12 July 2014

Messing around with GPR

I'm still learning about my GPR system and what it is capable of, so recently I have been trying out a few things to find out what the results are like.

Whilst you wont see much of a cut in clay with GPR, since the geology and clay soil have similar EM properties, I wanted to try out what a cut in chalk looked like, since chalk bedrock is going to have a different response to chalk soil. Firstly I visited the BHAS dig at Ovingdean where they are excavating a medieval manor complex. They had a barn like structure with large square post holes, which showed up nicely. While the normal soil-chalk boundary was fuzzy due to the top layers of the underlying geology breaking down into a soil, the bottom of the post holes were a much cleaner cut with a much clearer change on the radar.

Pleased with this success, and happy that I now have something that will quickly assess chalk sites (unlike magnetometry, which is rubbish on chalk without tertiary geology), I decided to try it out on another chalk site. This time, a neolithic causewayed enclosure. There were sights (and sites) to see on the way. Firstly, climbing the hill itself was illuminating. Below you can see layers of flint within the chalk. To the left and in the middle are two layers of nodular flint, which look broken up. To the right, you can see a more solid looking layer of tabular flint. These layers seem to be diving down into the ground, but as I was climbing the hill, they are actually pretty level.

Flint Layers. Click to Enlarge

Reaching the top of the hill was a Bronze Age barrow, unfortunately looted in antiquity. Again, because there is no height data, it looks like a pit. The layer visible is the underlying chalk. It seems there is not much solid chalk in the mound of the barrow.

Bronze Age Barrow. Click to Enlarge

Finally, I reached the neolithic causewayed enclosure, and was rather disappointed. I couldn't make out much difference in the ditch cuts. It didn't help that flint layers kept getting in the way, pretending to be archaeology and becoming the latest in a long line of 'annoying geology'. I later saw an old picture of a ditch section from Whitehawk Hill, which ASE are digging at the moment, and it looked like the ditch fill was mostly chalk rubble rather than soil, which would explain the lack of difference in EM response and why I couldn't see much.


Another thing I wanted to try out was a bit of Roman road hunting (of course). A lot of the Roman road network has unfortunately been built on, making it difficult examine. Unlike with earth resistance and magnetometry, it is possible to use GPR to examine built-up but flat surfaces, such as a modern road or pavement. I took the opportunity to visit the excavation at Ewell, where Stane Street and associated roadside settlement is being excavated in a small field (church meadow) near the church. The course of the road to the south-west is somewhat unclear, as there is a turn in the road somewhere in the area. With the help of the local archaeologists, we found the Roman road crossing the more modern Church Street. You can see the curve of the agger in the centre of the image below.

Stane Street. Click to Enlarge

I've still got a lot more to learn about GPR. I hope to learn a lot more when I do a week of GPR surveying at Bridge Farm shortly. Watch this space.

12 May 2014

Digging up the Geophysics at Oaklands Park

Readers of my blog will remember the magnetometry and radar at Oaklands Park, Sedlescombe. As well as the final report being published, some excavation has been going on, run by the Independent Historical Research Group. Just for a change, I was the site director! Despite thinking that geophysics is way better than excavation, this site provided a genuinely interesting archaeological question that only excavation could answer, which is whether or not the site was run by the Classis Britannica. It is an industrial scale iron working site, connected by iron-slag metalled roads to other Classis Britannica iron working sites and situated on what would have been a navigable river. The answer was surely yes and this could be proved by finding the special CL:BR stamped tiles that can be found on the sites that they ran. Here is the trench layout.

Trench Map. Click for a larger image

Trench A) This was the main trench. I was hoping the building shaped structure on the geophysics would turn out to be some sort of administrative building with a roof of CL:BR tiles. The walls appeared on both the magnetometry and radar, so I was expecting the foundations would be constructed of iron slag, which they were. Apart from that, things were quite different than expected. The feature in the centre of the building, which I had presumed was a central support was actually a trample layer. A smith had worn a shallow groove in the ground whilst standing at his anvil, leaving a thin layer of iron-slag, charcoal and hammer scale. The inner north wall (there are two) was quite substantial, being composed of layers of iron slag, on top of which a wooden building would have stood. The outer north wall was more slight and was most likely a lean-to against the main building. This only went half way across the trench from the east, so the entrance to the building may have been here. Unfortunately, the floor layer of the building had been ploughed away, but the lack of floor material and the trample layer in the centre of the building suggest that the floor was bare earth. No tiles were found in the area, so the building most likely had a wooden shingle roof. The identification of the building as a smithy raises further questions, such as what they were making. It may have been as simple as hammering the raw blooms into a shape suitable for export. On the geophysics, the middle of the southern wall of the building shows a strong feature jutting slightly into the building. This is most likely the forge.

A slot through the inner north wall in Trench A

Trench B) The original purpose of trench B was to target what seemed to be a pile of moderately magnetic material to the west of the smithy, in case a tile roof had been stripped and dumped to one side before the building was robbed, which the radar suggested had happened on the west side. The results of the excavation were slightly different. At the lowest level, at the level of the natural, some stake holes were found. The remains of a burnt plank was found resting on top of the natural. On top of this, a layer of redeposited clay and earth mixed together was dumped, presumably to level the area out in order to construct a wooden building on top. Unlike the main building, there was no foundation, just a beam slot and some postholes cut into the redeposited clay. On top of this, a destruction or occupation layer filled with charcol  contained a lot of pottery and iron slag. The form of the building is unclear given the small area sampled, but it is more flimsy than the smithy.

A complete pot found at the bottom of the occupation layer in Trench B

Trench C) This trench targeted a massive pit feature, about 35m x 10m, to the north of the smithy. The radar suggested that it was lined with a lense of dense material and it was hoped that some tiles had found their way in. The very deep (1.5m) test pit found pretty much what was expected. The bottom of the pit contained thin layers of burnt clay, slag and charcoal, unfortunately tile free. Above this was a very thick layer of silt, presumably because the pit filled with water, followed by a layer of hillwash with a much greater clay component. The natural here was a very clean clay. As this massive pit was only partially filled with rubbish and lacked any iron stone in the natural (the quarries for the iron stone were at the top of the hill), the quarry was most likely for the extraction of clay to build the bloomeries. To further vex us, we found a modern cable and a land drain cut without a land drain in it.

Trench C: Hello down there!

Trench D) After quickly exhausting the first 3 planned trenches much quicker than anticipated, thoughts turned to what to do next. I toyed with the idea of opening up a huge trench over a large (3m x 7m) bloomery, but decided against it, as that would not have furthered the research question. It was decided to further sample other parts of the site in the hope of finding the CL:BR stamped tiles, so trench D never happened.

Trench E) This trench was in a similar massive pit to trench C, but within the rectangular enclosure to the north-east, near the well. This quarry was 20m x 10m and the test pit found a similar thin layer of material at the bottom, but this time composed entirely of burnt clay from a destoyed bloomery. Below this layer was a thick layer of redeposited clay with little in it apart from pottery at the very top. We never reached the natural. Unlike the quarry at trench C, this quarry did not silt up due to being filled with water but was filled with hill wash.

 Some of the burnt clay layer in Trench E

Trench F) This trench targeted the main slag heap against the northern edge of the field. It was recorded that coins and other occupation material had been found in the slag heap as it was being removed for road building, so it was hoped that some tiles would be found. After about a metre of topsoil that had been ploughed down the hill, the slag bank was reached. It was black with charcoal and contained few finds, with only two pieces of pottery being found in 30cm of excavation. It was decided to close this test pit before the slag heap bottomed out.

 Trench F. Nothing to see here, move along.

Trench G) Targetting a different feature type and a different part of the site, what looked like a rubbish pit on the geophysics, about 2 metres wide, was half sectioned at the western end of the site. It did indeed turn out to be a rubbish pit as expected, with a huge amount of pottery in a black charcoal layer at the bottom. There was a dump of burnt clay on top of this on one side of the pit follwed by a silting up of the feature and some sandstone blocks being dumped near the top. There was quite a bit of post-medieval tile in the plough soil above the feature, but no Roman tile in it unfortunately.

A partially excavated Trench G

We didn't manage to prove the site was Classis run by finding their stamped tiles, but we found out a lot about the site while we were looking. Finding a Roman site with so few tiles of ay sort is quite an oddity in itself. A big thankyou must go to Pestalozzi for letting us dig up their land. Thanks to all the diggers, especially Robin, Brian and Cameron for all their sterling work.

13 April 2014

Latest Results: Plumpton

Late last year, I was involved with a magnetometer survey of Plumpton Roman villa with Chris Butler Archaeology Services on behalf of Plumpton Agricultural College. This means that his team did all the hard work and I processed the data. As the villa foundations were constructed of flint, it doesn't show up, but the enclosure around it does. There are also strong magnetic readings at either end of the enclosure, which may relate to a hypocaust system. There were also signs of field systems associated with the villa to the south, and a possible Roman road to the west, which I was very pleased about. This pattern of a road passing rather than heading to a villa is repeated elsewhere, such as Barcombe.

Plumpton Villa magnetometry

At the time time, David Millum of the Culver Archaeology Project led a team surveying the villa itself with an earth resistance meter, which showed up the walls very nicely. It seems to be a standard winged villa, but with the western wing extending further to the north and west than it should. The villa as a whole looks like it has at least two phases. At either end, where we get the strong magnetic responses on the magnetometry, we also get strong high resistance areas, showing more than just the walls. Maybe it is just rubble, maybe it is something to do with a hypocaust.

Plumpton Villa earth resistance

This weekend, I taught a geophysics dayschool here as part of the Sussex School of Archaeology. As well as covering the usual earth resistance and magnetometry, we also did some resistance tomography and a single line of radar, both along the green line in the images below. I have layed them out horizontally at the correct position next to the green line that they were surveyed along, so just imagine that they are vertical slices through the ground along that line.

First the tomography. Several of the walls show as red points towards the top, plus at either end of the villa, we get broader and deeper high resistance confirming the standard earth resistance. There is also a hint of something heading further down under the ground at either end.

Plumpton Villa resistance tomography

Finally, the single line of radar we did. The ground was a bit wet, so the top layer is a bit of a mess, but many of the walls are still visible below that. The two walls visible in the centre of the villa are at the expected depth, but those to the west and east are deeper, further suggesting a hypocaust system.

Plumpton Villa GPR traverse

The Sussex School of Archaeology will be running a training excavation on the villa in the summer of 2014.







09 March 2014

Version 1.13 of Snuffler released

It's time for a new version of Snuffler, and this version is all about importing stuff. There are two new files imports for getting data into Snuffler.

The first is for importing TerraSurveyor (was ArchaeoSurveyor) grids and composites. TerraSurveyor is a (very good) commercial equivalent to Snuffler. Someone sent me some data in that format, so I took the opportunity to write an import for it.

The second is Surfer ASCII grid files, not to be confused with the ASCII Grid files that ESRI produces. Surfer is software for displaying 3D geographical data. It wasn't data from that software specifically that I was trying to import though. ReflexW, the GPR software I use, has an export function for time-slices in Surfer ASCII grid format and I wanted to play around with slice data in Snuffler. Here is a time slice from the site I covered in my last post display in Snuffler.


You can download the new version at the usual place.

06 March 2014

Geophysics on WWI Camp in Eastbourne

World War I is very much in the news at the moment, due to the centenary, so here is some WWI geophysics to entertain you. The site is the extreme south-east corner of Summerdown Convalescent Camp in Eastbourne, which is very near where I live. I got to use my new GPR system, which is always good, and it showed up the archaeology quite nicely. In the area surveyed, there are ablution and barrack blocks. The former have fairly solid foundation structures while the latter were built on a series of concrete piles. See if you can spot the in the image below. I would compare this to what was excavated... but no-one told me when the excavation happened. You can see the full report here and watch a movie of the time-slices here.


15 January 2014

News and more talking

Hello everyone. It's not a time for geophysics, being winter, so it is writing and talking time. For those of you who wondered where my website for Snuffler disappeared to, the company that did my web hosting fell off the edge of the world, so I had to spend some time regaining control of my domain name and sorting out new web hosting. All is well now.

I've mostly been spending my time writing a chapter on Roman roads for a book entitled Archaeology and Land-use of South-East England to 1066, which is being published as a tribute to Peter Drewett, who died last year. Having only knowledge of the Sussex area, I have been furiously researching what has been going on in Kent and Surrey since Margary's day. Thanks to a number of people from those counties, my task has been made somewhat easier. The book should hopefully appear in 2015.

My first talk of the year will be at the Lewes Archaeological Group, where I will be talking about Roman roads in the Lewes area, where I seem to be doing a lot of my geophysics surveys. The talk is at Lewes town hall on the 31st of January 2014 at 7:30pm.

Readers of my blog will remember me talking about Oaklands Park in Sedlescombe here and here. Well now the final report is online. If you don't fancy reading a long geophysics report, you can hear me talk about it on the 17th of February 2014 at 7:30pm. The venue is Pestalozzi, on whose land the majority of the work was done.

Finally, I will be running a series of dayschools at the Sussex School of Archaeology. It is a two day course. The first day will be out in the field, learning about earth resistance, resistance tomography, magnetometry and ground penetrating radar while the second day will be indoors, covering theory, data processing, interpretation and the effects of geology on geophysics. As there will be limited space on the first day, I have split that into two groups on the 5th and 12th of April 2014, with the two groups coming together for the day in the classroom on the 13th of April 2014.

Then, I will be all talked out and will return to the field for more geophysics.

21 October 2013

Version 1.12 of Snuffler Released

Another small update for Snuffler today, which will hopefully make the destripe filter easier to use.

A feature of the destripe filter that is useful, but many don't understand is that when reading through each line of data, it will restrict any readings it comes across that are outside the display bounds to the level of the display bounds, so if there is a reading of +60nT and the display bounds are +/- 2nT, that reading will be treated as +2nT. The benefit of this is that particularly strong features will not affect the filter too much, which would stop stripes being left visible even after filtering. The downside is that if your striping is particularly severe, you would need multiple passes of the destripe filter to get rid of the stripes, as it would only remove a maximum of 2nT (the display boundary) each time. This downside became most obvious a few versions back when I changed the default display boundary for mag data to +/-2nT rather than calculating a range from the readings in the file, which is more suitable for resistance data. The fact that you needed to do multiple destripes was not obvious to all of my users, and to be honest, it shouldn't need to be run multiple times, so I created a new destripe type, which is now the default, which will do multiple passes of the filter, so you don't have to. You still have the option to do a single pass if so desired.

The other noticeable change is that help file should hopefully work for more people now. I had several complaints that it wasn't, but couldn't replicate the problem on my machine. Having found a machine where it doesn't work, there should now be a workaround which will make it work for some people at least. If it still does not, let me know.


Changing the subject, if you fancy seeing me talk about geophysics and Roman roads, I will be at the CBA SE conference in Faversham on the 16th of November.