You are on page 1of 20

DART heritage remote sensing horizon scanning workshop

Tuesday, September 17, 2013


Tony Cohn Welcome Introduction to the day. About 4 years ago the UK research councils were looking to fund projects on science and cultural heritage. This was the only one looking at remote sensing. Now we have the data and have to think about how to use it.

Ant Beck DART prcis Ant gave an overview of the DART project which consists of a large consortium and has 4 PhD students. There is a range of people involved. Ant thanked those from industry who have provided equipment and support to the project. The DART project has looked at contrast identification, looking at processes which lead to the features we want to detect. Different methods have been used, for example aerial, looking at crop height and colour, topography and thermal effects. There were many work packages, most have been achieved, likewise for milestones. Communication methods have included the website and all data is in the public environment, including videos, presentations etc. Summary of the sites used. Each region had to have sites with heavy and free draining soils in close proximity. Generally features are more difficult to detect in clay. Diddington Harnhill Initial surveys were used to characterise features. The general framework was to excavate then insert sensors. Geophysics was done on an area close by. Data was collected multiple times. Probes, such as TDR, to measure soil moisture and temperature were placed at several depths down the archaeological feature to see if the feature acts as a sponge changing the measurements. The clay soils had more sensors placed.

Samples were sent for lab analysis. The samples were taken directly over the geophysics area with the assumption that soils would be similar close by, but that assumption was found to be not true. At Harnhill thermal experiments were done. The originally proposed excavation site had to be changed as the villa area was scheduled a week before excavation was planned to start, so it was moved to a field boundary. Achievements of DART include, conferences, presentations and an impact in the media. Ants mind maps used for this presentation are available online.

Research summaries by 3 PhD students

Dan Boddice Civil Engineering University of Birmingham Dans work is on electromagnetic (EM) radiation and is still on-going, so talk about how research is progressing. EM radiation is used in many techniques. Dan gave an introduction to the physics of the reflection of an EM signal. EM signal responds to magnetic properties, electrical conductivity and dielectric permittivity of the material. Consider the magnetic response not to change too much over time. EM response changes with the seasons, measured by sending a pulse and taking measurements of time delay and how much energy is lost. These measurements were taken every hour. The dielectric permittivity is related to the moisture level of the soil. Factors affecting the conductivity and permittivity are rain, climate and interactions in soils. Contrasts between features and surrounding soils are affected by the maximum storage of the soil and by infiltration, for example a fast rise in moisture after rainfall. TDR allowed a range of frequencies of signal to be used. Different frequencies gave different findings. Temperature is considered important as it has seasonal and diurnal changes, affects the thickness of the water layer and affects ion movement. The extent of importance of temperature is debated. Dan showed plots of results for conductivity and permittivity for Diddington clay. There were periods of dryness. When the TDR probes were installed the soil dried out and there were not any results until after wetting.

A plot of permittivity against conductivity for different temperatures showed a high relationship between temperature and conductivity especially when wet. Temperatures were taken every half hour. For the Diddington well drained site, the plots show strange behaviour towards the end which Dan is still investigating, but is possibly due to the water table. At the Diddington pasture, the probes again took a long time to settle. The relationship with temperature was very similar to at the clay site. There was little difference in pattern between the archaeology and the surrounding soil. Dan is looking at different types of infiltration, for example in fine or coarse grained soils, and looking at the lag after rain. Dan used autocorrelation and found infiltration slightly faster within archaeology than surrounding soils. Dan compared measurements from TDRs to IMKO sensors for moisture content. For permittivity, he found different absolute values but similar trends. For conductivity, the absolute values were vastly different but some similar trends. IMKO sensors have some advantages and disadvantages. For example, they are easier to install but may need calibration. Dan had some problems with loss of sensors and still needs to look at the Harnhill sites.

Robert Fry University of Bradford Roberts work is mostly geophysics, above ground measurements, mainly using resistivity. It is known that there are seasonal problems with geophysics in that sometimes you see a feature and sometimes you dont. This problem is often only examined on well drained soils. DART gave the chance to examine this by comparing different soils. The English Heritage guidelines state that geophysical surveys should be done when moisture contrasts are at their most accentuated. Resistance measurements were taken using a multiplex system with an increased spacing of probes. An overview showed that seasonality was not found at the DART sites, responses were completely different in the same month of different years. Data plots were shown for each month of survey at each of the four sites. On the free draining soils, a clear ditch anomaly could be seen in every month. For the Harnhill clay site, some months did not come out very well. For the Diddington clay site, eventually a negative anomaly was seen, in one month only.

The results were analysed as for a commercial study by looking for high and low readings. For the free draining soils every month showed, but prompted the question how does the magnitude change? Two methods were used to calculate the contrast between the feature and surroundings. A detection test used the difference between ditch and surround, essentially using a correlation coefficient. The magnitude test looked at the percentage difference. Results were shown as bar graphs. Even on free draining soils, the magnitude changes. For clay soils, the best results at Harnhill were when it was dry; but at Diddington the best results were when it was wettest. Looked at the relationship between the weather and the measurements taken. Harnhill was wettest at the end of the study. Diddington was generally much drier than Harnhill. Plotting the responses with the moisture levels. For Harnhill free draining site, when excluding the driest three months, there was a strong (negative) correlation between the moisture level and the responses recorded, that is there was a better response when drier. A similar pattern showed at the Harnhill clay site. For Diddington, the opposite trend was found, that is the contrasts were bigger when wetter. This is still being investigated. In summary, no seasonality was seen. It is not clear that wet is best. To look further at when moisture contrasts are accentuated, electrical imaging was done with a line of probes. This data was used with the excavation data. At the Harnhill free draining site, the best contrast was at the bottom of the ditch. This site could be considered a text book ditch and showed increased contrast in dry months. At the Harnhill clay site, the greatest contrast was in the shallowest area, around rather than above the ditch. This is possibly an impact of the drainage pipe in the ditch which acts like a sump and draws water down. In summary, each area is very different and the research generates more questions than answers. There is potential to look at modelling afterwards.

David Stott University of Leeds David has been looking at the airborne part of the project, vegetation marks on arable land. He is looking at contrast and how it changes over time with different weather and land use. His project changed slightly as he was not close enough to look at crop changes.

The challenges in this project have been the high dimensionality of the data which has redundancy and/or repetition. Looking at images from Harnhill, a clear radial pattern is seen, which relates to the pigs on the land. Methods used have been spectro radiometry, recording crop height and density, tillering (counting shoots thrown out from plant), airborne measurements. The data has been analysed using self-written software for spectral analysis. Dave showed examples from Diddington taken over 4 weeks between June and July 2011. On the earliest photo, crop marks are seen as variations in greenness and biomass. Two weeks later there was still variation in colour and differences in the ripeness of the crops. Two weeks later, before harvest, differences in biomass were clear but not colour. Dave used spectral analysis and normalised the effects of the biomass. He looked at an absorption index, or ration of minima and maxima, to see changes over time. Lidar data was analysed to look for biomass variation. Dave showed lidar results from Harnhill (Quarry Field) where he subtracted the lidar data for bare soil to that with the wheat crop present to get a model for the mass of vegetation. Lidar use for crop marks has not been generally explored. From the Diddington clay site, the lidar results on vegetation height were not as clear as the aerial photo. Using absorption ratio gave much clearer details of ditches. No single sensor is clearly better. At the Diddington free draining site, there was good definition of the features on the aerial photo but not as much from the lidar and much less detail shown using absorption ratio. The data was collected the same day at both Diddington sites. Different sensors seem better at different sites but further work is being done to explain this.

Ant Beck Bringing it all together In archaeology, we collect evidence and then connect those dots of evidence through theories. There are many possible models and as we collect more evidence we cross off some of the theories. Theories include assumptions, we prefer those which use the fewest assumptions. Different models may be equally valid for different questions. We want to pull all available data together. Some data we have no access to, there are some things we are not aware of yet. DART has a range of data and we want to change how people think and work.

DART has an open science philosophy, all data should be in the public domain and accessible. This has been difficult to do. There is a data repository which uses CKAN and open licences. We encourage tracking of the data use. Demonstration of the DART portal. We see data at a summary level and can select one item. We see metadata, and we can cite the collection of data. We can select an individual object and then see metadata specific to that object. DART has an integrated spatio-temporal database with all relationships shown in UML. We can use collaborative analysis, as the code is openly available, we can take python code and run the same processes. This gives an audit trail giving the full process from raw data to final results. From this, others can understand the assumptions made in data processing, for example filters applied to data. Through this, DART gives access to data which is traditionally inaccessible, and through the open process, others can check whether the results are robust. An open process is not a black box and encourages collaboration and more efficient engagement. Through DART we see synergy between the goals in different areas. Open science improves the impact. We encourage an open citizen approach in which different groups come together and this can be small scale. Why have an open science philosophy? The EU says so. The UK government say so. The Royal Society say so. The EPSRC say so. Please use it.

David Jordan Modelling the data Mainz soil geoarchaeology David uses soil methods to look at structures. In Mainz they make detailed analyses of structures by using techniques such as micromorphology, conductivity, studying samples from excavations, and field surveys. They try to describe structures more rigorously or mathematically using models. The typical representations of an archaeological site are through photos and drawings. They can create models of distributions of properties, for example of resistivity and compare with what they see in a survey. Models can be based on the physics of soils with properties such as conductivity or resistivity and built in terms of pathways

through soil. They can take known soil properties and calculate the expected electrical properties. Using a crude model, for example with ditches and layers, they can predict electrical resistance. In the model they can change the conditions, without, for example, pouring water on the field. Realistic soil models are very difficult to create. Archaeology makes things vastly more complex as it interacts with the environment. There are problems knowing when to target methods on a particular place or time. Use of models on the DART sites. At the Diddington free draining site, there was saturation at the bottom of the ditch. They generated a description for how the hydrology controls the geophysical properties detected. For the Diddington clay site, they created a model with stages and used the daily rainfall figures from Cambridge. There were changes in water content towards the end of the fieldwork period. They used the water content properties in a model to calculate the expected resistivity at different strata and compared the calculated to the measured resistivity values. They had problems at the boundaries between materials. The equation used was bad for gravel, this is still being investigated. They can continue to use the DART data to try to improve the models, this is work in progress. One problem is the way that structures in the ground are traditionally translated into strata, but the materials in the ground are actually continuous and there are big variations even within individual strata. In the translation from the physical environment to the record, much information is lost. This is a necessary simplification. This makes it very difficult to model from archived information. A model may be built on particular properties, but when you displace only a very small distance, the material of the soil may be very different, for example different fills of a ditch, so the purpose of modelling is less clear. Models are only valid in some circumstances. In a different example, images taken with different sensors showed an army camp in the US. Different sensors gave different clarity. Principle component analysis was used to combine the images without referring to the processes. Hope to be able to go further when including the physical properties of soil. Need more understanding of how models can represent soils, for example, pathways of conductance.

Tony Cohn The future: mining the data University of Leeds Tony is not an archaeologist but has done a lot of work with video data which are spatio-temporal and have uncertainty.

The DART data are described by a UML diagram. There is a summary of the data. Some of the data is static, for example the trenches were only dug once. Other data have spatial and or temporal aspects with different levels of granularity. The data are very heterogeneous, describing different kinds of things. Data mining can be split into supervised and unsupervised. In supervised learning, concepts which we want to recognise in future data are labelled in existing data. The machine learning then finds features of the data which are relevant to that concept. In unsupervised learning, the machine finds interesting patterns in data. It might be interesting that one feature remains constant. The machine can try to find clusters within data, similar data points which go together. Data mining techniques typically involve a lot of pre-processing of data. The representation chosen for data is often more important than the actual algorithm used for data mining. Choosing a representation is a black art and depends a lot on the skill of the person doing it. A representation which is more expressive allows for more features, or more values which a feature can have allowing for more nuances in the learning, but needs more data and may result in over fitting the data. Typically a less expressive representation is used. Another issue is the need to discretise continuous data. Many different algorithms can be used for data mining, some are better at some concepts. WEKA is open source software with many algorithms, but often people write their own. Tony gave an example of video analysis in which a high level description of whats going on is given and we want to be able to explain future videos. What would we like to mine from the data? For example trying to automate a decision about when moisture context is most accentuated. It depends what is interesting to the archaeological community. Remember the data are online and free.

DART research discussion session Introduction by Chris Gaffney with a short overview of whats been covered. The DART research has generated plenty of questions. Along the way there have been errors in methodology but if we dont make mistakes we dont progress very far. Q. For Dan on the contrast of the 2 probes used. Dan. The IMKO probes were in bore holes about 2 or 3 metres away from the section. In the trench there was the problem of the soil drying out, which was not

seen in the bore holes. However the borehole tube may have allowed water straight down. There is still a question of how much the digging of the trench affects the results. Q. (David Jordan) for Rob. Have you calibrated the TDR for resistance measurements. Rob. The TDR sensors were initially calibrated against standard data. Q. (Paul Cordes) Does farming affect the data, e.g. fertiliser application? D Stott. Yes. Harnhill is organic, but at Diddington we see the use of tractors. Ant Beck. Harnhill includes crop trial areas where different ploughing techniques etc. are being assessed. Effect of fertiliser on conductivity is an interesting question, would have to go to Cranfield on that one. Q. On the selection of sites did you look at land use or farming practice, for example using 19th century records? Chris Gaffney. If we had had more complete initial knowledge, we would not have chosen a site with a ditch with a pipe still in it. Ant Beck. Access to the sites was the most important factor in site selection, sites with the different soil types in close proximity, and permission from landowners. Q. How do you extend the results to the wider landscape? Land use may be a factor in how they can be extended.. Ant Beck. Scaling up is a problem, we have lots of data on four small holes. As well as data we have a network of people who can feed in their insights. Soil marks can be extrapolated, soils are roughly similar in large parts of the country. We have created a framework which others could use for similar projects. Q. (James Lyell) Are you worried that the last 3 years have not had representative weather, for example the driest winter. Chris Gaffney. We have seen highs and lows which do not match the expected seasons. This may be useful as we saw a drought cut off point in the results. Toby Mottram. We are trying to develop an inventory of greenhouse gas emissions. We find that the last 3 years actually match the norm and are not an extreme period. David Jordan. It is important to understand the soil whatever the weather and so it doesnt matter if its extreme. Sue Stallibrass. Things can be very variable even within an individual site over time. Comment. Most archaeology happens due to planning regulations, that affects the timing of geophysical surveys. Chris Gaffney. Curators do not generally specify earth resistance as the main measure. We will come back to these points this afternoon.

Comment. We have so far been avoiding discussion of the afterlife of DART. Ant Beck. There are 3 PhDs to finish. Ant himself is still looking at the data. They are making an application for follow on funding, but that would not be to undertake new research. Funding is difficult to acquire which is partly why we wanted an open project. John Wells. DART could be very useful to community groups who can get out to sites and use for example kite aerial photography. David Jordan. We would need more data for example to follow moisture content. John Wells. Groups can go back to sites year after year and take images in visible and infra-red. Ant Beck. Part of the follow on application is to look at reasons why data are not shared readily. It is possible to build small scale devices, for example using arduino. Data can be stored in open formats. Data can be available in the proprietary format but also convertible to be used as required. With the proprietary environment it is a black box and you dont know what processing is done. DART data is open and accessible through URI without login or payment. Each photo etc. has its own URI and is citeable. You can cite at different levels. The URI is not persistent, to 2015 as a start. We still need to determine what data would need a persistent URI. Ant is looking at getting DOI so that when the data moves, it can still be referenced using the same reference. Q. Has the soil analysis data been incorporated into the PhD projects? A. Each PhD is a separate project and this data is not incorporated so far. Chris Gaffney. There is still the issue of fusing strands, the PhDs cant integrate all the available data.

Afternoon session Practitioner impact Armin Schmidt. Try to stimulate discussion on what is in DART and thoughts on it. From a geophysics point of view, can we make predictions about how or when to do resistance surveys. Archaeological features impact on the soil and plants and then we take measurements. The problem is that we have to invert this process to identify features from data.

Initially we though DART would solve this problem, but Rob has shown that we still dont know. Propose a question. Is it feasible or desirable to be able to predict the best survey methods from other parameters, for example soil types? Martin Roseveare. Yes. But mostly commercial surveys dont have information like land use, soils or rainfall history. Christine Hopwood-Lewis (Kat). When the data is available it is useable. Some data is downloadable, for example rainfall data for England and Wales. There is talk about making data available, for example the environment agency. Data can be available depending on circumstances. We might be able to add a charging system. The UK government has an open data policy. Universities can access data for research already, but charges may be made for commercial use. Martin Roseveare. There shouldnt be extra costs in making data available, thats the ideal situation. Ant Beck. Recently there have been big changes in the OS with regard to their funding model. Chris Gaffney. Remember that free data doesnt mean its the best or the most recent. Armin Schmidt. Even if we have the data, Martin would you go back and suggest that you wait before doing the survey? Martin Roseveare. Yes, we would make the case. Simon Crutchley. If you were told that you have to do the survey now, could you do something? Martin Roseveare. Yes, I could caveat the results differently. Comment. When its commercial work, sometimes you have to do the survey as they pay. Chris Gaffney. You can say that the outcomes are rubbish or suggest another technique. Comment. The data may not be predictive, you cant know the weather in advance. If you think theres a ditch but dont see it, you can try to explain why not. Some sites are visible on a survey for 6 months of the year, others there is only a 1 month window. Can you create a characterisation so that you can classify which situation applies to a site, or how sensitive a particular site is to the conditions? Or can you compare a site to a benchmark? Chris Gaffney. Most commercial work is on land not sites and there may be sites within that land. It can be difficult working in a field about which nothing is known.

David Jordan. If we could produce a set of simulations and present a client with a set of options then that would be more concrete than just telling the client that you dont think it will work well. Models allow you to make concrete something that we interpret personally. Kat. For ecologists we can usually break the time of year into categories like nonoptimal, sub-optimal etc., for example to exclude disturbance during the breeding season. Could we add archaeological constraints, for example when crops are planted? We could then have a list of alternative times.

Oscar Aldred Lets look at the perspective from the air. DART has a potential impact of looking at correlations between images (e.g. lidar) and the interpretation in terms of underlying features. There is so much data, as a non-scientist I would like to know how best all that data could be dissected and what types of archaeological questions the data might help answer. Back to the idea of not necessarily being able to specify the best time some aspects of DART may help to explain why we see nothing, for example in clay there is a good chance of not seeing anything. Comment. There is not necessarily a lot of new understanding coming from more data. There is a national map of soil types which might give a chance of crop marks, in deep soil there is less chance. For crop marks, we usually think of visible light but they can also be see in the near infra-red, images should be taken when it is overcast. Thermal cameras are becoming ever cheaper and thermal imaging is potentially very useful. One can look at thermal crop marks in any period of transient dryness during the growing season. Ant Beck. The discussion so far implies we are building an bermodel, but what would be more possible would be to constrain results based on sensor data for example subsurface monitoring. We could look at giving a reference site for a particular type. Cost is a problem with heritage management, e.g. DEFRA. Q. How do we deal with the vast amount of information already there, for example soil moisture figures? Comment. An important issue is land use change, for example increasing areas of land under plastic, such as to force wheat crops. This needs on-going monitoring. Kat. In Natural England we are creating character area descriptions and looking at future trends, for example growing of energy crops, wind turbines. So some of the information does exist.

Oscar Aldred. English Heritage are looking at historic landscape characterisation.

Keith Wilkinson. So far the focus has been on resistivity and aerial data. DART is about much more than that, it seems that the project has come to an end too early. The data are all in one place, also in the database are magnetometry, magnetic susceptibility, EM surveys, but not over entire timespan. We could look for correlations between the magnetic or moisture data with the trench data, for example the grain size data. We could do the same for within and outside the ditch. A by-product of DART could be to look at how archaeologists recognise contexts and how these relate to, for example, grain size. Keith Wilkinson. Geotechnical engineers use borehole data which is recorded in a different way to archaeological data. The different practitioners could compare and extract each others meaning. David Jordan. English Heritage have a report about this which contains case studies on how archaeological details can be extracted from geotechnical data. Matt Guy. Geomatrix. You could consider volunteer based follow on work, by outlining small bits of work required from the data, asking volunteers to do the task and create a structured report. Ant Beck. Yes, part of the follow on fund application concerns communities of practice. There is the challenge of getting set up, the data are not a problem but the problem is in getting started. It is difficult to set the problems to be tackled as different people are not necessarily interested in the same problems. The data are also available for example for final year or masters projects. Q. If anyone contributes then there is the problem of quality, have you considered a peer review type of idea? Ant Beck. We have considered hackathons, with people with different expertise in the room sharing ideas and techniques. Peer review is hard as if people feel they are being judged, it puts them off contributing. If the code and ideas are open, anyone can review them. James Lyall. Community projects tend to focus on work in their own local areas, with interest in local spots. Chris Gaffney. Also communities want to see outcomes. Ant Beck. We are talking about different types of communities, not necessarily based on locality.

Armin Schmidt. There is a difference between community archaeology groups and science or computing geeks who may not be interested in archaeology as such but in problem solving. Chris Gaffney. Then they would need some kind of direction. Q. Do you see DART as a reservoir of masters projects, what role is there in directing the next generation of researchers? Keith Wilkinson. Its an easy hit for masters projects as the data have already been collected. Chris Gaffney. Students also need experience of collecting data themselves. They could use DART to examine a particular aspect and then test elsewhere. Armin Schmidt. Open data have many problems. Research is a competitive environment and so we have to protect our assets. How do we protect our intellectual assets, for example each PhD has to be novel and we cant compromise this? We need funding to make the data useful, just being on a server is not enough. James Lyall. I really appreciate the open data aspect of DART. Local societies could use the same processes to put their data online. Comment. I share Jamess enthusiasm. It is important that the data are citeable, if someone reuses an idea, they have to cite it. Armin Schmidt. We are in a transition between protecting and making available. Chris Gaffney. Regarding REF 2014, the impact of work has a bigger effect on funding than before, this will be more so in 2020. The quality of publications is important, it makes more sense academically to write good papers and then release the data. Ant Beck. We have been in a rapidly changing environment ever since the start of the project. Here at Leeds there was a digital humanities event but they have no repository. Universities have to catch up to allow reuse of data. We need to build relationships so that groups can share and reuse data. ADS needs to be more than an archive, we need to make data available for reuse. There is a question of when and how to provide data. In an interdisciplinary project there are different practices for dealing with data. Q. Will people still be able to add to the DART archive. Ant Beck. We have a server until 2015, afterwards much of the data will go to ADS. Open data are only open because of the licences used. Ant Beck is working with the heritage lottery fund on data licencing. They specify by attribution and non-commercial. So commercial users would have to request exemptions.

There is a danger of getting islands of data. James Lyall. The data show anomalies or crop marks but archaeologists want interpretations. Is there a mechanism for people to add interpretations and feed into the archive? Ant Beck. We could build an open archaeology map, but this is not that kind of archive. There are ethical issues, would night hawks misuse such information?

Policy/Curatorial impact Quinton Carroll not available. Dave Cowley. Open data are a fundamental outcome of DART. Government policy is only partially implemented and much useful stuff, for example high resolution images, is behind pay walls or restrictive licences. As an aerial surveyor for the royal commission, reliability of data is an issue. For example in aerial photos what you can see depends on the weather. Most archaeological distributions reflect survey bias. How do we address the reliability of data? In earlier discussion, commercial contractors may request survey at the wrong time. What role does DART have in making more informed curators? Martin Roseveare. Contractors dont have any contact with curators. The archaeologist is in the middle, unless we can persuade the archaeologist to go back to the curator. Kat. Our practice is to give constraints and then ask the contractor to suggest a suitable technique, this gives flexibility. The farmer asks for the survey and then gets paid back through grant aid. Kat was an archaeological fieldworker in the past and has experience of county archaeologists doing a spec and not asking what the best design would be. The curator may change a project, which changes the cost and developer doesnt want that. Sue Stallibrass. Usually the details available are just 25 hectares pasture. What is the best way to seek further information? Comment. It is difficult to see where the DART output should feed in. At the technical specification point? Comment. Theres no such thing as a standard brief.

Chris Gaffney. Can we change tack a bit and return to a point Sue brought up earlier?

If we had to choose probe separation, we might choose 25cm, but that does not see deeply enough to look at ditches. Theres a lot of information locked at the bottom of a ditch. There are two problems, rapid changes and depth. In a previous workshop we discussed EM systems and measurements at 3 depths between 0.5m and 1.5m Using electrical and magnetic properties gave a good estimator of conductivity. We saw archaeology in one channel on every site tried. But these were sites we were not going blind. EM has also been used in Ireland where they claim that they dont have a seasonality problem. But they do as at certain times, the ground is too hard for resistance. Using EM you can still measure contrasts. So new EM instruments might take out some of the guesswork. Kat. We want to be able to characterise the extent and significance of features. EM gives the character but not significance. Chris Gaffney. Why not? Perhaps your information is based on old instruments. New versions are hand held like a magnetometer. Ant Beck. Remote sensing is not just about features, but about decision processes, like the risk to whats there. There can be a change to risk when taking land out of pasture. New technology can change the risk. Kat. Farmers dont want to be paid to take land out of use, they may earn more for crops. If we scanned the whole country, every farm would go out of business. We need to look at risk and significance to determine what to actively protect. Ant Beck. Is scheduling an appropriate mechanism for protection? Technology can allow data to be streamed to tractors to raise the plough over areas judged higher risk. Kat. Lots of farmers dont like precision farming because of the impact on weed management for example. Some would be mortified if they thought they had damaged a site, others focus only on money. Martin Roseveare. Most commercial geophysical surveys are for developments for housing etc. Comment. How does DART communicate e.g. to give guidance to local councils, through ALGAO? Comment. Could discuss with ALGAO so that they know to provide soil conditions etc. Should this be part of the follow on fund? There is a difference between publications and a short guide. This is a crucial point for DART having impact.

Armin Schmidt. You could give questions which should be asked but not a decision tree. James Lyall. People only want to pay for whats in guidelines and not go further. Comment. Difficulty being prescriptive, they need to be guidelines.

Community impact John Wells West Lothian archaeology is not just local to West Lothian its a national co llective. John gave some background. He had no background in archaeology but set up a community website in 2006 and flew cameras on kites to photograph an old mining town. He found that local archaeology was not good, housing developments were built without surveys. John formed the archaeology group in 2010 for people with all backgrounds. People throughout the country wanted to join and use kite archaeology. Last year they set up a fund to give away kite kits. People think that kite flying is difficult but its easy with a bit kite. Community groups cant look at full spectrum, but its cheap to use visible or near infra-red which have lots of potential. The techniques are simple and fish eye lenses are available from under 200. The information they would like from DART is what conditions are best for aerial photos? Ant Beck. UAVs and kite techniques are more effective in communities. John Wells. I dont agree about UAVs. In the community there may be only 1 or 2 people able to analyse data but 12 year olds can use a 700 kite kit. The Scottish National Aerial photo scheme is wider than just Scotland. It is difficult to get groups to use kits, so we have been targeting people in organisations like the National Trust and giving them a standard kit. They go out and are getting good results. Kites dont make it a toy. Five year olds are willing to try it, but professionals less so. A secondary school in Ireland held a remote sensing course and we provided kites and cameras. Since then West Lothian council has become interested. Derek Magee. How does the data analysis work? A kite is less well located than a satellite. John Wells. You can put ground reference points in. Initially we wanted to publicise the archaeology of West Lothian. Comment. UAVs can be gps enabled so you could study temporal changes.

John Wells. With UAVs you would get specialists in the group. Kat. A group has better longevity if there is a leader. What about nature improvement groups? John Wells. A local agricultural college wanted instruction for their teachers in aerial photography. This is due to start soon funded by the council. Kat. Scotlands rural past has a good website. John Wells. We are starting to show aerial photos taken using the kits on the website, just Google West Lothian archaeology. Q. Could you use smartphones, they may have gps and infrared. John Wells. Yes we have done. Comment. When aerial photos are available through the year, if soil data are also available that might give more useful information. Chris Gaffney. We could see if the same sort of responses are seen elsewhere. John Wells. So far, the archaeology side is slower than the environmental side, e.g. the RSPB. People are also starting to make 3d models from the data.

Building Bridges Effective collaborative networks Bob Evans Building bridges things to consider Do they want to speak to you. It takes a while to get people interested Find out what people want Need dialogue and common ground. Bob described using soil surveys to tell planners what they needed. Using potential soil moisture deficit, you can work out when crop marks will show. But do evaporation and transpiration always occur at predicted rates, or are there more constraints. Bob showed a case study on levels of nitrates in streams where there is an EU directive on the quality of streams. Farmers are told not to use fertilisers when the field drains are running. The data show that when in solution, things can reach the river even when the field drains are not running. It is important to be able to access data.

Toby Mottram

Experience in telemetry systems for milking data, agricultural/engineering data and sensors. We should develop low cost sensors which can be left without environmental problems. Agricultural issues impact an area. How do we make data available for checking etc.? Drive towards more sustainable systems More food to be produced from less land. Theres a move towards arable farming Theres more plastic and tunnels used More animals are being kept inside, for example to keep cattle from contact with badgers. What is already available? Farm vehicles often have gps. Government surveys. Things are driven by money, we predict less traditional arable farming and more high value vegetable crops, these would be less regular in aerial photos. Automation of data management, this takes a lot of investment to get it to work. Chris Gaffney. We are coming to the end of the project but do not have all the answers. Technology doesnt answer all the questions. Ant Beck. To try to put the work into a broader context, we dont have land or resources so we need to work with other organisations. We have a wonderful relationship with RAU. Comment. Over the last 20 30 years soils have become more compact. Kat. There are similar interests at Natural England which promotes catchment sensitive farming based on run off but my be ignorant of the archaeology. Policies are based on run off rather than flow through the soil matrix.

Ant Beck. Final Remarks. Regarding policy, we want to bring something to ALGAO and IFA. We need to engage curatorial environments. If people think of anything weve missed today then please send an email. Ant made thank yous.

Tony Cohn thanked people and hopes that some of those present will be involved in follow on work.

If you do use our data, wed love to know, also contact us if you think we can help you.

You might also like