WEBVTT 1 00:00:06.000 --> 00:01:06.000 Speaker A: This episode of GeoMob is supported by the Open Geospatial Consortium or OGC. If you work with maps, location data, digital twins, earth observation, mobility, AI or really anything geospatial, there's a good chance you're already benefiting from work happening inside the global OGC community. Even if you don't realize it, OGC brings together people from across industry, open source academia, startups and government around the world to work through the practical challenges of getting systems and data to work together better. They recently launched Individual Membership, which gives independent developers, researchers, practitioners and generally curious geospatial people a way to get more directly involved in the community. Conversations with working groups, events and collaborative projects happening across OGC. You can learn more@ogc.org now onto the episode. 2 00:01:06.000 --> 00:01:47.000 Welcome to the GeoMob podcast where we discuss geo innovation in any and all forms, whether for fun or profit. Hello and welcome Back to the GeoMob podcast. Before we begin, I have a quick ask for you, the listener. If you enjoy the GeoMob podcast, can you please leave a five star review on on whatever platform you use to access the show? This really helps us out. Right now, on with this episode. Today we're talking about reintroducing beavers with our guest, Davron Smith from Natural England. Hi there. Oh, sorry, can we maybe start with introducing yourself to our listeners, please? 3 00:01:47.000 --> 00:02:24.000 Yeah, absolutely, yeah. Hi everyone. Hi Alistair. I am a principal data scientist at Natural England. I've been here for, yeah, a few years now, but yeah, no academic background. Sort of, yeah, all things environmental really. Environmental science, environmental consultancy. And yeah, I've been doing things in the GI realm since leaving university longer ago than I'd like to admit now, but you and me both, yeah, done a few interesting things in the engo sector and then, yeah, moved into, yeah, Natural England a few years ago now. 4 00:02:24.000 --> 00:02:38.000 Oh, that's really cool. So just before we get into the sort of topic of the podcast, do you. Do you do quite a wide variety of spatial stuff and environmental stuff at Natural England in your specific role? 5 00:02:38.000 --> 00:03:29.000 Yes, yeah. So yeah, I won't speak to the entire organization, but yes, certainly within my role, yes. So as principal data scientists, I actually sit over a team of primarily species modelers, so everything from great crest newts to bats, dormice voles, all sorts of things really. But then actually my role does stray a little bit further outside of that as well, so into things like modeling nutrients that are sort of reaching watercourses. So yeah, modeling and air quality and recreational disturbance as well. The number of visitors that are sort of likely to be visiting any protected site trying to model that as well. So yeah, it's actually quite a diverse mixture of different GI projects. 6 00:03:29.000 --> 00:03:57.000 Very cool. We had a guy, Tom Foxley from University of Bath a couple of weeks ago on the podcast and he was talking about modeling landscapes for bat populations. So sort of sounds like there's some sort of crossover there in terms of what you're doing, what he's doing. So that's really interesting. But today we're talking about beavers. So I guess first up, why are beavers important? 7 00:03:57.000 --> 00:05:33.000 Yeah, beavers are really enigmatic creatures, aren't they? Yeah, it's interesting that and thank you very much for the invite onto the podcast. But it's interesting of all of the sort of different pieces of work that I look at that the beavers was the one that potentially caught your attention and hopefully will sort of capture your listeners imagination as well. But yeah, beavers just in, in general provide a wide range of benefits for the environment from restoring natural processes, reconnecting rivers with floodplains, and I think there's term quite commonly used about them, ecosystem engineers, you know, building dams and streams, creating wetland habitats and things like that as well, which obviously has the knock on impact of, you know, potentially reducing downstream flooding, improving water quality, filtering sediments and increasing resilience as well in you know, changed environments that are being impacted by potentially climate change, etc. So and typically the, these wetlands that beavers create are often quite, well, very biodiverse, very rich in wildlife, be it plants, insects, the fish, mammals, birds, etc, that use the wetlands created by beavers. And yeah, there are lots, lots of these different species which thrive where beavers are found, you know, from, from those bats that we were talking about before through to water voles, willow tits, etc. That rely on the dead wood and sort of wetter habitats that the beavers can create. Really? 8 00:05:33.000 --> 00:06:15.000 Yeah, I think I'm getting a bit of a reputation within the podcast hosts of being the guy who talks about ecology quite a lot. But I just find there's so much going on in ecology at the moment with all the sort of nature restoration and rewilding types of things or whatever you want to call it. And all of that has some spatial component and it's using spatial data sets and what have you in order to really understand either where the species are, where they might go or what the habitats are. So with that in mind, could you maybe explain to the listeners what the beaver consideration assessment toolkit is? Because that's sort of what caught my eye initially and maybe say a little bit about why it's been developed. 9 00:06:15.000 --> 00:07:14.000 Yeah, absolutely. And I'm glad it did catch your eye. Yeah, this was a number of years of work, work really, to create what I would. The actual technology underpinning the tool perhaps is not groundbreaking, but what we're trying to do with it is fairly novel for a tool like this. From Natural. From Natural England, really. And the important thing is about it is that it is this publicly accessible tool developed by Natural England in collaboration with the Environment Agency. And it's designed to help indicate where there might be potential impacts and where it might be suitable for beavers to be reintroduced into watercourses. So, as I say, it's taken multiple years, but we've pulled together lots of different data sets to make this possible and hopefully give the public and people interested in potentially releasing beavers sort of a first, first port of call in their journey to understand what, what might be impacted by a beaver in any given place in the landscape. 10 00:07:14.000 --> 00:07:40.000 Okay. And, and so, so how, how interactive is it? If I, if I don't know, there's a small stream not too far from where I live. If I was to look that up on the, the toolkit on the map, would I be able to sort of zoom into that area and then put a dot on the river and it would tell me how big an area would be impacted by a beaver dam? Or is it giving me information about stuff that's there sort of thing? 11 00:07:40.000 --> 00:09:03.000 Yeah, no, it's the latter. So you would absolutely be able to zoom into your small stream. And we've used quite a. Well, I should say, actually it was the University of Exeter that did our initial modeling for us that have used quite a detailed river network data set which you can zoom in on. And it will probably have your stream and it will tell you things like how likely a beaver is to dam in that area and it will also tell you all of the human, so things like abstraction points, but also some of the sort of natural factors that occur there, be it ancient woodland or whatever else where it might not. It does not claim to tell you the exact impact a beaver is likely to have, or indeed that area. We, We've, we've stopped short of that and we might get on to reasons for that later, actually, but we've stopped short of telling you exactly what the impact would be and exactly how widespread, but it would give you a good idea of what's in the area so that you can at least think about some of those potential Impacts and there is quite a wordy guidance document that goes along with it because beavers are. Yes, although it's great to visualize it and great to get that information into your hands, there is quite a lot of thinking that needs to go on around actually interpreting what you see in the tool. But it is, it is a good first, as I said, first port of call in answering some of those questions or at least getting you to ask the right questions potentially in the first instance. 12 00:09:03.000 --> 00:09:15.000 So in terms of a beaver, that would sort of be along the lines of is there the right type of woodland, is it the right size of woodland, that sort of thing. It slopes too, too steep. 13 00:09:15.000 --> 00:10:08.000 Absolutely right lines, yeah. It sort of looks at it from both directions actually. So beavers are pretty well generalists, but you do still have to. Obviously they do need to forage and the habitat does need to be suitable for them. Things like the, the slopes of the banks. But so it's looking at it from what is suitable for a beaver, but it's also looking at it from the side of things that where at ancient woodland actually could be impacted in ways that might not be particularly favorable for, you know, if it's a protected site, potentially there are communities there that might be impacted by rewetting of that woodland. So it's really a mixture. It's both the opportunity and some of the considerations that you might need to have for the beaver's welfare and the sort of the features in question that are already at the site. 14 00:10:08.000 --> 00:10:53.000 Okay. Okay. Yeah. So I, I guess then it's. It. There's a couple of things that are coming into my mind as you're describing that. That one of which is your choice of data sets is going to be really important in terms of understanding what it is that is going to impact. And, and because you've got those two focuses, you've got the focus on the beaver, but you've also got the focus on the impact of the human environment, as it were. Then I guess you can't just chuck sort of data sets at, at the model and hope it all comes out okay. You've got to really think specifically about that. So could you maybe talk a bit about how you went around choosing your data sets? 15 00:10:53.000 --> 00:13:24.000 Yeah, absolutely. And I think there is the old sort of, well, the saying that 80% or more of a GI analyst job is more about clean or any statistician or whatever is more about cleaning the data than it is the actual final process that you, you can often click run on that and it's done in Five seconds. But actually sort of cleaning your data and getting into the right state is one of the harder parts. So in this case there was that element, the usual data cleaning, but there was actually this other quite interesting element where we were working with species specialists, we were working with the Environment Agency, we were with universities, we were working obviously with internal Natural England folk. And we had to try and build consensus around which data sets were suitable and which were not, which could be interpreted in the right way and how to interpret them. So we had a few constraints. We wanted this to be a national scale tool, so we didn't want to overwhelm with potentially hundreds of really detailed data sets that we only had information for, either patchy information or we only had information for certain areas of the country. So we needed it to be national in scale, we needed it to be relevant. And there had been some really great reviews around the factors that beavers influence. It was a Natural England report that was sort of produced a few years ago now. So we had an idea from that. And we also needed the data quality to be of sufficient standard, and that means from all sorts of angles, actually. So there was lots of discussions around what good enough data quality looked like, and then the resolution as well. It needed to make sense of the scales that people would be viewing doing this at. So some of those, actually it's fine to map an entire catchment in terms of its ability to potentially have groundwater restoration and the potential, if beavers were to create wetlands, how likely is it within that catchment that actually that could get down into the groundwater and recharge aquifers, etc. So it might be fine to map something like that at a catchment scale, but it's not very much good saying within this catchment there is a particular priority habitat that's a little bit too coarse. So we needed to have these data sets at appropriate resolutions on a case by case basis. So it was almost a little bit political which data sets went in there, but also a lot of that sort of technical consideration as well around what was and was not appropriate. 16 00:13:24.000 --> 00:13:53.000 I think you've just blown my mind with the groundwater recharge statement. I never even contemplated that that would be a thing, but of course it would be. Yeah, yeah, I can see how that would work. That's really interesting. So were there any data sets that were discussed? Well, if you can tell us. So were there any data sets that were discussed and considered but weren't used? And what sort of spatial data sets would you like to be using if you had access to anything? 17 00:13:53.000 --> 00:16:36.000 Yeah, so there were a handful of data sets that are sensitive and that we weren't able to put out in the final version of the tool that there are two versions of this tool, I should say, one for the public and one for internal use. And unfortunately we couldn't have that. We would have loved to have those be a mirror to make sure everyone was, you know, playing with the same sort of data. But unfortunately some of the data sets were sensitive things like abstraction, licensing some particular species. Also some of the data around exactly where they are couldn't be shared. So there were a few layers like that. There were some layers which were just data deficient. I mean, things that we flagged at an early stage with the species specialists and the hydrologists that could be really useful. Things like detailed, high quality data sets of riverbank structures. So you mentioned the slope there. It would have been great to have a really good national data set of exactly the sort of shape and structure of riverbanks. Unfortunately, that hike lidar or whatever sort of data we might want to do quite got for that we couldn't quite get our hands on. So that unfortunately has been taken out. And that sort of data set would be hugely useful not only for this project, but I mentioned water volume modeling that we do as well. You know, you can imagine how useful a data set like that would be and could have been, but unfortunately, yes, couldn't put that in there, which the other sort of data set we'd have loved to have generated. But it would have required another stage in the modeling that unfortunately we didn't have the resources for was to. I mentioned the layer that the University of Exeter produced which was around the likelihood of damming. That would have been great to actually use some of the factors, some of the other data that we had around stream order, local topography, soils, etc. To be able and actually generate the likelihood of potential flooding or a flood risk indicator. We stopped short of that just because we found we didn't have the sufficient evidence or the resource on the project to actually put that in there. But we'd have loved to have a spatial layer that actually showed how likely potentially, you know, flooding and how widespread that flooding would have been if beavers were sort of reintroduced to a given segment of a watercourse. But yes, we stopped short of that. So it was an interesting mixture really of things that simply don't exist yet, as far as we're aware, didn't exist at the right spatial scales or just politically or rather, yeah, just were rather too sensitive to put In. So there was a whole host of reasons that we had for not including some data and including others. 18 00:16:36.000 --> 00:17:38.000 Yeah, that whole data set around understanding the sort of aerial impact of damming would be super useful. I mean, just from a layperson's understanding, it's very difficult to get a concept in my head as to how big the water pond behind a beaver dam is. And I know it'll depend on the local circumstances, but sometimes you hear people saying, like, oh, an entire agricultural field will be lost. And then other people say, oh, no, it's all within, pretty much within the margins of the riparian strip. And I'm guessing that in the majority of cases it's somewhere in between. But, yeah, that would be a super useful data set to have. Oh, so you've mentioned some of the data sets that you've been using. What sort of processing were you doing for that type of work? 19 00:17:38.000 --> 00:20:21.000 Yes. So I would say this is where things are probably less groundbreaking. So doing all that work with the specialists and with all of these different agencies to get to the point where we agreed upon data sets was sort of challenging enough. And then to clean that data and to get it into a format that people could interrogate it themselves. The actual processing that we've done on any given layer, particularly because many of these layers that we've got in there are being updated quite regularly, we decided not to put too heavy a processing step in between the raw data and the final sort of output, so to speak, the actual app that people will be interacting with. We did play around with some different ways of doing this, wherein we sort of did some hotspot mapping for, on both sides of the equation, where, you know, you were seeing a lot of good foraging habitat, you were seeing some, you know, some really good potential consequences of having beavers in an area. But we found actually defining what was an opportunity and what was a risk came, became quite tricky. So we had to step back from a lot of the more sort of exploratory stuff there. And we've actually left it a lot up to the user to make some of those decisions and to evidence those on a case by case basis. So primarily what we're doing is we're drawing together a lot of different modeling streams. So some of those have been performed by the University of Exeter, as I mentioned, some by the environment agencies, such as this layer on groundwater recharge, others by Natural England as well, that we've worked on. But the actual modeling for this project was relatively limited. It was more about how do you bring together all of these really rich sources of data into an easily digestible format. So there was a lot about symbology, there was a lot about making sure that the traffic, the actual app that we've put out there was going to work for people and not run incredibly slowly. So almost more on the software engineering side of things, some of the more interesting technical aspects were in terms of how we were going to use caching and segmentation of data to allow that rather than on any specific modeling that went into this. For me, the shining light of the modeling that went into this was probably done in that collaboration between Natural England and the University of Exeter. Looking at the sort of the level of modeling and also the foraging quality of each stretch of watercourse across the country. That was the really interesting part of sort of real new modeling GI analysis that we did. The rest has much more been about. Yeah, as I say, bringing together some really interesting and rich sources of data into a format that people could use. 20 00:20:21.000 --> 00:20:53.000 Yeah, that's. That's pretty cool. So is the plan to develop it further, in particular in light of the fact that I guess that there are more and more licensed releases of beavers and possibly unlicensed, but there will definitely be more beavers around. So I'm guessing quite a few of your data sets will either you'll want to change them or maybe you'll want to update them or the processing around that in order to sort of mirror what's happening on the ground in real life. 21 00:20:53.000 --> 00:23:45.000 Absolutely. So, as I mentioned, we have pretty light touch in how we process data sets after we've extracted them in their raw format. They are put out in most cases as they are without too many pre processing steps in between on the app, which means that we are actually drawing in. So we're using API endpoints etc. To actually get the newest versions of data at least monthly. And there is in the documentation if, if people do fancy a bit of a bit of a read at some point in the documentation there. Each. Each layer is actually sort of. There is a clear explanation about what licensing there is behind a layer, how often it's updated, what the source data is, the sort of limitations and how the data can be used, how it should be interpreted. So yeah, there's a mixture but we are always up. Yeah, we are always updating the data sets that are going into this. There is some more exciting work to actually redo some of the modeling strands. So potentially on that foraging index and the dam capacity that could be updated in the future. But one of the Pieces I did want to mention in this is the fact that we are looking at quite exciting projects with a couple of partners in looking at beaver dispersal and population growth and actually understanding that, okay, we might be able to tell you what will happen or guide you as to what will happen if you put a beaver in a certain waterway. But beavers don't stay put. They do seem to spring up everywhere, whether assisted or unassisted. They do seem to be moving about the place. So it would be quite good to understand what a population might look like 10, 20 years down the line and where they may have dispersed to and therefore try and capture some of that thinking at the earliest possible stage. And also it might give people a bit of steer about what area they might want to actually put into the tool because we allow people to draw a polygon in the tool and it will extract all of the potential receptors or considerations that are within that area. And this might be quite a good way of doing that. If people think, oh, I might release two beavers in this area, okay, where do they spread to? Let's have a look at all of the different things that could be impacted by beavers in that area that, that could be a really interesting tool that we're looking at adding or a new facet to the tool. The pathway to adding a brand new data set into the tool is a bit lengthy and I already talked you through that sort of sign off from loads of different stakeholders. So in terms of adding brand new data sets, we would be open to it, but it's quite a rigorous process that we go through before we add a brand new data set. But as I said, they are being updated, the ones that are currently in there. And yeah, there are some of these new functions that we're looking to develop as well. 22 00:23:45.000 --> 00:24:04.000 Yeah, no, that sounds pretty cool. So have you got examples of where the tool has been used by someone looking for a license? Or is it a case of you've made it available and it's up to the people looking for licenses to get on and use it rather than sort of you being made aware of what's happening. 23 00:24:04.000 --> 00:24:56.000 So before we put the tool out, we actually had a few groups test the tool and all of that sort of feedback was, yeah, pretty, was, was taken on board and we, we actually adopted some of, some of the changes that people suggested. We are monitoring now there is the opportunity for people to tell us if they're using the tool. But we have put it out there. It's sort of no pressure. It is out there now. We've already taken feedback on board from end users so we are keeping our ears to the ground for anything that people might want to see. But yes, I don't think currently we've got quite such a focus on gathering feedback in the same manner that we were before. Yes, it's now out there in the state that it is and we will make changes as and when we. Yes, as and when we hear about new elements that might be. That might be useful. 24 00:24:56.000 --> 00:25:15.000 And you might have alluded to this in some of your previous responses, but are there any areas where you see the tool being more important? So either geographic areas. You've mentioned that it's for the whole of England, but are there specific areas of interest to you or specific scientific areas? 25 00:25:15.000 --> 00:27:07.000 Yeah, so yeah, as exactly. You've hit the nail on the head there. It should be useful nationally. That's the whole point of the tool. So geographically, national usage, really across England anyway. And yeah, so the potential benefits and any risk from beaver activity might vary geographically depending on lots of different factors from topography, land use, river use, you know, all of these things. But the tool specifically allows us to predict what the key beaver behaviors are that people might need to watch for in different areas. So. And how it might interact with those local land management elements. So I would say in terms of examples of where this has been used, we have the example from Kent where water was quite heavily modified and the beavers have actually built fairly few dams. And that wasn't what we expected. We expected in heavily modified watercourses, people to. Sorry, yeah, we expected the beavers to have built quite a few more dams than that. So the tool allows us to. We're sort of going to be developing the tool with that in mind and the science will at least we can sort of track what we expected beavers to do against what they might actually be doing in a location. So I think there are elements where the tool can certainly be informed as the science expands and allows sort of other new branches of science to be brought into the tool later on. Yeah, I, I think there's. Yeah, it's sort of broad, broad applicability. But this is not the be all and this is probably not the final form of the tool. And we are expecting it to both influence the science of reintroduction and be influenced by a growing understanding of what beavers do in different areas of the country. 26 00:27:07.000 --> 00:27:11.000 Yeah, sounds like the beavers need to use the tools so that they understand what they're supposed to be doing. 27 00:27:11.000 --> 00:27:25.000 If only they would. That would make things Fun. I think that would actually, I think that goes without saying. Well, I think that would be very useful in many areas of ecology, wouldn't it? If they could see our models first and could just conform to that quite nicely. That would make my life a hell of a lot easier, that's for sure. 28 00:27:25.000 --> 00:28:00.000 Exactly. So just before we, we come to wrap up, are there any sort of limitations of the toolbox that, that you're sort of aware of or that you think people might sort of get tripped up on? I mean, obviously you've put out a tool that is ready for use, etc. But it's always interesting to get the impression of the people who develop these tools to really know about the ins and outs and whether there's little bits that you think you want to develop a little bit more. Aside from maybe some of the data that we've already touched on. 29 00:28:00.000 --> 00:30:36.000 Yeah, so I think there are various sort of limitations, various assorted limitations here and I think I've touched on many of them already. But yeah, to be a bit more explicit about some of them, I mean each of these constituent layers within the tool, a lot of them are modeled data sets and each one has their own caveats and limitations. And as I say that those are all documented to as greater extent as we can in the accompanying documentation. But yes, that will always be there with any data sets you use, particularly as you get to second and derivative models that have been models upon models. There's always that sort of layer of limitation that's derived from each other. We're also working at a national scale, meaning as I mentioned, we missed out some of those local high quality data sets that do exist. So certainly limitations there where someone, if they were to develop a tool solely for one area of the country, may be able to populate with even more high quality data sets that could be useful. I already also mentioned some of the licensing constraints. So not every data layer is in there that could be useful because unfortunately some of those data sets, just the data licensing didn't work out for those. They were too. Yeah, it just wasn't going to be appropriate or possible to share them. Then the final thing I would, and we've touched on this numerous times actually throughout this discussion, is just how local knowledge is required to interpret the, to interpret the tool. We didn't feel that we had enough, enough data, enough time as well to actually do some of that, you know, that really interesting stuff that we, we've already talked about which, which may have either been that geographic extent of impact or indeed the, the exact impact it would have. As I think we talked about ancient woodland earlier, we did not feel that we were able to say at a national scale whether the beavers coming into an area were going to improve or potential negative implications for any given habitat or feature that was currently there. So, yeah, local knowledge still critical, but it should give broadly. The tool was only ever supposed to give you sort of that starting point to start asking the right questions and start thinking about the right beaver behaviors. 30 00:30:36.000 --> 00:31:02.000 Yeah, that's been brilliant. Thank you so much for your time and for taking us through a new species for my discussions around ecology on this podcast, but also new data sets and new methods for understanding the impact and the opportunities around those species as well. If the listeners want to know more, how can they get in touch with either you or the team or the project? 31 00:31:02.000 --> 00:31:40.000 Yeah, so there are a few, sort of, few entry points for that. I mean, if people want to play around with the tool, I do encourage it, do give it a go. The beaver management website would be the place to go for that, both to play with the tool and to get advice on beaver management. And then if you wanted to contact the folk and natural English who work with beavers, that is beaversnaturalengland.org would be the right email address to send any queries to. But yeah, thanks for having me, Alastair. It's been great talking to you and yeah, keen to potentially have more conversations on different species in the future as well, if it's appropriate. 32 00:31:40.000 --> 00:31:52.000 Excellent. Yeah, I'm always up to talk ecology and data science. My two favorite things. Definitely. Yes. Thank you so much. And with that, I think we'll say cheerio. 33 00:31:52.000 --> 00:31:58.000 Brilliant. Thank you very much. 34 00:31:58.000 --> 00:32:34.000 Thanks for joining us today on the GeoMob podcast. We hope you enjoyed the discussion. You can get the show notes over on the website, which is@thegeomob.com and while you're there, you can sign up for our monthly newsletter and find links to our accounts on Mastodon and LinkedIn. So you can stay up to date on all our upcoming events. We hope to see you at an event soon.