 Bwrw, roi'n gweithio gyda'r Yntroyd, rwy'n Paul Brown i'r Gwyl Shallos Gwyl Rhifordd, sy'n gwyshion rwyfgrifyr i gwbl iechyd meddwlau Lŷch. Mae'r Gwyl Rhifordd ar y statfodaeth roi'r Gwyl Rhifordd. Felly, rwy'n gobl eich drafod i wneudio'r Gwyl Rhifordd ar gyfer y Cwyl Rhifordd. O'r rhesymlu mae'r cymaelfaeth sydd yn gweithio'r gymryd pan daniaeth bwysig,cymiwn i g такa'r gweithio. Newcastle University bringing the remote sensing, GIS, data science and social sciences to the project. University of Strath Clyde bringing the data science and the modelling. UK CEH, Michael bringing his ecological and social science and citizen science expertise. Work with charities, specifically the tree council and also working with the tree council warden network in Norfolk who have been providing the tree condition assessments and the internet of things network management as well. And also Norfolk County Council providing A, the land for the sites that we used and also stakeholder engagement workshops and tree inspections done by tree health inspector. So the challenge, why monitor tree health? So trees are under increasing stress from pests, diseases and climate change. This project was looking at multiple threats. We weren't concerned with one particular issue. We're looking at trying to make an early warning system for the spread of current threats or the incursion of a new threat to our treescape. The loss of trees is implications for conservation, biodiversity, ecosystem services and public safety. And we can see on the images there, images of tree health problems, powdery mildew, lesions on the tree trunk and then also the results of those problems felling of that tree scape next to roadway because it becomes a safety, a public safety concern. The timeline here is just a little snapshot just showing you the amount of diseases threats that are threatening our treescape over the last, sort of, since 1997. Most notably, you will have all heard of in 2012, the incursion of ash dieback, which has been really affecting our ash trees population here in the UK. So that's the challenge that we wanted to look at. So what were the aims of the project? We were looking to develop and demonstrate a monitoring system to detect stress from multiple causes able to be deployed at a landscape regional scale. We want to scale this demonstrator up moving forward. Combine citizen science in the tree warden network with technology and modelling methods. And we were looking to design and demonstrate a blueprint for future deployment throughout the UK. For this little schematic, this little diagram is a real nice catch all of the other Sentinel treescape. You start with the monitoring system. It's been deployed since May 2021 in Norfolk just east of Norwich in the UK. We've got woodland sites, roadside sites and farmland sites. We're observing the stress recorded on 150 ash oak and sweet chestnut trees, including 60 that have been monitored by the Internet of Things, the tree talker sensors. So these, on the bottom left hand corner, are these wireless tree talker sensor network. It measures the physiology, the canopy condition, the growth and stability alongside environmental variables like temperature relative humidity. Combined with that is the citizen science observations. So the tree council tree warden volunteers have been undertaking visual surveys using an app that we designed to monitor the tree health of the trees with the sensors on and the further 50 trees in the three woodland environments. We've also been capturing drone data and satellite data to look at the upscaling aspect, whether we can see what we're seeing on the ground observations, whether we're seeing that spectrally in the canopy spectral response. And then it's just how we integrate that data and model and scale up from this demonstrator project. And here's a few images, a tree talker sensor, and then the tree wardens out inspecting the trees, did a huge amount of engagement with the tree wardens, worked with the tree council doing spring in the woods, summer in the woods, autumn in the woods, and would go round on a tree walk looking at species identification, then also discuss the network, discuss how the observations are going, what improvements can be made. And one key aspect that came out of this engagement and evolved during the project is we kind of started with these Internet of Things sensors and these tree warden observations. And to my fault, I kind of looked at them as separate entities, but they very much started to merge. And what we found is the technology, the IT technology was driving the engagement of the tree wardens. And he wanted to get more involved in that network. And at the start of the project, once we got the network out in the field, we realised that they wanted to help maintain that network. So they were changing the batteries, monitoring the battery life, making small repairs when a rabbit ate through a wire, something like that. So we actually developed a dashboard, an online dashboard, where they could monitor in real time the battery life of each individual sensor so they could literally go and change the batteries that needed changing rather than having to haul 20 very heavy batteries back and forth in one time. So that was a really brilliant part of the project, seeing that engagement and how they wanted to get more involved. So I'm just going to touch very briefly on some sort of results because we've got the poster out there. We can discuss it more extensively during the breaks. But these are the tree health status from the tree warden observations. So this is just looking at average health status across three sites, correlating all their observations over an 18-month period to really get that general picture of the health of those individual trees. And then what we can do is we can look at the spectral response in those certain canopies of those trees and see if we can see those stresses in the remote sensing data. And then another one is just species classification. So looking at machine learning algorithms using e-cognition to classify the treescape from the remotely sensed data. I'll finish down the slides. I think the main challenges with this project is the sensors were very much at a research stage in the sensor. When we started the project and the company that developed the sensors has actually sort of exploded and they've got too much demand for the supply. So having the support from them was a little bit lacking. And the hub failures, the 20 sensors go to a hub and then that hub comes back to the database at fara in real time. But we had hub failures quite often so we had to go down a five-hour journey just to swap out a hub and then you'd miss data, you'd have like a couple of weeks whether there wasn't data coming in. Luckily the tree talker sensors actually hold the data internally themselves but it means walking around each individual one, taking apart serial input, downloading the data. So that was a little bit of a challenge. A positive, we're really looking to scale this up and see how we can understand how we can distribute these sensinal treescapes in a region and then extrapolate between with modelling. So we've actually got a PhD starting at the end of this month who's been awarded funding through Farra and Newcastle University's joint institutes IAFRI to continue the project for another three years. So that's great. Thank you very much.