Spot that boundary…

DSC_0976 copy

Spot the boundary. It’s very slight, but it is just about visible on the ground. Scroll down for the answer…

Having put my various datasets into the GIS environment (see previous posts), the time has come to put them to work. First up – and most time consumingly – to map the known coaxial boundaries within the national park. This might sound like an obvious thing to do, but many of them have not been accurately recorded before. Some of the large systems of Swaledale have been systematically surveyed as part of the Swaledale Ancient Land Boundaries Project (their website/reports are here) and several systems in Wharfedale (between Grassington and Kettlewell) have had attention paid to them by antiquarians and more recently (see Raistrick 1937, Horne & MacLeod 1995), but many are recorded in the HER only as a ‘coaxial field system’ without details pertaining to individual boundaries.

By using a combination of digitized sources, I have been able to map the remains of the boundaries in 24 known coaxial systems. Perhaps the most useful of these sources (beyond the HER, which gave me an idea where to look) was the aerial photographic imagery available through the ESRI ArcMap setup. Other sources of aerial photography are available, but mine was provided by Bing!, which worked out well as data provided by some of the others are not so clear for this part of the world. Where available, the lidar data complemented this photography nicely, illuminating lumps and bumps that were not otherwise visible due to the lighting or vegetation. It’s amazing, though not surprising, how much easier it is to get your head around a system when it’s seen from the air – and slightly ironic, considering the people who built it didn’t get to see it that way. Of course, the detail is always on the ground, but this approach gives a useful broadbrush overview – a starting point for further investigation.

The maps below show the coaxial system near Horton in Ribblesdale – like several of the other systems, you can walk though it (there is a Natural England self-guided walk through the nature reserve in which it is located here).

Horton_topo copy

Coaxial field boundaries at Horton in their topographical context. (Contour data provided by OS Terrain 5 data service via Digimap.)

Horton_ap copy

Coaxial field boundaries (in white) at Horton in their geographical context.

 

arrows copy 2

…the boundary runs straight towards the camera, between the two red arrows. Much easier to see from the air (although, admittedly, this is a particularly-difficult-to-see example!).

Advertisements

Data and Deliberations: Part 2 Lidar

As you probably know, the availability of lidar data (lidar stands for Light Detection and Ranging) has taken the archaeology world by storm over the last decade or so. The collection of airborne lidar data involves sending out laser pulses from a plane, which are reflected back from the point at which they hit the ground (or any intervening objects). Given that the speed of the light is known, the distance each pulse has travelled can be calculated from the time it takes to return; in conjunction with very precise GPS positional data for the plane, the result is an accurate 3D location of each point in space, and therefore an accurate model of the land surface. This can then be viewed in a GIS, where it makes a superb tool for detecting subtle earthworks, such as coaxial boundaries, against background clutter.

Unfortunately, there is not 100% lidar coverage of the Dales yet. The lidar I am using has come from the  UK Environment Agency, who collect data as part of flood monitoring and environmental asset planning programmes and therefore target specific, required areas. The image below shows the 1m resolution (there is a point value for every square metre) lidar data available for the Yorkshire Dales National Park. It is very clear from this map that most of the data collection has focussed on the river valleys, with coverage often not (or only just) reaching the top of the valley sides. Which makes it difficult to rely on when prospecting for coaxials, which often, frustratingly, survive just above this level!

Figure 1: 1 x 1m lidar data for the Yorkshire Dales National Park (Data copyright Environment Agency 2015).

Figure 1: 1 x 1m lidar data for the Yorkshire Dales National Park (Data copyright Environment Agency 2015).

In places, however, coaxial field boundaries do show up clearly in the lidar data, running across the contour as in figure 2, and I have been mapping them from the lidar in conjunction with aerial photographs and maps, in order to characterise the individual systems. Lidar data can be viewed in various ways: figure 2 contains a simple 2D hillshade plot that ‘makes sense’ to the human eye, but various parameters can be exaggerated, or the point values used to analyse assorted elements of the terrain. The data can also be viewed in ‘3D’ (as in figure 3), which helps to visualise and understand relationships between the archaeology and the landscape. Usefully, it is also possible to drape other data sets over this digital elevation model – such as the aerial photography in figure 4.

Figure 2: Coaxial field system boundaries visible  north of Grassington in 1m resolution lidar.

Figure 2: Coaxial field system boundaries visible north of Grassington in 1m resolution lidar (Data copyright Environment Agency).

Figure 3: Hillshade plot  showing part of Upper Garsdale, viewed as a 3D surface model. It is possible to rotate and 'fly through' the model landscape. (Data copyright Environment Agency 2015.)

Figure 3: Hillshade plot showing part of Upper Garsdale, viewed as a 3D surface model. It is possible to rotate and ‘fly through’ the model landscape. (Data copyright Environment Agency 2015.)

Figure 4: The height data from the lidar has been used as a surface over which to drape this aerial photograph of Halton Gill, Littondale. (Data copyright Bing Maps/Microsoft/Environment Agency 2015.)

Figure 4: The height data from the lidar has been used as a surface over which to drape this aerial photograph of the valley side at Halton Gill, Littondale. (Data copyright Bing Maps/Microsoft/Environment Agency 2015.)

One of the bonuses of lidar data is the facility to alter the direction from which the (artificial) light source is shining on the landscape – depending on the orientation of individual features or their position in the topography, they may not be easily visible when illuminated from any given direction (as is the case in an aerial photograph, for example). Figure 5 shows the same medieval lynchets in Wharfedale illuminated from 3 different directions, and the difference this has on the visibility of the archaeology is very clear. The same goes for the height of the light source above the ground (think of the difference between viewing earthworks at midday and in late afternoon!). Obviously, while particular light directions and heights reveal more detail in landscapes, the flip side of this is that others cause detail to be obscured and it requires a little trial and error to work out which are the most advantageous light source positions.lynchets0lynchets45

Figure 5: Medieval lynchets near Grassington, Wharfedale, illuminated from 0º (top), 45º (middle) and 315ª (bottom) azimuth. Note how the visibility and prominence of features on different alignments varies under different lighting conditions. (Data copyright Environment Agency.)

Figure 5: Medieval lynchets near Grassington, Wharfedale, illuminated from 0º (top), 45º (middle) and 315ª (bottom) azimuth. Note how the visibility and prominence of features on different alignments varies under different lighting conditions. (Data copyright Environment Agency 2015.)

One way to get around going through 8 or 16 different images by hand for each area, is to use a combination of the images, which essentially includes the ‘best bits’ of each one. For my first foray into the world of computer coding – aided and abetted by my infinitely more code-minded colleagues – I am in the process of writing a piece of code that does this, by applying principal component analysis. You can see the difference between the images below…it’s often only subtle, but in places it may make interpretation just that bit more accurate.wharfe135a_06_07 copy wharfe090a_06_07 copy

Figure 5: The first two images show hillshade plots of Knipe Hill, Kettlewell, lit from 135ª and 90ª respectively; the third image shows the results of the application of principle component analysis, in which elements of these plots were combined with 6 others lit from various directions. Note how each emphasise different features. Click  to enlarge. (Data copyright Environment Agency 2015.)

Figure 5: The first two images show hillshade plots of Knipe Hill, Kettlewell, lit from 135ª and 90ª respectively; the third image shows the results of the application of principle component analysis, in which elements of these plots were combined with 6 others lit from various directions. Note how each emphasise different features. Click to enlarge. (Data copyright Environment Agency 2015.)

You can find out more about lidar here: http://content.historicengland.org.uk/images-books/publications/light-fantastic/light-fantastic.pdf/

This is hardly a scientific test, but…

…the following ‘word clouds’ give a quick idea of the contents of some of my datasets. Wordle is an app that generates such clouds: the relevant text is fed into the app and the outcome is a nebulous collection of words that condense in such a way that the most important or frequently used keywords are rendered most eye-catching in the final cloud. So, if I was to round up all the records from the Historic Environment Record database that refer to prehistoric sites, features and finds… I would end up with a cloud like the black one below. It’s interesting that the word ‘field’ features comparatively prominently in this cloud, giving an indication of the importance of the field systems among the other recorded prehistoric sites. It is also somewhat telling that, when the text of the records relating to the prehistoric field systems alone is used (bottom), ‘unknown’ is one of the larger words.

Fig. 3 Word cloud representing the contents of the ‘prehistoric’ HER records. ‘Field’ appears centre-bottom, right and centre.

Word cloud representing the contents of the ‘prehistoric’ HER records. ‘Field’ appears centre-bottom, right and centre.

Word cloud representing the contents of the ‘prehistoric field system’ records from the HER. Note the relative prominence of the keyword ‘unknown’!

Word cloud representing the contents of the ‘prehistoric field system’ records from the HER. Note the relative prominence of the keyword ‘unknown’!

Data and Deliberations: Part 1

This is the part of the project where, having acquired my datasets, I have been cleaning them and converting them into a format I can use in my GIS. This has been more longwinded than it sounds, and what follows is a bit fiddly, but it’s a crucial part of the process. Don’t forget to click on the images to see bigger versions. The main datasets I have been wrestling with recently have been the Historic Environment Record database (kindly supplied by the National Park Authority), the National Mapping Pilot Project data (kindly supplied by English Heritage), lidar data (available from the Environment Agency) and aerial photographs (at the moment, I’m using a combination of prints from the HER and a layer available from ESRI that is similar to google maps). Historic Environment Record There are over 80 HERs in England, each of which contain details and records of the known archaeological sites, finds, interventions and surveys that have been recorded/have taken place within their area of jurisdiction – in this case, the Yorkshire Dales National Park. This makes them one of the major go-to sources for archaeological research. The Yorkshire Dales HER contains nearly 32,000 entries, the majority of which contain details including location and interpretation, and can be searched by various categories. But while the database is a mine of information, there are the inevitable limitations: for example, it has been compiled by numerous people over the years, resulting in a plethora of conventions and standards.

Fig. 1 Each of the red dots on this map of the Yorkshire Dales represents one of the 32,000 records from the HER.

Fig. 1 Each of the red dots on this map of the Yorkshire Dales represents one of the 32,000 records from the HER.

Obviously the records pertaining directly to the individual field systems are of interest, but so is the rest of the database – it provides a useful context against which to examine the field systems, acting as a background view of the prehistoric landscape. Fig. 1 shows the ‘raw’ data with each separate record represented by a dot. The distribution is interesting: the majority of known archaeological sites and findspots cluster in the valleys, with more white space visible on the higher ground – is this a real distribution of past human activity, or a reflection of present activity happening away from the less hospitable peat covered moorland where archaeology-spotting is more challenging?

Fig. 2 Prehistoric features and finds from the HER.

Fig. 2 Prehistoric features and finds from the HER.

So the first step was to sort the individual records into broad chronological phases, based on the ‘from’ and ‘to’ fields that have been filled according to the interpretation of the inputter. These can be selected from a ‘thesaurus’ of terms, such as ‘Late Iron Age’, so there is a reasonable degree of uniformity, but there is also a spectacular number of different combinations, making life difficult for me. I went for the low-tech approach and, with the help of the ‘find’ function in Excel, removed records for all those sites of unknown date and later-than-prehistoric origin. And then the remainder were broadly categorized into Early Prehistoric, Late Prehistoric and Prehistoric as it was felt that the majority of the features could not be accurately dated more narrowly than this. Their distribution is shown in fig. 2 below. National Mapping Project   The NMP dataset is a body of data that dates from the 1990s. It was collected as part of a project conducted by the Royal Commission on Historic Monuments of England with the aim of identifying visible archaeology from aerial photos: the Yorkshire Dales served as a pilot project to assess the effectiveness of the method over uplands. Aerial photographs from the Dales were transcribed, and recognizable features recorded with the use of a system of symbols. As it was done with pen and paper it has since been scanned to produce a digital copy.

Fig. 5 A tile of data from the National Mapping Project. The dotted areas with arrows represent ridge and furrow, the solid lines mark lynchets. The crosses are grid reference points.

Fig. 3 A tile of data from the National Mapping Project. The dotted areas with arrows represent ridge and furrow, the solid lines mark lynchets. The crosses are grid reference points.

When faced with a national park-worth of this data (fig. 3 shows how it arrived), it is quite overwhelming and difficult to work out what’s going on. It arrived as a set of raster files, so the first step was to vectorize it to make it possible to select each symbol individually. This also allows the removal of any smudges, grid reference points or paper ‘edges’ that are not real features. Then it was just a case of assigning each polygon to an appropriate layer category using AutoCAD. When I say ‘just…’, it was actually very fiddly and time consuming, but as a result I have a series of useful, colour-coded layers to add to the GIS (fig. 4).

Fig. 6 An area of NMP data near Kettlewell. The dark green polygons are features interpreted as field boundaries and lynchets; light green represents ridge and furrow; brown dotted lines are tracks; brown represents buildings.

Fig. 4 An area of NMP data near Kettlewell. The dark green polygons are features interpreted as field boundaries and lynchets; light green represents ridge and furrow; brown dotted lines are tracks; brown represents buildings.

It is not possible to assign dates to the features in the NMP data but it was possible to assign them to broad categories that cover, for example, evidence of extraction industries, ridge and furrow, settlement, and lynchets and field boundaries. Of course, some of those field boundaries belong to prehistoric field systems; the aerial perspective makes them relatively easy to identify (much more so than on the ground). The NMP report (Horne & McLeod 1995) identifies at least 35 prehistoric coaxial field systems. These are shown in fig. 5, alongside those known from the HER data (with varying degrees of confidence), and it is clear that there are already some overlaps and underlaps between the datasets. The NMP dataset in particular gives good coverage of the landscape as a whole, and will, for example, help to explain ‘gaps’ between recorded field systems where more recent ridge and furrow or lead mining is present.

Fig. 7 Comparison of field systems as identified in the HER and NMP datasets.

Fig. 5 Comparison of field systems as identified in the HER and NMP datasets.

One of the main reasons for going through these datasets so thoroughly by hand is that I am now much more familiar with the data. I have moved on to processing the lidar data that exists for the Park – more about that next time… Horne, P. & McLeod, D. 1995. The Yorkshire Dales Mapping Project. A report for the National Mapping Programme. Air Photography Unit: Royal Commission on Historic Monuments of England.

Visualizing the prehistoric landscape

Conveniently for me, my dad is a graphic designer. Also conveniently for me, he is retired and can be persuaded that he has time on his hands to apply some of his skills to his daughter’s phd project… hence the image below, which I hope gives some indication of what the landscape at Conistone Dib, Wharfedale, may have looked like in use in late prehistory. It’s a combination of dad’s illustration and my photograph, and is based on the archaeology and geomorphology of the area that is coming into focus in this project.

“Oh no, it’s crow stew again for tea”. Two iron age kids entertain themselves while the grown ups look after the animals. Illustration: Bryan Brown.

“Oh no, it’s crow stew again for tea”. Two iron age kids entertain themselves while the grown ups look after the animals. Illustration: Bryan Brown.

Conference life

One of the definite perks of phd life is the chance to go to various conferences, meet other researchers, compare notes, get feed back on your project and, of course, experience the local sights and sounds. This month, two relevant conferences came along at once – the annual meeting of the European Association of Archaeologists, held in Istanbul 10th-14th September, and the 3rd Landscape Archaeology Conference, held in Rome from 17-20th September. It’s a really tough life, being a phd student!

The Istanbul meeting was a large-scale affair, with over 3000 attendees from all over the world and numerous parallel academic sessions. I gave a paper in an interesting session on Comparative Perspectives on Iron Age Landscapes, which was intended to juxtapose various projects and methodologies from across Europe and subjects ranged from architectural monumentality in Mallorca, to woodland in Anatolia, cultural landscapes of French oppida and the Northumbrian landscape. My paper was nothing particularly ground breaking – an introduction to the Dales and their archaeology, and an overview of the data collection phase of the project and some of my suspicions – but there were some useful questions and in the subsequent coffee break I chatted to several very interesting people, and was introduced to a couple of new-to-me coaxial field systems (in Lincolnshire and Hertfordshire) that may well prove useful comparisons.

The Landscape Archaeology Conference in Rome was a much more modest affair, but equally stimulating. On the whole, it was quite biased towards Italian, Dutch and German research; it’s always intriguing to see what is going on elsewhere, and sessions ranged from ‘archaeomorphology as landscape archaeology’ and ‘integrated approaches in landscape archaeology’ to ‘computational modelling in landscape archaeology’ and ‘methodological approaches to social landscapes’. I presented a poster and the poster session was definitely one of the highlights – held alfresco with coffee and fresh peaches!

Let’s go fly a kite…

Over the summer there have been a few days in Wharfedale with just the right amount of wind to lift a kite without ripping it to shreds. And on some of those, I have managed to get out and about and take some more pictures with the camera rig that the lovely John Wells at West Lothian Archaeology has provided. These images are still somewhat experimental, but some of the results are here – the individual pictures have been stitched together by hand using ArcGIS in order to get a feel of the surface archaeology with more detail than is visible on aerial photographs taken from a higher level. The problem still remains, however, that the field systems are whole landscapes rather than individual sites and as such are difficult to photograph satisfactorily ‘in one go’.

Prehistoric enclosures below the scree of the scar line at Conistone (north is down).

Prehistoric enclosures below the scree of the scar line at Conistone (north is down).

Prehistoric enclosures and field boundaries below the scar line at Conistone (north is up). Some of the coaxial boundaries are visible on the bottom left and far right of the image.

Prehistoric enclosures and field boundaries below the scar line at Conistone (north is up). Some of the coaxial boundaries are visible on the bottom left and far right of the image.

Scar line (bottom) and limestone pavement at Conistone. A couple of the coaxial boundaries are visible continuing over the scar towards the pavement - the fact that boundaries continue across the pavement suggests it was originally vegetation-covered and has since been subject to processes of erosion.

Scar line (bottom) and limestone pavement at Conistone. A couple of the coaxial boundaries are visible continuing over the scar towards the pavement – the fact that boundaries continue across the pavement suggests it was originally vegetation-covered and has since been subject to processes of erosion.