at a bare soil initially factoring up for presence of edge and considering cover and the two approaches did give us um quite reasonably close um data and quite reasonable agreement. So to that extent we can say that it does appear that those strategies do work. Um and no I don't know of anyone that's held their hand up to run the growth model side of web. >> Yeah. Well that's great. Um I have another question for you. Uh so similar to the different types of modeling should measured data also be reflective

02:15:43
of reflective of the stage of planning and design and should uh small sample size in planning stage with additional data collected at detailed design or be collected at detailed design I guess. Yes. Well, um generally um some old chap who'd been in the mind mining industry since whenever and he said on reflection of his career he could never think of a project where he felt that they had uh overspent on getting samples analyzed. >> Yeah. Yeah. >> He he could think of many projects that

02:16:26
suffered from a lack of data. Um and I think the problem you have um is to say oh we'll do half do a half sampling or a very low intensity sampling now and we'll go and do it later and later for some reason either it doesn't happen or it's done differently so that the two data sets don't align. um and whatever it ends up to some degree people wishing to hell they just done it the first time and I don't think compared to many things that happen and compared to the costs of running a

02:17:05
landform design study um an extra dozen or so or two soil samples is um a trivial cost. >> Yeah. put putting that work in up front and spending the money I guess and um in a focused way to collect the right information sets us up for success I guess. >> Yeah. And but I suppose a little bit maybe a little bit different from Rob but um the at the at the early planning stage like when you're still looking at prefeasibility even trying to decide whether you're going to put the project

02:17:39
forward. You know, there might be cases where you do need to know what you're dealing with in terms of erosion risk, but you you can deal with a higher level of imprecision or inaccuracy. Um, so maybe in that case it' be a few fewer samples would be would be fine and and analogous to that would be exploration exploration drilling versus resource drilling, those sorts of things. Um, or so survey done at a reconnaissance level versus at a planning level. Um, so I think there's probably that to say, but

02:18:12
I think Rob's point's really important in that, um, the two examples I just gave. Um, they're always followed up with, um, with the more detailed testing. So, never do you never does the exploration drilling then not result in further drilling when you're actually mining. Whereas I think um in in the environmental space often we do fight um that battle that um additional sampling u is still required to to prove up and improve the accuracy or the the confidence in that data. >> Yeah. Yeah. That's excellent. So we uh

02:18:48
we're coming to the end of the question time. Um we have one here that is I guess reflecting on the different perspectives. So uh what is the benefit is what is the benefit of this work for consultants, mining companies and regulators? >> Um could you repeat what you said earlier again please Jason? >> So the question is >> no no no not the question there was a comment you made right at the very start. Yeah, it it does cover >> back through that, doesn't there? I

02:19:18
mean, the great thing about this work is that we can con consider that um you know, it's useful for consultants who are preparing the the studies and also for regulators who um who might be uh trying to assess um what what they're looking at. So, it helps set a standard there. I think also from the mining company's perspectives, I think it makes it clear as to what the expectations are. So, um, looking at a project scope, uh, approving those and trying to work out what's what's necessary [snorts] and

02:19:48
what's not, I think is, um, are all good aspects of that work. But, yeah, if you wanted to throw your thoughts into that. Well, I' I'd emphasize that what this can lead to is a consistent and thorough methodology for looking at um landform design that can be applied by miners, by consultants and regulators. So that each should be talking to the other and all within the same framework of expectations. And there's a fair bit of clarity coming out of this in terms of expectations. Um, and it's given a fairly clear

02:20:29
guideline of what to expect from a report. So, it tells uh consultants and mining people what should go into the report. Um hopefully it'll make life easier and they'll be u a few regulators with smiles on their faces um cruising through reports happy with what they see and uh uh not wanting to make the sign of the cross and get the priest in to exercise the demons out of the report. So it should give that consistency and and efficiency across the whole process. >> Absolutely. Evan. Yeah, I think um I

02:21:08
think the industry industry really values clarity in some of these sorts of um areas. So I think the the work that um Rob's put into producing this document um provides some clarity in this in this space around erosion modeling. I know it's even even now it's already cuz I'm a consultant. It's actually uh enabled discussions between the client and myself about what's the expectations that are actually there and and being clear on that allows um allows for the work to just um um happen with

02:21:44
more certainty knowing that what's being put forward is going to be going to be accepted and there's no holes missing and those sorts of things. So I think um I think that certainty is is really important and I think um it will also help improve our practices that you know in modeling some of those examples that have been given are are basically real examples. Um and so if if those sorts of uh uh practices are are reduced and we're getting an improvement in the quality of the work that's being done,

02:22:15
then um it's good for everyone because the the confidence in the modeling is improved, the outcomes are improved, the environmental outcomes are improved and the regulators uh have to um probably have an easier job at regulating um because they're uh they're look overseeing more and more success rather than having to deal with failures. So >> yeah, for sure. I think it really um just shows how complex it is to actually undertake these, but having a a basis to be able to um step through, you know,

02:22:48
what the expectations are in a step-wise manner is really um important. So I'd encourage everyone to download the the report, have a look in the back, um check check through the um have a look at the checklist that's there and um you know, see if you can apply your next projects that are coming up. But um any any final statements, Rob or Evan? >> Yep. Well, >> I was going to say two and a half hours of us is long enough. >> Okay, that's great. I wanted to thank you both for um bringing your expertise

02:23:19
and your experience into this document. It's been a great process and um I hope valuable for everybody out there. So I would like to wrap up and in doing that just wanted to provide some takeaways uh from my perspective. So really I think we can see that the model should be data driven and we need to accurately represent the materials that are present and the situations that are there. Um the erosion processes need to be considered properly. We want to look at um you know the development of

02:23:53
reels and gullies as well as a um as a key issue in that the model should be fit for purpose. There's no point doing you know more than is necessary but that they do need to be an appropriate standard to represent uh the erosion that's happening on site. They need to be matched to the stage of the mining from early on through to middle of mining and approaching closure. You have different different types of modeling there. um and there's the need to apply the checklist and ensure that all the

02:24:20
aspects of leading practice are covering off. So I would like to thank you all for joining us today. Uh it's been a great turnout to this event and finally I would point you to uh a survey that we are preparing. So your feedback is important. Um please scan the QR code on your phone using your camera. It will take you to a very short survey. gives us an opportunity to understand um you know what work is working, what isn't working, what we can do better in the next um seminar. So there were a bunch

02:24:55
of questions that we weren't able to get to today. So apologies for that, but um please keep the discussion going and feel free to email the QMRC inbox. There's um you know a link through our website to do that and we can follow up those questions with you. Uh, finally on our website as well, there's a mailing list that you can join to register events like this and others into the future. So, thank you. I hope to see you at the next QMRC event.