 I'd want to welcome everyone to our fourth webinar in our North Dakota Reclamation series. We're going to have a few housekeeping items to take care of, and Mike's got our slides up, so he's going to hopefully move it along as I go. And then we'll turn it over to him. My name is Natalie West. I'm a research ecologist with the Pest Management Research Unit in Sydney, Montana. I work on weeds and classical biological control weeds in range and wildland systems. And I'm looking forward today for our speakers, Mike Rowich and Brady Alred, and to talk about using technology to enhance reclamation. So a few kind of housekeeping reminders. We do encourage you, please, to put your name and where you're from in the chat. Make sure it's visible to panelists and all attendees. So we all kind of know who's in the community. I want to remind you that all webinars are reported and will be posted at ndreclamation.com. So if you missed something and you want to go back, the information is there. We do encourage you to use the chat to discuss any ideas that you have. However, please do use the Q&A to ask a question to our presenters. And you can have it answered live. We will have kind of questions at the end of the seminars just to remind you that that's how we do it. Each of the speakers will speak and then we'll moderate the questions at the end. So get your question in there and either live or by text will be answering them. So and also just a reminder that our next webinar will be next Wednesday. That'll be our final webinar on March 31st. And it will be reclamation to achieve the most bang for your buck. So please do join us then. So, Mike, if you can switch the slide. Just as a reminder, NDSU is a non-discrimination institution. And the rules are there for you to read and see. And then so the last slide. I just want to then introduce, just to keep us moving, our speakers. Our first speaker is Mike Rawicz from Rumble. And he is a geologist and geographer specializing in the implementation of geospatial technologies to optimize environmental problem solving. And with this company, his focus is on remotely sense data analysis, collection and visualization applied to environmental monitoring. So Mike's team specializes in aerial imagery and image analysis to understand site conditions, reduce uncertainty and help efficiently characterize and monitor environmental sites. So I'll turn it over to Mike then. And remember, put your put your questions in the Q&A and your chats in the chat box. Thanks, Natalie. Thank you for having me here today. As you introduced, I'm Mike Rawicz. I'm an innovation project manager at Rumble. And I'm excited to be here today to talk about how drone imagery data, artificial intelligence, satellite imagery and some of these other advanced technologies can be used to promote innovation and reclamation monitoring. So as part of my work, I spend a lot of time talking with reclamation managers. And there's a number of universal pain points that there just exists in reclamation monitoring. And this is sort of where we try to aim our technology to provide value on the projects. So making sure we're not just doing something cool, but also doing something really useful is key in all of this. So oftentimes, monitoring reclamation projects is expensive and resource heavy. It requires a boots on the ground approach. And it's very difficult to track changes over long time periods that reclamation projects often take between years to decades on many of these projects. It can also be tough to make really objective decisions. And also sometimes it's hard to proactively make decisions about where to improve and spot opportunities for improvements. So out of understanding these sort of universal pain points, this is where my team comes from. We want to develop a technology to improve some of these processes. And with my background, trained as a geologist and a geographer, wanted to find the best image mapping technology and utilize some new technologies become more available like AI and machine learning. And I like to say that we're trying to turn ecologists into superheroes, giving the regular ecologists the ability to do more than they've ever done before, reduce the amount of cost that takes to do certain tasks, reduce the amount of time and improve decision making over the work. So Ramble, which is the company that owns us, is an international consultancy. They operate in 34 countries with more than 16,000 employees, and that's allowed us to grow quickly. You'll see some of the examples here today from around the world. Over the past two years since we started this program, we worked on five different components. So it's a pretty exciting place to be working at. So generally, we go through this three-step process, and you'll see this in our different examples here today. But first, we select an imagery source. This can be drone imagery, which you'll see in some of the examples today, or it can be satellite imagery, aerial imagery, or even we're investigating currently the use of stratosphere for lens for capturing imagery data. Then we can analyze that imagery, obtain a baseline survey of metrics like vegetative cover, the aerial distribution of invasive species, here are some of the other examples you'll see today. And we can go back and monitor that year after year and show document progress across maybe sites. Because all this information is digital as well, we can automate some of the reporting that has to take place. And so sometimes it just takes a long time to write a PDF report. We can take what used to take weeks or even months and compound that to just a few days to get some of the basic information described to our clients. So this is all great. But let's zoom in and actually look at a few tangible examples of how this all works. We'll start with this mining client. This is actually in central Queensland in Australia. They have an aluminum tailings facility where they're monitoring vegetation dieback around the tailings facility. So in this case, there's some leachate that might be impacting the surrounding vegetation. And they're having to, they're regulated to actually look at this change over time and assess the environmental impacts from the mine facility as well as trying to understand what the environmental impacts are from droughts. The big challenge here is that this facility is more than 160 square kilometers. So the approach we took is a spectral approach. A little bit of basic physics here. We have the electromagnetic spectrum and with the naked eye, you can see what the wavelengths can basically see on the screen here. But there's a number of other wavelengths which are beyond what you can see with the naked eye and they provide specific insights into things like vegetation management. So the graph that we see here on the y-axis is a measurement of reflectance. So that's the percentage of light that's reflected back off of an object. So if it reflects back 100% green, that object will probably look very green. On the x-axis, you can see representation of the wavelength of the electromagnetic spectrum in the end of the years. So for a typical healthy plant, this is what a spectral curve will look like. So at different wavelengths, it will reflect back different percentages here. You can see for a healthy plant, the green bands reflects back pretty high relative to the red bands or the blue bands. And you also see that in the near infrared, this area just beyond where the red is here is also relatively high. In contrast, a stressed plant might exhibit a curve that looks similar to this. And so what you see here is that green is actually a little bit lower, but so is the near infrared. So there's a little bit of information that we can glean here in the near infrared and another wavelengths that you may not get just by looking at a plant or taking a picture with the naked eye. And so what we do is we use satellite-based measurements that take discrete measurements at these wavelength intervals, and then we can calculate indices based off of that and make deductions about how a plant is actually performing in terms of its health. So for the site in Australia, the key question was how do we tease apart droughts? And so what we did is we used freely available satellite imagery from the European Space Agency called Sentinel-2. Sentinel-2 orbits around the site more than once a week. And we were able to get 62 images that were cloud-free over a period of about three years from 2017 to 2020. We looked at these images and we actually took those bands and calculated an indices called the Natural Difference Vegetation Index. So this is using band combination of the red and the near infrared and green to kind of calculate what vegetation health is. And so what we see here on the y-axis is NDVI values and the higher that value, the healthier we can expect that plant to be the lower the value, the least health we can expect. You see four curves on here and what we did is we actually looked at four separate areas. We had two potentially or three potentially impacted areas. That's the green curve, the blue curve and the yellow curve. And then we had a reference area which had a similar soil type, similar vegetation characteristics, similar hydrology precipitation ratio in here. And one of the key findings that we had was very simple by just observing this curve that there was a drought that happens in 2019. We can see all of these areas dipping sort of in unison. What this helped to do was actually provided an important line of evidence for our client and trying to understand, are all of these impacts that we're seeing related to the mining facility or some of them related to droughts. And we're beginning, this is an ongoing project, we're beginning to be able to tease this apart. The nice thing here is that when we met with our teaming partner, the ecologist, it's actually on site in Australia. This was the feedback they gave us. They said, this is a really useful graph. We've actually been guessing. We just had a hypothesis. And until now we didn't really have any evidence to support that. So this gives us that big picture of looking at 160 square kilometers. And we're beginning to sort of understand how drought is impacting versus mining impact in here. So ongoing work that we hope to continue to live right on the report. For this next project, we'll move to a different continent and a different technology sensor here. This is working for a client in England called Highways England. They essentially manage the interstate system in England. And they're responsible for managing the soft estate adjacent to the roadway. So about 30,000 hectares, similar to maybe what DOT and the United States would do. So the problem they have here is they have a unique species called giant hogweed. And if you're unfamiliar with giant hogweed, it was first introduced as an ornamental plant. Now it's got amounts of the wild. And it's actually known as the UK's most dangerous plant. This is in other parts of the world, such as in the Pacific Northwest and New York State. And it can cause painful burns, even permanent scarring when the plant releases a sap. And that sand comes into contact with sunlight and moisture, it can cause a very severe burn. So hopefully this little boy can get out of the way here. There wasn't harnes, but this shows you kind of the scale of the plant as well. So working for Highways England, they had this 350 acre area. They were looking to have this test technology on. It's remote access here with only a few incorrect trails you can kind of see running throughout the site. And what we did is we used the drone to fly over the site and captured more than a thousand images of the entire area. We then stitched that imagery together into a cohesive map of the entire area. And it was very high resolution. And had a pixel size of less than one inch per pixel. So we took all this information together. And what you can actually see here is that the drone imagery is high enough resolution. You can actually detect some of the giant hogweed. That's these brown, patchy areas you see. So the next thing we did is we actually took an algorithm and trained the algorithm to identify the unique signature. So it looks for both the shape and the color of giant hogweed. And we're able to map the aerial extent of giant hogweed with a high degree of accuracy. The benefits here is that the drone flies over the whole site in just a few hours, keeps the ecologists out of way. It's very fast. And in terms of the image analysis, we had an ecologist look at this for several hours and try and point out where in the imagery released where they could find areas that the algorithm missed and they couldn't find any. That's not to say we may not have missed some of the areas underneath the canopy cover, but this provides us a really fast and accurate first glimpse at where the giant hogweed populations are. Not only that, but we're hoping that this can be a part of a five-year remediation program to actually go ahead and eradicate these species and prove that we're eradicating those over time. And the drone gives you the advantage similar to how you can cover 160 square kilometers where you can see this white box on the bottom of the screen. You can see there's individual plants that are sort of popping through and they may not be visible from the road. So this type of mapping using drone imagery and AI, you see the ability to quickly look over large areas, hundreds, the thousands of acres and come up with an estimate of what the aerial extent of a given species population might be. So this is where I say, oh wait, there's more. It's exciting because there's other species that you could do this with. And we're still kind of pushing the envelope about how many species we can actually detect this with. It's highly dependent on the species, seasonality and the actual visual extents of those plants as well as some of the phenology. But as you see an example here, many of you from out west will recognize this red grass as cheap grass. And we've developed a prototype for identifying cheap grass at one of our mining clients in New Mexico. We also have been able to identify frag mites for some of our clients up in New York state. So it's exciting. There's a lot of different opportunities to expand this on sort of a case-by-case basis at the moment. So the last kind of thing I'll present here is one of our ongoing development projects with well-panned reclamation where we're using drone technology to actually improve this process. Right now we're working with Dr. Michael Coran. Some of you may know him. He's in a number of societies recently graduated from the University of Wyoming with his PhD where he worked on spatially balanced design and optimizing field collection processes. So Dr. Coran's process that we're building off of so far has been to use sample point. This is a method photographic technique where just like the first on the screen here you walk out, take a picture and then you can take that picture back to the office and select pixels and classify those pixels as belonging to fairgrounds or much the same way you deal with align point intercepts. The difference here is that by collecting one of the bigger differences by collecting this digital photograph you can actually then take that back into the office and it's about seven to 10 times faster walking around taking these photos in the field. You can do the same thing with a drone and it's actually 25 times faster than align point intercept method. And you collect many more data points than you would with align point intercept method. Traditionally on well pads, you would have just one transect maybe on the well pad in the reclamation area and one adjacent to the well pad in a reference area. What we're able to do here is using Dr. Coran's spatially balanced approach. You randomly select one point and then there's an algorithm that automatically distributes the rest of the points across the site. He's taking a step further by optimizing the actual path you would take using what's called the traveling salesman algorithm. So when you randomly assign that first point and then it develops the spatially distributed points the image on the left here shows what the order of those points are. We can actually make it so that's the most efficient path possible here. This may be intuitive when you're walking but the real benefit is of being able to program the drone to walk for you and capture this data. So this is a screenshot taken from Dr. Coran's papers where we're actually looking at flying drone automatically around this well pad capturing both the reference area as well as the area in the reclamation area and then performing a sample point analysis on that. The big leap that we're trying to take now moving beyond this is to actually automate another part of this process. So we're taking this raw imagery which you would normally bring in the sample point and manually classify and we're looking at instead of just random points in this image we're actually gonna measure what's happening across the entire image using artificial intelligence. So this is a tremendous increase in accuracy because instead of trying to take small representative sample you're actually measuring the entire image and you're able to assign accuracy and measurement to that. Typically what you see is a trade off of scale versus accuracy, the smaller the scale you can actually zoom in and look at things the less you're able to get that overall picture and using artificial intelligence we can analyze every pixel and every single image and there's no more trade off on scale and you get a look at this very fine scale and you can do that across the entire site. So this is sort of what that first prototype begins to look like where you have images on the left that are actually predictions of the image on the right. And so you can see there's this purple flower implants in the center here and we're able to predict you know what's the aerial extent of that what's the aerial extent of bare ground versus some of these different shrub types out here. So very early stage but we're hoping to continue developing this further into the summer. So if you have well pads that you're trying to reclaim and you'd like to perhaps consider this method or test this method we're actively looking for those partners now to do this. So some of the conclusions from the work that we've done so far is that using these digital and automated results you can still you can reduce some of the bias that may be subjective. You know if you send a couple of people out in a pickup truck to spend the summer monitoring well pads. Some of them are going to take you know better notes than others. And so this helps make sort of a uniform uniform reporting results. You have one kid you know one person with a high school degree and one person with PhD in ecology they may record things very differently. In fact we found in the literature that two people with PhDs may record line point intercepts in different ways even though they're doing it just right after each other. As we've also shown this has the potential to significantly reduce the amount of time I spend in a field and for a given site. And so you can now cover more sites than you ever could before in a given time period. This is important when we're talking about the phenology of specific plant species. Of course very important to all this is the fact that we can help reduce the amount of time you spend let's say on a given well pad or a given area on a site and that allows you to answer bigger questions than you ever could before. The digital record aspect of this is also really important because we can take that data go back to the office and analyze it and if that data is scrutinized you can then have somebody else replicate that analysis and actually scrutinize this in a very defensible manner. These type of analysis are also very repeatable and they're designed to go back year after year and show change whether that be improvements or degradations. And the digital aspect also allows us to automate some of the reporting which is very exciting. What we see in the future is some opportunities for data-driven optimization of seed mixes. And so what we mean there is that by identifying let's say in well pad reclamation we can fly 500 well pads, analyze that data, identify what species were present and compare that to the seed mix and then decide well this particular species was only present on one well pad then you shouldn't be spending money on that in seed mix. And overall we're hoping that we can help bring all this together to make every ecologist like I said at the beginning of this and to a super ecologist give people the power to make more informed decisions and to address bigger problems than they ever could before with the same amount of resources. So with that I don't wanna take up too much time because I know Dr. Alred has a very interesting content himself. And so I guess we'll take questions at the end. I have my contact information on the slide here as well. Fantastic, thank you so much Mike. That was excellent. I wanna encourage everyone to put any questions they have for Mike or any of our panelists or Brady I guess in the Q and A section and we will be talking about them after Brady's presentation. So thank you again, Mike. And we'll move on then to Brady Alred. He is a rangeland ecologist at the University of Montana. He works with the Natural Resources Conservation Service to spatially target and evaluate farm bill conservation programs. So working with an excellent team of scientists. He has led the development of the rangeland analysis platform which I'm sure he'll tell you more about an online tool that empowers landowners and research managers to track vegetation through time. So with that, I'll leave it to you, Brady. Thank you, Natalie. And thank you, Mike, for that wonderful presentation. It's fascinating to see all the wonderful technological advancements that everyone is doing. It's just fascinating to think that the things we were thinking about and dreaming about 40, 30, 20, 10, maybe even five years ago are now possible. And so I'm excited to see where the future takes us. Today I'm going to talk about the rangeland analysis platform or the RAP as it's more commonly known. So let me get started here. So what is rangeland analysis platform or what is the RAP? The rangeland analysis platform is actually two things. The first thing that it is, it's a set of data or a group of data sets. And that's really the engine behind the rangeland analysis platform. There's two primary data sets that I'll be talking about today. Our rangeland cover data set, which measures percent cover of rangeland functional groups and our rangeland production data set. And I'll talk about those in a moment. The second thing that the rangeland analysis platform is, is it's a web application. It's available there on your screen at that URL, rangelands.app.app. And if I have the time at the end of this presentation, I'll give a brief demonstration of that web application. So the question, why the rangeland analysis platform? And Mike hid on it in his presentation. The rangeland discipline has a tremendous history in fact, we are the leaders. We wrote the books on plant sampling. We do that really, really well. But the one thing that our rangeland monitoring does not do is it, these methods do not scale. There are not enough, there's not enough time. There's not enough people. And there's definitely not enough money to monitor the rangelands at the level we would like to monitor them at. And this is a problem that the discipline has grappled with since its inception. And we've come up with various techniques to do our best in various sampling methods and sampling design and statistical analyses. But when it comes down to it, these methods simply do not scale. And so we can use technology, specifically geospatial and remote sensing technology to add to these methods. And I say add, I don't mean replace but to add to these methods and to add value to what we do on the ground. And so that's why the rangeland analysis platform was developed. So the data sets. As I said, we've produced a rangeland cover data set that measures percent cover for rangeland functional groups. That is produced at an annual time step. We've also produced a rangeland production data set which measures forage measured in units of pounds breaker, the common currency of rangeland production. And that is also available at an annual time step and at a 16 days time step throughout the year. And so I'm gonna talk a little bit about our vegetation cover data set. I'm not gonna get into the nitty gritty details. I can later on if you'd like, but we were able to produce maps of rangeland cover. And this is continuous cover. When I say continuous, meaning from zero to a hundred percent, a continuous measure of rangeland cover for the entire Western United States. So for a large geography. And we do this for five different functional groups, perennial forbs and grasses, annual forbs and grasses, shrubs, trees and bare ground. And as I said, we provide this data set at an annual time step, meaning it is available from 1984 to the present day. In fact, we just ran 2020 data last week and it is now available. These data sets are available at a medium to fine spatial resolution of 30 meters. And to put 30 meters in context, it's about the size of a baseball diamond more or less. And so if you're wondering how many baseball diamonds are across the entire Western United States, it's a lot. It's somewhere between five and a half and six billion. And so there's a lot of land out there. And for each one of those baseball diamonds or each one of those pixels, we provide an estimate of vegetation cover. Now, a lot of people ask, how did we make this? This is my one method slide for this data set. And I've packed it very simple and I'm not gonna get into the nitty gritty details. And so the first thing to recognize is that this data set in particular is based off of on the ground data, data that was collected by or through the NRCS NRI program and the BLM AEM program. And we used about 60,000 on the ground plots that measure range land cover, vegetation cover. And these have been collected since 2004 across the entire Western United States. So a good number of plots. Now, when you combine, now what we did is we combine that on the ground data with the satellite data that's been available for a long time. The Landsat missions that many of you may be familiar with, the first one was launched in the early 70s, but really starting in the early 80s, 1984 with Landsat five, the Landsat missions really took off. And we've had Landsat satellites up orbiting the earth every 16 days since 1984. And in some cases, we've had two sensors up like we do now where we had two satellites orbiting the earth. And that has provided a wealth of information for natural resource management, not just for range line management but just natural resource management in general. And so the other thing that's changed through the years is the computational power has increased specifically the rise of cloud computing. With cloud computing, we are able to throw massive amounts of computation at these satellite images and at these plot data that we just couldn't do before. And so what we did is we combined the on the ground plot data with the satellite imagery and using cloud computing and some artificial intelligence modeling, machine learning modeling, specifically neural networks. We're able to produce maps across this wide geography from the Great Plains to the Pacific coast and through time from 1984 to present of our range line vegetation-covered dataset. And I'll demonstrate that in the web app here shortly. Our second dataset is what is our range line production dataset or what we call our biomass dataset online. And that's a measurement of herbaceous biomass measured in pounds per acre. And similar to the cover dataset, we separate that out into two different categories, perennial forbs and grasses and annual forbs and grasses. And we also combine the two so you can look at herbaceous total. But what this gives is this gives an annual estimate of above ground herbaceous biomass from 1986 to present. And we also provide a 16-day estimate. So for any given year, you can look at the change in range line production within any given year and see how things, how production is increasing, how production is decreasing and see the change that's happening within a year on that 16-day time step. And similar to the vegetation-covered dataset, these datasets are available at a 30-meter resolution about the size of a baseball diamond. This dataset in particular was made us a little bit differently than the range line cover dataset. And we actually use a very long-standing, long-used model called a light use efficiency model that actually measures the amount of growth that is occurring on the ground. It measures the amount of sunlight that is being used by the plant and models that into production. And it does that for actually every day of the year and we just divide it up into 16-day time steps. The other thing is we modified this algorithm quite a bit and we were able to partition that production so we can look at how perennials are doing relative to annuals and depending on what part of the geography you live in, that can be very important. Lastly, before I kind of demonstrate the web application, just want to kind of end here with this last slide of some guiding principles of using maps and using remotely sense data. Many, many times people think or want remotely sense data to replace on the ground data. And I'm here to say that that is not the case and should never be the case. Really these datasets should be used together and should be used with other local information whether it's local data or local experience or other management frameworks. Remotely sense data and satellite derived maps they're just another tool in the toolbox for rangeland management and rangeland monitoring. And they should be used along all the other tools that we use. And in particular, they need to be used in a decision-making framework. A lot of times people think maps are just gonna solve the problem but we have to remind people that maps don't solve the problem. Maps just provide us information that help us make a decision. And there's some other important key notes there that I'd like to point out. Maps provide us the opportunity to look at the landscape, look across space but also through time and get a better idea of that landscape variability. Maps aren't perfect on the ground data isn't perfect. As Mike pointed out, none of our data collections are perfect and so it's important to keep error in perspective and use maps when they're helpful. If they're not helpful by all means, don't use them. And then also think critically about contradictions. Maps will oftentimes contradict what we think we know about a landscape or they may contradict other datasets or they may contradict other maps for that matter. Sometimes those contradictions are the maps fault. Sometimes those contradictions are our own personal biases and our knowledge of a landscape isn't as complete as we might think it is. And so it's important to think critically about contradictions when they occur. So lastly, and I have about 10 minutes or so, I'm gonna give a very brief kind of overview of the application and so you guys can see what it's like. It's a publicly available web application available at rangelands.app. And so anyone can go there, you can go there right now and this is what kind of pulls up. We have a landing page which provides information. I encourage you to check it out. Specifically, we just launched a new support page. And so if you click on the support page, it will take you to a different site that you can come back to that has articles, videos, demonstrations of how these data can be used and how to use the web application. So please check that out. We just launched that this week. And so I'm actually just gonna jump straight into the web application here real quick and kind of just give a crash test a dive into this to see what it's like. The rangeland analysis platform is just built on a simple Google Maps type interface. And so it's very easy to use. My children are experts at using Google Maps. And so what we have here on the right is kind of the map version of it, the map panel where we visualize these datasets. These datasets are very large because they cover large geographies and a large time period. And so we wanted to make, the reason we made this application was to be able to easily visualize and allow people to interact with these datasets without downloading literally terabytes worth of data. And so on the right, we have the map visualization. On the left, we kind of have the data panel where you can control what you're looking at. You can kind of toggle on the satellite layer with Google Maps toggle that on and off as you would any Google Maps type interface. What we're looking at right now is the perennial forbing grass layer for 2019. 2020 will be updated very, very, very soon. But you can turn that off and you can also look at our biomass layer. And what we're seeing now is the herbaceous or the combination of perennials and annuals for 2019. And I'm gonna darken that a little bit so we can see kind of a better picture. And so you can change the specific year. Like I said, we can go back to early 2000s and we can see how things look 20 years ago. You can zoom in to an area, to an area that a project that you're working on or an area that you actively manage and see that visual representation. One of the things we wanted to be able to do with this platform is to provide the ability to do simple analyses because although maps are great and they allow us to visualize things through space and time, being able to summarize that data is even more important because that's what can help us make decisions. And so we have the ability where you can upload a shape file or a polygon or you can draw a polygon and you can do a simple analysis. And so I'll demonstrate that right now. And I'm just gonna show a BLM grazing element in kind of central Montana, the Billings Field Office in particular. I'm going to turn the cover dataset back on just so you can see better the outline. And so it will load that polygon or that shape file. And what you can do is you can click on it and you can calculate the time series. And what this will do is it'll go out and talk to our servers in Silicon Valley and calculate the average for this polygon for each year. And so on the right, we have what we call the analysis panel. And what we're looking at right now is a time series of cover for this particular grazing and allotment. A little small, so I'm gonna make it bigger here and blow it up. And so what I'm gonna do is I'm gonna turn off a few of these things. I'm just gonna look at perennial forb and grass cover and bare ground cover for this particular allotment. And so what this provides us is this time series from 1984 to present or 1984 to 2020 last year of annual cover. And you can see that in the early 80s, this particular allotment maybe had a little bit reduced cover than what it is now. But for the most part, it's been mostly stable. Bare ground on the other hand has gone down through time. And I don't know the specific management practices with this allotment, so I can't comment on the management, but there has been some favorable precipitation through the years, which may have helped with that. And so you can look at the different functional groups and see this data, see this particular time series. You can do the same thing for annual biomass. And so what we're looking at here is a comment as the annual biomass estimates for annuals, perennials, and then annuals, perennials and combined for herbaceous. And so I'm just gonna turn that one on. And so this particular one, we can look at and say, through time, the production on this allotment has been increasing. In the 80s, in the early 80s or early 90s, it was between 500 to 800 pounds per acre. And now it's producing on the orders of a thousand pounds per acre. And there's annual variability, as you would expect due to weather and climate and other things. Again, I don't know the management behind this specific allotment, so I can't comment on any management changes that were made. Lastly, I'll highlight our 16-day biomass dataset. This provides 16-day biomass for any given year, including the year that we're in now. And so it shows 2021 kind of a near real-time estimate. Our last estimate was about 16 days ago in March 5th, and we're about ready to run. Literally today, we will run our newest estimate for the new 16-day time period. But you can go back and you can look, say, 2020, and you can look at what is that growth curve for 2020, and how did that production change during the year? And again, this is separated in between annual forbidden-grass biomass, perennial forbidden-grass biomass, and then the combination of herbaceous biomass. And so you can see that for any particular year that you're interested in, whether it's a dry year or a wet year, you can go and see what are those, you know, introsesimal dynamics. The last thing I'll demonstrate before I sign off here is we wanted to be able to make these data available so you can use them in reports, you can use them in your own analyses. And so if you are interested in downloading the data that go into these graphs, you can just click here on these buttons here and download them as a CSV file or as an Excel file. I'll just go ahead and download the Excel file and just to give you a demonstration. And then you can open that up. And for this particular polygon that we were looking at, it will give, in particular, the 16-day biomass is the one I downloaded. And so you can get this data for the entire time series and you can use it in your own charts or in your own reports or any of your own analyses that you may be interested in. Lastly, and I'll put my email address in the chat box when I'm done. If you have any questions, feel free to reach out to me directly and I'm happy to help you kind of walk you through this and kind of answer any questions that you may have. Oh, and one thing I also forgot is we also, you can download these data, the images themselves, the actual data themselves and the links are available on the website. So if you have any questions, I'd be happy to answer them. These have been really great presentations and so we do have some questions in the box. So I'm gonna start with some questions for Mike. So Brady has a little break. So Mike, do you fly well-planned individually or have the drone fly multiple pads in the area? For instance, can they scan multiple pads in one flight or is it just analyzing one at a time when you're thinking about planning these flights? Yeah, that really depends on sort of what field you're operating in. Some of the preliminary work has been done up in the Jonah Field in Wyoming, which is up 2900 well pads and they're in pretty close proximity. So it may actually work to fly multiple well pads just on one go. In other parts of the world, the well pads are a lot less dense and so you'll probably want to do one at a time. So far the preliminary work we've done, we've just been doing one well pad at a time. So that worked. Great. Okay, and so a couple more questions about kind of using these processes. Do you think it would be feasible to utilize publicly available satellite data for right-of-way monitoring at a field wild scale? So could someone go in and look at all the right-of-ways with the North Dakota mapped at field wide with like satellite imagery? Do you think that there's enough and how often do those satellites actually pass over, do you think? Well, you know, Brady I think jump in here but it's definitely possible to do monitoring. It really depends on what the question you're asking. Yes, I think Brady's presentation showed there's definitely some questions that you can ask and you can begin to understand that. Other questions you may not be able to answer like the invasive species, you're not gonna be able to necessarily see that as extensively on satellite imagery just because of resolution requirements. But as Brady's shown, there's lots of questions that you can answer using that type of data at that scale as well. Yeah, Brady, do you have something to chime in on that? No, yeah, I think Mike hit it on the head. It really comes down to the question that you're asking and using the best tool available for that. So if you're really interested in very local site-specific management, then you need to use those tools that help you do that. And from a technology perspective, that could be drones and that could be combining it with on-the-ground data. If you're interested in a more broader scale management or going back in time or continuity through time, you might use a different tool. And so it's fitting the right tool with the right question. So just as a follow-up for you guys, what if someone was considering what the best technology for the imagery would be, whether it was a drone investing in analyzing with the drone versus what's available in satellite imagery? What are some key ways that people can kind of move through that decision tree? Do you have some pointers for that? If people are considering what the best tool is? I guess I'll, you want me to start, Mike? Yeah, go for it. You know, I think it comes down to just, there's a lot that goes into it. I mean, our goal with the Rangeland Analysis Platform was to produce an off-the-shelf product that can be used for a wide variety of things, whether it's monitoring, reclamation, and across the course of wide geography. And so that is probably the initial path of released resistance, right? It's something that's already made and that you don't have to go and make something yourself. But again, that won't work for every single use case. It won't work for every single question. And so I kind of view it as a gradient where you start there and if that doesn't work, then you have to explore different tools and different technologies. But then it may be working with someone like Mike and his company to develop a product specific for your project that will work really well that's just not available off the shelf. And doing that requires expertise, whether that's through a company or in-house, you have to be able to do that and manipulate all these satellite images or drone images and take all the steps necessary to do it. So it's kind of a gradient from off the shelves to more complex and more specific. Yeah, I'd agree with that. I mean, there's sort of like a Swiss Army knife of different data sources available. And then the same thing when you look at how you analyze that data and sort of interrogate it and understanding the assumptions around that. We looked at Landsat as well as Sentinel-2 imagery today, which are two of the publicly available resources, but there's also all kinds of privately available satellites. I followed this guy on LinkedIn that's a space reporter, which is an interesting job title, but he continues to be reporting on launches of private satellites. And it's really just staggering how many satellites are in orbit capturing Earth observation data all the time. A lot of them you have to pay for, but there's just a ton of different options out there. So gotta formulate the right question and then you kind of dig down and see what you can answer with what's out there. Great. Okay, so to go back to the question and answer session, how quickly could a specific anomaly be identified with today's current technological abilities? So could have anomaly, so people, for instance, a spill or something else, could it be identified within hours a day and could identification of these potential anomalies be utilized to maximize effectiveness of field inspection or survey? So could you identify them and could they be used to feedback and improve our detection of anomalies? And maybe Harold can give us an example of an anomaly. That would help guide that answer. Yeah, I mean, I get think again of an example of sort of a way to think it would be really useful because the advantage of a drone is that it's basically an on-demands platform. So if you have a drone, it can come in a box that's this big and you can drive out to your site and pull it out, set it up on the tailgate of your truck, take off and you can have some form of data to look at just a few minutes later. But you're only going to be able to capture maybe a few hundred acres at a time with most systems that are out there. So spills, yeah. So you may be able to document something like that and then the quality of data is another thing because there's different methods for sort of post-processing drone data. You like the invasive species map that we looked at in England. We took more than a thousand images and crossed that to a cohesive worth of Mosaic map. The same time, there's a lot of utility just having a drone and being able to take a geospatially taxed photo of a location. Then on the other end, you've got satellite imagery data which can be at different intervals. It could be, you know, there's some platforms on the private side that advertise daily capture. You don't have to pay for that but it's going to be a lower resolution than the drone. So this is sort of that decision tree that you were looking to earlier is how often, what quality you need, what's the spatial resolution that you need to incorporate spectral characteristics as part of that as well, kind of teasing those things about it that day. Yeah, if you're looking for that day, just seeing Harold add a little information in the Q&A. Usually the most effective means would be a drone for depends on, I guess, the area you're looking at. That's really hard to turn around satellite imagery. I don't know of any way to do that within like a same day kind of method. Yeah, I think there was a question about whether changes in surface vegetation, standing liquid and significant changes of surface moisture be identified. So it sounds like it's plausible but not in the same day sort of thing. Unless you're there with a drone in that day, you're probably, it's unlikely to be that immediate or to notify within a daily time period of a spill alert or something like that. Is that what you're saying? Yeah, I think you'd have a hard time doing that. There's certain emergency response situations where they can fly a satellite over and I'm talking about by perking response or something like that, where some of the private companies might be able to fly over a site and have the data downloaded maybe just a few hours later. Usually that's for like the circumstances where I've seen that have not been on my projects. They've been for like the National Geospatial Intelligence Agency. So above my pay grade officially, I'm getting the service. Great. So Brady, do you have any tutorials on using your rangeland app in a classroom? They are coming. And so I highlighted that new support site and on there we will be, we are specifically developing and we'll put on there kind of a curriculum on using the rangeland analysis platform in a classroom, in a lab setting. And something we will have modules and lab exercises that teachers, professors, they can use to help teach this with their students. So it looks like we've answered most of the questions in the Q&A box and Brady and Mike have provided their information. So I'm sure they're willing to follow up with you. If you've got further questions, I would like to ask though our panelists each if they could give us a brief kind of final thoughts, take home message for all our listeners in the internets that you would like them to take away for their practices today. We can start with Mike. Sure, yeah. So Brady, it was fascinating seeing your research as well. So thank you very much for sharing that. If I had one takeaway sort of eludes the one of the thoughts that actually Brady brought up which is I have one of our clients had this quote and I wrote it down and I say it all the time, we're making maps that we've never been able to make before. And it almost gives me goosebumps when I talk about it because my job that I do right now didn't really exist even just five or 10 years ago. And now there's this demands and the ability to service people with the services we provide. So really exciting time and it's changing very fast. Yeah, Mike hit it on the head again. It's very exciting. And I guess I would just encourage people to think how can we use this type of information in our workflows, in our frameworks, our decision-making frameworks and where do they fit in and where do they not fit in? Those are the questions we need to be asking is how do we learn to use these types of data? And once we start asking those questions, these data will be incredibly useful and there'll be a large return on investment. So thank you everyone. Thank you very much. I just want to remind everyone that all of the presentations are recorded and they will be available at ndreclamation.com. Thank you very much for joining us. Next week is our reclamation to achieve the most bang for your buck. So follow up there. I want to again thank our panelists for this very interesting webinar. And as I said, their contact information is available. So please feel free to follow up with them. That sounds like they're willing to listen and to provide you some feedback. So thank you again. And I hope you all have a fantastic day.