 I will be talking about some of the activities we are pursuing at European level, trying to look at land use at 6000 BP. I've got some results to show you. I'm also going to talk a bit about some of the issues that we've been having to deal with and how you develop a methodology for doing something like this. It's a fairly open presentation. Yn ydych chi'n gwirio o'r ddif chi'n masbwytym Llan 6K, ac yn unig i gyfnodran ni'n gwybod y llanduwymp yma sefydle Llan 6K. A i'n gwahanu ar hyn o'ch gwaith cyfnod. Mae'n gweithio'r hwn e'n gwybod ar Eurpa. So, this is one of the models which has been created by the modelers. This is the work of Peace. Forget his second name now. This is one of the hide maps which have been created for Europe. You can download these, and there are different time slices represented. I'm only going to show you one. This is for 4000 BC, which is the time frame that we're interested in. ac rydych chi'n rhesnod i'ch bod ni'n meddwl y bydd ei ddechrau, i rhan i'r gyfathion i ddod i'r thymianedd o'r arddangosio erbyn yn dod allan iawn, a byddaf ni eisiau beth sy'n mynd i bwyddoi'r cynyddol, ac rydych chi'n gwernio pam unrhyw dechrau i bod eich gweithio i gwaith crindiannol. Ryddwn ni'n gallu digwydd credu i'r hynny, na fyddwn ni'n yn unedill ac yn ystod i'r unig yw dowod. Mae'r cyfgwyrd yn ymddangos gweithio cyfnodd, ac mae'n gweithio i'r cyfnodd y cyfnodd yn ymgyrchau ac ymddangos i'r cyfnodd yn ei gweithio'r llan. Mae'n gweithio'r cyfnodd yn y CK10 yn gweithio'r cyfnodd yn y gweithio'r cyfnodd yn y cwnghwyl yn y cwnghwyl. The hide and KK10 are pre-represented here. You can see that they are similar. But they are clearly not the same. This is the reveal reconstruction from pollen data of land cover. You can see that according to the pollen data, both the models are significantly underestimating openness at 6,200 to 5,700 bp. O'ch meddwl am y cyfnod, os ydych chi fod yn y peth yn gweithio adeiladu cyfnod, yn y peth yn gychwyn i'r hyfforddau, a'r gallwn lle, mae'n rhan, mae'n edrych yn gweld o'r gweithio'r gweithio, mae'n gweithio'r gweithio'r Cyfeunus, ac yn y cyfnod, Cephe 10, yn gwneud o'r cyfeunus, yn cyd-gweithio'r cyfnod, mae'n gweithio'r cyffredinol yn gweithio'r gweithio'r gweithio'r gweithio. Ac mae'n gweithio ei oedd gallwn arweinyddol yn gweithio'n ddefnyddio. Efallai, na nhw'n diolch yn gweithio ar y hollol, y Llandus, o'r rhancofferdadau o ysgol yng nghyrch targedd, yna yn ymwneud yn ymddangos i'r adres, ac mae hynny wedi'ch armellaf y cynnig oedd yn eluio gael y profiad ymddangos. Mae fenyw di'n rhanom yn nodi ar fynd i'ch datblygu hwn i' Caswn. Ac mae'n meddwl pethau o'n gweithio ar yr ymddangos y taith y nifer, felly byddwch yn ymryd i wnaeth i gweithio'r ddiffryd ddiweddol a'r ddiffryd ar y cyfnod at y cyfeirio i ddiffryd i ddiffryd ar gyfer y cwmbrannig. Felly, mae'r ddiffryd wedi gwneud o'r mawr a'r ddiffryd. Felly, drwy'r ffordd yw'r ddiffryd yn ymryd iawn, yng Nghymru ar gyfer y cwmbrannig, Ieithio ydych chi'n gweithio'r bwysig oherwydd rwy'n cyfnodd ac yn unig i'r gweithio'r bobl yn ymgyrch i'r Unedau. Rydych chi'n debyg yn ymgyrch i nhw, rwy'n credu ei rhaid o'r cyfrifau o gweithio a'r L-1 ac i'r cerddau i'r cyfrifau sydd yn gwneud o'r model. Mae'r cerddau i'r cerddau iawn i'n cerddau i'r cyfrifau o'r landfeydd, felly iawn yn cyfrifio'r gweithio. Felly, yn y chers dim ond, sy'n ddigonwyd, y byddwn yn fawr gennym â'r argyfwyddiadau, a'r ddigonwyd yn fawr gennym â'r argyfwyddiadau, y modelau sy'n ddigonwyd yn ei wneud eu ddechrau o'r ddweud cyfnod o'r oedd yn ei ddweud o'r ddata. Mae'r ddweud yn ymddangos i'r ddweud beth o'r ddweud o'r ddweud o'r ddweud o'r informatio both in terms of chronology and in terms of how we classify, how we understand the random uses. We've also revisited some of the categorisation, particularly coming together with us and agriculturalists. I'm not going to go through these because these are very much still at the work in progress, but I think they're not too far off what the main theme is, but my point in drawing this to your attention is that there's still quite a lot of debate to be had about what do we mean. We have a lot of discussion about things like monocops, what do we mean by that? So making sure that we're all speaking the same language is actually quite a big challenge in this project, especially since the definitions are still ongoing, and when we started this work we didn't have all the definitions to hand. So that's led to quite a lot of use or discussion around definitions. Moving on from that, we also had a lot of discussion around how we do this because there's no easy way to do something of this sort of magnitude. I guess most archaeologists are used to dealing with site data, I mean regional data, but not at a continental scale. So the task that we had set ourselves precipitated a whole range of other issues that we have only really just started to grapple with. The first thing is around chronology because obviously we're being asked to map at a particular time frame, and that's 6,000 BP. We were asked to go either side of that by 250 years. So we have used all sites which encompass the 4,000 or 6,000 BP date within that range. So that does mean that there are some sites which fall within that range but which do not encompass the 6,000 BP date which are left out. So it is very much a snapshot, and that's producing some quite interesting results. We've also been thinking a lot about data quality and uncertainty. We have sites which are well dated. We have sites which are much less well dated. We have sites where we have pottery evidence, where we know they're agriculturalists, but no archaeological data. So how do we deal with that level of uncertainty? When we suspect, we know they're probably agriculturalists, but we don't actually have the evidence. How do we record that? How do we make clear those assumptions? So we're having to really think quite carefully about the kind of decisions that we're making. The other issue that came up in our meeting, I don't think it's resolved, is how do we deal with bits of information between data points? So we have sites, and we have lots of space between sites. What do we deal with? How do we deal with that information? Do we assume people are moving in that landscape? Do we put a buffer around it? Do we just use expert knowledge? But all of those things require a decision process. They require transparency. And certainly, modellers are really interested in this because understanding that level of certainty and uncertainty allows them to run different thresholds with their models, different uncertainties, to really establish where the errors might be. And I think we're starting to move in the right direction with the rules of interpolation, but we've still got quite a lot of work to do. Recording of metadata, how do you make sure that there is a clear and transparent mechanism by which all your decisions have been made? And we've been talking quite a bit about this. We've also had a lot of discussion around, in Europe, we have lots and lots of data points. In other parts of the world, there are many less, there may be less data points, but more knowledge around a particular region. I suppose what you might call more arm waving. So how do you deal with those two levels of information? And I think certainly the Europe group were very much of a strong view that we have to work from the data points. And this is a slight disjunct between what the overall project wants us to do and how we feel we need to work as archaeologists. I think we reconcile that now from some of the work that I'll show you in a minute, but there's been a lot of discussion around that. In the end, what we've done is we've used both mapping, using radiocarbon dated sites and expert mapping, but the idea is we keep those layers separate so we are able to distinguish between different levels of knowledge. And we can maintain a level of transparency so that if anybody queries our decision, we can go back, we can revise our decisions. Also, the other issue that came up with the modelers is they want quantified land use. They're not actually just interested in if there are hunter-gatherers here. They want to know how much land use are these people doing. How much are they taking off the land? What percentage of the land? And we're like, I don't know. But the modelers will say, hey, if you tell me the hunter-gatherers here, I will put 100% hunter-gatherers and I will assume, or 100% agriculturalists and I will assume 100% agriculture in this cell. Now, we know as archaeologists that's rubbish. You can't do that. So we have agreed, somewhat reluctantly, to try and get some values on. But this is really important for modelling perspective. And I think there is this process of reconciliation between what the modelers want and what we are able to provide. And we're talking different language, different expectations. So it's been quite an interesting process. But actually having one of the modelers at our meeting was very helpful. So, these are some of the results. And I'm just going to show you a series of images that show the process. So the initial stage that we underwent was we basically used paper maps which had been printed out. They had already been developed within a GIS and they were done by hand. So we divided people up into groups and we just over process of an afternoon because we spent a day and a half talking about the methodology. We started to map out. So we literally just drew on maps. I can see Eric's face. It's just like a picture. So here we have hunter-gatherers. Here we have agriculturalists. And then the rest within this are agriculturalists. So it's interesting because different groups approach the exercise slightly differently. So in this group they just assigned land use category one and then lading everything up and then put the metadata in a sheet. This group was a more sophisticated approach. This group just kind of went, ah, hunter-gatherers all over the British Isles. We don't need to be sophisticated about this. So that's the first part. So we came back, we got it scanned in and this is what we got. Mmm, that's interesting. We also mapped out different levels of data and I'm not going to dwell on this but we started to think, okay, do we map out when we know for sure there are people here and they're dated or do we map, say for instance, where we've got settlement sites but we've got no archivotomy? How do we distinguish between those levels of information? So here is the land use map done at level one category. Now you'll notice that there is apparently nobody in this area in the world. That's not true. It's because the Spanish group haven't finished their work and Feram has now sent me a data sheet but actually for reasons that should become clear it's not a problem. So this is what it looks like. This transparent blob here is something that was sent to me later on giving sort of more of an extensive idea of I think early land use. So it's already starting to show different categories. So the next thing we did was Mark, and this came up at the meeting Mark van de Linden very kindly allowed us to make use of his radiocarbon dates. So these are all radiocarbon dated sites from across Europe. It's a very extensive data set but it's not a complete data set and it's mapped out here and it shows kernel density estimates as well so it's starting to show where you might have some groupings of sites and these are grouped at 250km and you'll see that the map looks different. It also has data points where we didn't have previously and I think having seen this and I'll show you in a second what the two look like we became increasingly more convinced that we have to start from the point data. This is really important. So we have rebelled I'm afraid against the original instructions that we received but I think we can reconcile them. This is the same information but starting to filter it out between chart from short-lived samples so most of this data set is all geo-referenced. It has quite a lot of associated environmental data with it but it is going to have to be populated in terms of other more detailed information for land use and we also need to start to populate for other parts particularly here other parts of Europe for which we don't have any data and we just agreed that we will try and do that over the next couple of months. When you then superimpose the block map maps I think what's actually quite interesting is how accurate we were. If you look at the British Isle as well it's not bad actually. Okay, up in Scotland perhaps we need to be a bit more sophisticated but over here it's not bad. So this is a first pass and I think our conclusion from all of this is that we have to use the radiocarbon data but also there are lots of parts of Europe and elsewhere for which there are many less well dated sites for which we know there are agriculturalists or hunter-gatherers during the timeframe and we want to make sure that we can include that level of information and essentially show it at different levels of resolution so the two will be put together. The last thing I want to show you is this so this is just revisiting the hide map at 4000 so according to this map I think we should primarily be in semi-natural woodlands remote. Now as you can see there are lots and lots of sites lots of things happening I doubt very much that really what we're seeing is remote semi-natural woodlands so the overlap between the data that we have in terms of radiocarbon dates and the model data clearly there's a disjunct between it but also in some areas I know, I think Mark was talking about some of these areas people are definitely cultivating they're having quite a big impact on the landscape so this tells us very clearly that there are big issues with the hide map we kind of knew that but I'm not going to see that. So this is the database which we will eventually hope to make use of so the idea will be that we will aggregate this information into the grid 8x8km grid squares but I think once you've made a decision about a land use class in a grid you can't really go back what you've made a particular decision to decide to take the approach that we have which is essentially to move up from the data point we will create our own data set of the European data and then it will be aggregated and already we have the grid behind the map that we've got anyway so the European map can be moved up relatively easily into the grid system that's all I'm going to say I'll leave it there