 I'm Steve Hallott, from Cranford University. Professor of Applied Environmental Informatics, along with Ron, one of the two digital environment champions for this project. It's absolutely wonderful to see you all here today. Thank you so much for coming. We now move to a part of the programme, which is to hear the voice of experts and stakeholders from the various partner organisations and other organisations who use digital approaches in their work. We have a panel discussion now, and a chance for you to put some questions to colleagues here, and then we have some presentations after lunch. We're just going to push lunch back a little bit, looking at the time of things. But I'm absolutely delighted to invite and to welcome here a number of colleagues. First of all, we have Sally Brown, the Principal Scientist Flood and Coastal Risk Management Research from Environment Agency. Thank you very much. We have Mick Wheeler. Now, of course, this morning we heard from Isabel Stephen about the SPF programmes. Delighted to welcome Mick, from the SPF Landscape Decisions programme, which is in fact coincidentally running at a conference here in this very room for the next couple of days. Kate Gill from the Q Science. Thank you, Gill, for coming from Digital Revolution and the priority lead of that. Lastly, Julie Gregory, Senior Program Manager from the Climate Change Task Force Network Royal. What I'd like to do is just invite each of the colleagues here to give just a few comments and introduce themselves a little bit about their work, and then we'll have a discussion. So, Kate, may I start with you, and we'll move down there. Hopefully you can hear me. Sounds good. So, Dr Kate Gill, I've come from the Royal Botanic Runs Q, but I am not a biology student remotely. I'm an aeronautical engineer, so basically I'm a data scientist who's done software, and that's kind of where my kind of head comes into this conversation. People ask me why I work at Q. This is my fifth government department. In terms of my conversations, I always come back to like a quote that I remember from a colleague of mine, actually, who's a Wilson, an American biologist, who basically said, we're drowning information while starving for wisdom, and that's kind of in my life for the last 40 years in various departments around government. I've landed at Q because I'm leading the Digital Revolution Project, which is basically digitising and capturing anything and everything that moves or doesn't move in Q, including the Millennium Seed back in Wakehurst, and that's really what I'm trying to do is kind of mobilising the data, mobilising the information and then helping the Q staff and other researchers worldwide to access that and kind of be the really exciting bit, which I'm kind of looking forward to in the next step, which is why we've got all this data. Now, what can we learn from it? And that's the bit I'm getting very excited about now. Thank you very much. So, yeah, I'm Julie Gregory. I work for Network Rail as Senior Program Manager, as Steve said, for Climate Change Task Force. So, this is a role that was created about eight months ago, but I actually got involved with this work with one of the projects in my previous role, where I was a sponsor for the South West Rail Resilience Programme, bit of a mouthful, but that basically was looking at the piece of railway down in Devon, between Dawlish and Tymeth, where you have a sea on one side and steep cliffs on the other side, twin massive environmental threats to the railway, and it's obviously a key link connecting the South West Peninsula with the rest of the network. And so, I became engaged in the Cream Tea Project as part of that, which looks at the overtopping at Dawlish, and we're looking to use that information to help us respond in real time to high tides and overtopping, but also to think about what we might need to do longer term to kind of make a resilient network. So, yeah, it's been fascinating to come along and just hear, you know, the breadth of the programme and lots and lots of insights, particularly interested in the, was it, the Alessandro's Talk, because we've got loads of cliffs and loads of earthworks that are causing a threat to the railway. So, yeah, loads of really interesting stuff to think about, and, yeah, echoing your point about information is what we need, not data. Thank you. The Sansom project there was, yes, thank you. Sansom. Please. Hello, everybody. My name's Sally. I work at the Environment Agency, where I'm in one of the national teams that look after flood risk, and I'm particularly interested in research. Research needs related to flood risk, sucking information in and getting it to the right people in the Environment Agency. I've been in my post for a year, and so I'm still exploring what the Environment Agency does. One of the things I'm trying to do is to start a project on digital technologies. How do we get the right digital technologies into the Environment Agency? So it's been really good to listen today to hear about some of the cross-cutting things that are going on, and particularly the processes, skills and techniques as well. So just some thoughts I've had so far listening to today, and also talking about 50 different colleagues from around the Environment Agency as well for some of those challenges. First of all is that data. There's a lot of data out there, we've just said again, but trying to get it in the right space is really important. It's also about the data quality, integration of that data, the storing, formatting. There's a lot there, a lot of historical data as well. It's not just what you're picking up now, but how do we actually get that into a friendly and usable way, particularly if you want to use it for different applications like AI, machine learning, how do we actually get that in a comprehensible way? The second thing is how do we use that data to make a decision? So is that actually lots of data really helpful, or do we have enough already? And that again is a question of affordability, is a question of the skillsets, is a question of the visualisation of how we can make decisions quickly and efficiently and effectively, and about the longevity of any model in any data set as well. It's okay running something for the duration of a project, but what happens next? Who's responsibility is it to keep that data alive and to keep that project alive? And then the final point is about the use of ethics, the appropriate use training as well for staff that may need upskilling to make sure that we continue to improve what we do. So that's another legacy, what I'd really love to hear today as well from people with different projects, how we're going to continue that to make sure the data gets to the right users at the right timing so we have the right information to make decisions. Thanks very much, and linking up the SPF projects. Mick? Thanks Steve, so thanks very much for inviting me here. Just by good fortune we, as Steve said, happened to be running our final programme meeting right here over the next couple of days. If by way of a plug, if anybody is interested in coming to that meeting is a bit late in the day, but there might be an opportunity for signing up, so please see me afterwards if you're interested in coming along. By way of background, I'm basically the University of Leicester, I'm a professor of environmental science, I have a background in hydrology and biogeochemical cycles originally, but over the last 20 years I've got into synthetic organic compound, fat and behaviour, so my own research is water quality. The reason why I'm sitting here this afternoon is because I also sit on the programme coordination team for the SPF landscape decisions programme. So the programme, as you've probably picked up, is there's a huge overlap, I mean one of the things that struck me this morning from hearing the presentations was the massive amount of overlap between what's going on in landscape decisions and what's going on in this programme. The strap line for the programme is towards a new framework for using land assets in the UK. There are 63 interconnected projects spanning a very wide range of disciplines, most of them are interdisciplinary, lots of work on developing new techniques, new mathematics and models and applying models for looking at landscapes, but also we have a work package which is called New Thinking and Communities, which brings in people into the equation. So thinking about landscapes and landscape decisions, it's really important to remember the people and not get lost in the data and the modelling. A few reflections on what I've heard this morning and maybe overlaps. The first thing that I think Ron picked up this morning was the ethics and legal thing and that's really quite important and I think the last speaker talked about drawing data from software that's used by farmers and that's a very touchy sensitive issue that farmers aren't always that pleased to give away their data. There's a lot of commercial sensitivities and legal sensitivities about that, so when we're talking about bringing data or taking data and using that to make decisions, we do have to be very mindful about the legal and ethical considerations. Lots of things going on. One of the drivers, I guess, for landscape decisions was the DEFRA 25 year environment plan, which basically is built around this idea of multifunctional landscapes. So a key concept for landscape decisions is that the landscape is multifunctional. It provides lots of different services. Traditionally, it was food and fibre production, but over the last 20 years we've started to become increasingly aware of the other services that are provided by landscapes. In the agricultural landscape there may also be disbenefits, so externalities that are farming, for example, may cause losses of biodiversity, water quality problems and problems with water quantity. A lot of what is going on in landscape decisions is thinking about multifunctional landscapes, trying to quantify ecosystem services. Some of those are easy to quantify. Some of them are less easy to quantify. For example, biodiversity itself, as a thing to measure, isn't always that easy to measure. We've also got other things like cultural services that are very important, how people value landscapes, that are very difficult to incorporate into models. So there's a big challenge there. Some projects in landscape decisions have been trying to grapple with that challenge. I'm going to show up in a minute. Obviously, this leads very strongly into policy. Most of you will wear more than me that DEFRA are trying to develop a new set of agri-environment schemes around this concept of multifunctional landscapes, public money for public good etc. Very central to that will be collecting the evidence for the benefits that are generated by different landscape uses and interventions. It's really important that we get it right. Data is going to be really important for that. Models are going to be really important, sensing everything we've been talking about today. The last thing I'm going to mention I picked up from several talks this morning is the value of visualization and interaction and producing software that's actually engaging that pulls people in. It's great to do all this fancy modeling and data analysis, but if you don't have a user that's actually going to be interested and going to be drawn in, then it's not going to be of much use. So I think visualization and software developers have a really important role to play in all of this. So I'll leave it there and we'll open it up for one of the speakers. Yeah, thanks Mike. So it's so true visualizations are a key part of this and that's a very nice example. I'm also interested, you mentioned the ethics and so on as picking that up and I just drew my thoughts to the, we did a webinar series as another plug for the YouTube channel by the way. We did a webinar series on the ethics in AI and the intellectual property and bias in AI and there was some really insightful talks there. It's a fascinating area. So what I'd like to do now is just throw out a couple of themes really to colleagues here and I'm hoping we'll have some questions from all of you as well so you can be thinking of your questions, but maybe we could just start by just putting to all of you really what you think the advantages, the strengths, the opportunities that digital approaches have in the way that organizations like yourselves conduct business and what is the art of the possible in a sense for some of the things we're hearing about. Who'd like to have a crack at that? Julie perhaps? Yes please do. So we clearly do need data to drive our decisions. Earlier I said we don't need data, we need information. We need information to drive our decisions for that information comes from data doesn't it? So in the real way we have to make operational, we have to make safety and performance decisions every day and very often they're not driven particularly by particularly robust or comprehensive data. So particularly in the kind of the weather resilience space and climate adaptation we need to make sure that we've got those kind of real data-led kind of decision processes. We're constantly balancing two things really in the railway so Network Rail for those that don't know manages all the infrastructure of the National Rail Network and we're constantly balancing safety and performance. So obviously we want to keep passengers safe but if we wanted to keep them 100% safe 100% of the time we wouldn't ever run a train so we have to make decisions about performance as well so we need to basically route these decisions in as best data as we possibly can so we can balance those two things and run trains safely but also run them as reliably as possible as well. So yeah it's really what decisions are, is the data going to drive and that's absolutely key I think. Thank you. Sally and the Environment Agency how are things done? I was going to say health and safety is one of them if you're out on site or where your limits are particularly in extreme weather conditions really really important. It's also about efficiency and how we work so if we've got a lot of data do we have the right skills in again making sure you're right the skills in the organisation are correct so if AI can do something you know can we use up skill people in one area to make sure we can use their time effectively and also can we avoid doing boring tasks I think that's the other thing of actually how can we actually be efficient in what we're doing and use different technologies in different ways to do the things that maybe we don't really want to do. Yeah that's right. Kate just thinking we had a discussion before and I was fascinated to hear about the span and breadth of all the work that you and your team do at QScience. I mean what's missing do you think at the moment technology skills onboarding I mean is a familiarity of these technologies with management what's missing do you think? I think thanks for the question. I think in terms of the the conversation we were having because again I've had different environments and different conversations it's that cultural confidence about digital conversations is in every job I've been in it's case of I didn't know how to ask the right question every time or I didn't know I could do that or I didn't actually kind of think oh it's got to be a specialist person who does that and I think collectively and this community as much as every other digital community I've spoken to it's just like we will have to have a little bit more confidence in asking the question because sometimes you kind of go I don't know what I don't know but actually I just need to throw myself into it so in terms of what's missing it's that confidence conversation sometimes and also that kind of acceptance of the fact you're going to fail. My last job I failed spectacularly in so many different ways but I also succeeded and that's the kind of like that conversation is sometimes we're a bit risk adverse in terms of that data conversation and sometimes you should throw yourself into it and you will learn something from the journey not necessarily what you thought you were going to do so in answer to your previous question about what are we missing that predictive AI I always find really interesting conversation because everyone says well that's great AI is going to solve world peace it's not because it looks backwards but in terms of that predictive AI is we're going to find some stuff that we never thought we didn't know and we just need to kind of cope with that and have that conversation and then understand why we didn't think we didn't know that etc that known unknown kind of conversation piece and just have a little bit more comfort with that I think that's one of the real things that I think is missing. I mean I think the make we talked about the convergence really of some of the things that you're some of the themes you're looking at and and and what you've heard here today I'm just you know what where are we going with this what's what's the future for for these sorts of digital approaches and looking at things like land landscape decisions where do where do where do the these technologies fit in what is it going well I think maybe just go back a little bit I just just to reflect a little bit on how far we've come you know thinking back to when I first started in in environmental science you know how hard it was to get data how little data there was out there now we are it is absolutely amazing I mean when you when you reflect on the technology that's in our pockets that's in our homes the the computational power on our desktops compared to the computational power that we had only a few years ago I distinctly remember when we got a machine that had a gigabyte of not the memory of a gigabyte hard disk and we were you know what's ever going to use a gigabyte how can you and and of course how things have come along so I think we are now in a position with the technology of sensing technology particularly remote sensing it is unbelievable that that that this space based technology is now at everyone's fingertips and many of these data sets are you know open source and are available to everyone so I think that is absolutely incredible I think going forward I think there's so much opportunity to use these data to gain insight so it's all about getting the insight you know we've got a lot of data a lot of numbers flying around we're drowning in information but we're we're we're scarce of knowledge I think that's very very poignant for this conversation and a key thing we want to try and get insight we want to learn something from the data and key to that is to be mindful that the data do need to be quality controlled and there needs to be a you know some work done on on making sure that data are consistent I think Matt mentioned that the data are being collected by a whole range of different organisations made perhaps using different methods and aren't always comparable so that's one thing that I think we need to be mindful of it's very easy just to throw numbers into an AI and we get nonsense nonsense in nonsense out etc but I think that the the potential is tremendous for gaining insight from all of this data I'd like to see and it's happening more in situ sensors for example for greenhouse gas emissions but also for water quality part of the sort of an outcome from the recent outrage about water companies spilling raw sewage into into our rivers is that combined sewer overflows will be much more heavily monitored in future and that will degenerate a whole set of very very useful data so I'm looking forward to seeing more more sensors being used also to seeing the data being provided in you know the right format in an open access format I can't remember the the phrase I quite like at the beginning somebody mentioned I wrote it down analysis ready as one of these phrases that was that was it analysis ready I quite like that so analysis ready data sets with good metadata so that they're you know they're easy to use and quick quick to use so that that that that's I think within our grasp I think we we're definitely able to do that now and I'd like to see a little bit more of that so stop there yeah the print the principles of sort of fair data sets that's findable accessible interoperable and reusable this is something that's really emerged hasn't it in the last few years and the emergence of curated described analysis ready data sets vast bodies of data through centres like cedar and the jasmine super computer and Daphne colleagues in the audience from both of those facilities is a you know it's extraordinary development. I also actually should just note that this is capturing that insight that you meant from the data is a key factor in the NERC digital strategy which is something we should all have a look at. I think there's an opportunity now to ask you to pose some questions to our panel members so who'd like to start proceedings with a question for our colleagues here should put your hand up and if you'd like to say who you are and thank you. Hello my name is Rachel Chakros I come from the marina fisheries directorate within DEFRA as part of the digital strategy team. I'm interested in understanding how you determine your baseline on which to build your evidence platform. If you start from today we've already had a lot of degeneration and so therefore if you're going onwards from there then how do you determine how well you're improving a system? If you start from a long time ago you may not have the data sufficient to give you that quality baseline. Thank you. Thanks very much. Baselining, who'd like to have a crack at that? Can I start? I think that's a really great question because if you're going to track progress over time you need to know where you started from and sometimes we do have good data already but not always so I can't really answer the question but I think it's a good point that's well made that that quite often we don't spend enough time thinking about how we're actually going to measure progress over time. One of my big frustrations in the in the sort of water quality world is an obsession with interventions without proper measurement of the effect of the intervention so the intervention itself is the good not the actual outcome and it's the outcome that we really want to achieve and we need to monitor the outcome we need to be able to measure the outcome and track that over time so it's a good point. Thank you. I think we've got time for a couple of quick questions before we break for lunch. Please at the back. Thank you. Did I see the second one over there, John? Yes, thank you. Please. Hi, I'm Elizabeth Cowdery. I'm a postdoctoral researcher at the James Hutton Institute. I do data synthesis with models and I'm responding to the comments about visualization and also earlier thoughts on essentially how does digital develop with the natural research and I was thinking about how often we start with thinking with working with data sciences we think about how we have to tell our story we make our figures we have to work with people who know how to interact with policy but these days there are so many more fields things like user interface UI user experience UX studies web design app design if I were to design a new project I could have a whole new area of team members who do all of these different things can you imagine projects moving forward that would include this entire new area of all these new people just focusing on these areas to help move digital the digital aspect of the research forward and help us support this side so that the data science can focus on what we do best even with the same say monetary resources that we have. Okay, thank you. You have programmes of capturing vast amounts of data from all of the money samples and so on at Q and visualization of this to support the science must be key. I mean how are you approaching visualization of all these? That's a fantastic question and I was desperate to answer it so thanks for giving me a question. Q and white curse we clearly have a lot of customer interaction and that kind of inclusion conversation is really really forefront how do we tell the story how do we tell it the story of the data in a way that lands with those communities both in the UK and worldwide and then also the visual kind of web based interface conversation as well in multiple languages in multiple ways and how do we actually get that conversation because we've got some very very informed citizen science communities now a lot more I would say than we had a few years ago and I think if COVID bought us one thing it's bought us that community that's said at home and actually engage we have a huge amount of those that engage with Q so we're being challenged constantly all the time as to is it a video is it AI is it an augmented reality is it we're currently doing that and trying lots of different things in Q now we've just employed some some augmented reality specialists for exactly that reason because we need to tell the stories in the way that lands and actually that's a really important part about the data so so well we're doing we're doing the project and I'm very fortunate that I'm actually in a in the middle of a four-year project that's funded at the minute and my biggest concern is what happens in two or three years time when that drops off the cliff so that I don't think I've solved everything on this conversation it's more of a case of totally recognising what you've said about how do we actually employ the specialist to tell the stories and there was a digital comms person that I employed about six or seven months ago who's now done some amazing things that I would never even have thought about thinking about let alone actually put into place and it's only because they came in with a completely different set of eyes and said well why are you not doing this again for me it's just like yeah I didn't even know I could ask that question so I'm challenging that environment as well and I think the more different skills you can bring into a conversation that the more kind of opportunities you have for that data exploitation so thanks for a fantastic question is the public interface to the data science isn't it it's the and if you're doing it public engagement and it's absolutely crucial so I think we have one last question before you break please hello I'm Ryan Chris I'm a PhD student at Cranfield I had a question about kind of using this data rather than just having this huge this huge data set is actually using it you know requires a lot of computational power and one of the things that kind of I've been reading about is that the UK kind of compared to places like the US lacks this kind of capacity for compute the kind of the hardcore data crunching how much an issue is this for your just kind of respective areas and how do you think it can be solved in the future so access to compute sorry just clarify yes access to kind of supercomputers and the kind of things that like particularly like AI based analysis the kind of high level it's okay thank you thank you very much do the how did the the environment agency and network where it will deal with supercomputers to you I mean a network rail with the new measurement train are collecting terabytes of data absolutely extraordinary amounts of data and I was lucky enough to go on the train and saw this huge amounts of data being pulled off and actually process before the train had even stopped and all the processing actually going on on the train but where does that data go and how's that as a supercomputers use for yeah I mean I used to work for the Met Office and we had supercomputers but I have no idea if network rail has supercomputers I would think probably not and I think that's it's really good to hear you know that at the cutting edge of data research and data science is is all of this thinking about what do we do with these huge amounts of data that we're now being able to download from various places and and the next steps you know of how we make them usable I can't pretend to know what network rail doesn't afraid but I just think it's um it's really key to kind of talk to the the ultimate users and you know trying to get that post processing into the most usable form possible yeah I think I think the the data opportunities are fairly significant and you know I was also lucky to go to the the data center and down in Bath the environment agency I mean the environment agency are also collecting and dealing with huge data data resources are supercomputers used in in do you know I mean I don't know the details there and I think I think modeling some of the the flows in the North Sea I've spoken with colleagues on the Ipswich offers and I think there are circulation models that do use supercomputers but it's you know that it's interesting to know how we how we actually give access to some of these these platforms really I don't know make if you have any thoughts on I'd like to just put the question out to the audience do people think that that computing resources in the UK are lagging behind the US and you know is there a need for additional resource in this area can we have a show of hands to people feel like it's a constraint I guess it depends on where you're coming from but there's a few people I mean there must be a few people think well you know I only need a you know a desktop laptop computer I can run everything on that but um yeah I don't know I guess it's I guess it's sort of context specific I know the Met Office have allegedly very high performance computing facilities probably other organisations that also have that but maybe generally across the board that's not the case we have a colleague from from Jasmine and from Daphne here just please a quick thought from you and then we'll break hi I'm Matt Pritchard from the Jasmine team at Rall and yeah I suppose I can I can talk a little bit um from the point of view of Jasmine but also the CEDA the Center for Environmental Data Analysis which is sort of run from the same same team so one of the interesting things there is that um you know whereas we started with a service that allowed people to download data from what was the British Atmospheric Data Center and the NERC Earth Observation Data Center that grew into the the CEDA archive covering not just those disciplines but also sort of climate model data and and it's been really interesting to see how with the advent of Jasmine it's not just about providing the data it's also providing um a platform a sort of collaborative space where people can come together and do do their stuff collaboratively and that's a really important part of of providing some of those big facilities that it's not just the data and the compute power and that sort of thing it's actually the the interactions that happen between groups between scientists between disciplines even and and facilitating that to happen. Thanks very much Matt and um with Daphne I think the Daphne user conference is just going that might just be time for people to sneak in uh to let's let's go um yeah okay just have a quick quick one on them. Yeah well it's just to follow on from this about user needs for HPC and things I from my point of view I feel like we're quite well served for heavy number crunching you know we have lots of big supercomputers in the UK that do a great job for that from from my point of view and I don't know if others feel the same I feel like there's a bit of a niche in the middle between desktop or local servers and the very heavy stuff something in the middle to you know where you can play around with data analyze data sets et cetera things like the sinodes on jasmine or the clouds offerings I feel like there could be more done in that space I think that's that's a bit of a gap that we have personally. Thanks Matthew. Right colleagues I think uh much as we like the discussion to go on here there's plenty of discussion over lunch uh we'll break now and um you'll notice the exhibitions are all there please take the opportunity to talk to the teams who've kind of