Anyone new to this blog or geophysics in archaeology is recommended to read the material on the “Geophysical survey in archaeology” page.
As a result of the hot, dry weather (sound familiar?) last year, the Chess Valley Archaeological and Historical Society noticed some parch marks in the fields along the River Chess near Little Missenden. They undertook some Earth Resistance survey in those fields. Earlier this year they asked if CAGG would be willing to come and undertake further surveys at the site. At the end of May we undertook three days of survey completing an area of mag survey, two blocks of GPR survey, and extending the Earth Resistance survey undertaken by CVAHS. Peter Alley also undertook a topographical survey using his UAV.
To take the topography first, we have two sources of information: the LiDAR data which can be downloaded from the data.gov.uk website and the UAV aerial photogrammetry data collected by Peter. Each method has its strengths and weaknesses. I have posted the LiDAR image created by John Glover and Peter’s imagery below.
Figure 1: LiDAR image. The field we were working in is to the west of the buildings in the centre of the plot. Image courtesy of John Glover.
Figure 2: Geo-rectified composite image of the area. Image by Peter Alley.
Figure 3: Elevation map created using Structure from Motion (photogrammetry). Image courtesy of Peter Alley.
Figure 4: hillshade representation of the photogrammetry data. Image courtesy of Peter Alley.
From these various images we can see that the site consists of a flat area either side of the river (essentially the flood plain) and then a steeper slope up to the houses and the main road. The flat area is alluvium, the slopes are “New Pit Chalk”. Although there are some undulations, most clearly shown in Figure 4, nothing much suggests archaeology. Some of the features, especially in the field to the east, are probably alluvial or periglacial.
The mag, working as quickly as ever, covered two blocks either side of the river (Figures 5 and 6).
Figure 5: Yvonne Edwards and Rhian Morgan running the magnetometer. Image © Mike Smith.
Figure 6: the magnetometry results.
There are a few features showing in the mag plot, especially the modern ferrous features to the south of the river. I have put arrows on the plot to make the discussion easier (Figure 7).
Figure 7: the mag results with arrows.
South of the river is a long line of strong magnetic features (red arrows). I’m not entirely sure what this was, perhaps an old fence line? Maybe an obsolete pipe which has been partially removed? Whatever they are, they have largely obliterated any other magnetic features in that area, as well as causing problems with the destriping routine!
North of the river there are two very strong features marked with dark blue arrows. These are easy to interpret (see Figure 8)!
Figure 8: the explanation for the strong magnetic features indicated with the dark blue arrows.
The only interesting thing about these is that the services connected to them do not show on the mag data at all. The most enigmatic of the features is the linear one indicated with the light blue arrows. This feature could be seen in the parch marks from 2018. To the west, the straight line looks like it might be a service of some sort, but then it curves up towards the pond, and then continues towards the east. I suspect that the linear feature indicated by the green arrows might be part of the same thing. It is very curious, and I am very unsure what it might be. The low and high magnetic readings keep swapping sides, which suggests it isn’t just the result of soil with a high magnetic susceptibility filling a linear feature. Sadly, I think the odds are on it being a service again, but it needs to be tested. The pond, by the way, is a relatively new feature in the landscape (Figure 9).
Figure 9: the pond.
The linear feature indicated by the white arrows in Figure 7 is the build-up of topsoil at the change of slope (cf. Fig. 3). The last remaining feature is a square of magnetic noise shown by the yellow arrow. I suspect that this might be the location of something like a wooden building, or parhaps a flat platform for something.
As well as the survey which had already been undertaken by CVAHS, we did a small block of Earth Resistance survey on the lower slopes. Thanks to Ruth Halliwell and Peter Alley for helping me with the survey (Figure 10).
Figure 10: the Earth Resistance survey underway. Image © Mike Smith.
As usual, we used the pole-pole configuration for the cables (i.e., the remote probes were at least 30m away and at least 20m apart), and took one 1m mobile probe separation reading and two 0.5m mobile probe separation readings at each survey point. On this occasion, the 1m readings didn’t add much to the results so I will just discuss the 0.5m survey. Figure 11 shows the results. Very unusually, the readings went from very low (6.1 Ω) to very high (250 Ω). The high readings were so high that I had to change the settings on the res meter, and made my ever patient helpers re-do a couple of grid squares.
Figure 11: the earth Resistance survey results.
A further problem can be seen. The long tail of high readings which can be seen in the inset graph in Figure 11 means that most of the survey is shown as very light low resistance readings and it is hard to see much of a pattern. If we simply clip the values (i.e., make everything above a certain value black), the high resistance areas become one nasty blob. The statistician in me is used to seeing skewed data sets like this in all sorts of cases. The answer is some form of data transformation, and TerraSurveyor provides this via its “Compress” function. This can apply a log-transform to pull in that tail and make the data look more normal (in the statistical sense of more normal). Figure 12 is the result.
Figure 12: the Earth Resistance results subjected to a log-transform.
In Figure 12 the transformed data values are shown as the inset graph once more. Hopefully, you’ll agree it allows us to see patterns in both the high and low values. There does seem to be a linear feature running across the plot. I suspect, however, that this is, again, part of the geology. I doubt that anything we can see in this plot is archaeology. The high readings might be a deposit of flints. The blobby mag results in the same area suggest depressions in the surface of the subsoil which have filled with slightly more magnetic topsoil.
The final technique we used was GPR. We completed two blocks, one to the east of the area, north of the river and the second to the west, again north of the river (Fig. 13).
Figure 13: the GPR being operated by John Glover of CVAHS. Image © Mike Smith.
Figure 14 shows the time-slices from the first block. Just to confuse you all, north is to the bottom of these plots. They have been created with a 40% overlap between the slices. Ignore the depths, they are incorrect.
Figure 14: time-slices from the first block of the GPR data. North to the bottom.
Slice 1 is simply the impact of the surface layers on the GPR, probably variations in vegetation and water retention. Often we can see things like mushroom rings in this slice. Slices 3 to, probably, about slice 8 are the actual deposits under the surface. Slices beyond that are probably “echoes” of the upper slices boosted by the software into visible images. We seem to have three things happening. Some linear features which are undulating and strong reflections shown in red. A broad curving low reflection area shown in dark blue, and a linear feature shown in cyan. Figure 15 shows these on the Google Earth image.
Figure 15: time slice 6 from the GPR survey of Block 1.
As can be seen from the figure, the dark blue swathe comes out from the current course of the river and returns to it. I suspect we have picked-up an earlier meander of the river. Similarly, the lines of red might be banks of river gravels from previous meanders. In Figure 16 I have indicated the outline of the noisy area of mag data shown by the yellow arrow in Figure 7.
Figure 16: the GPR slice with the magnetic feature from Figure 7 indicated by the white line.
As can be seen from Figure 16, the southern edge of the magnetic feature aligns nicely with the northern edge of the dark blue “channel”. Perhaps we have a building platform on the banks of the river?
The linear feature in light blue could well be another service trench.
The second GPR block was to the east, again on the northern bank. It was sited to cover one of the parch marks seen in the previous year. Figure 17 shows the first 12 time-slices. Again, north is to the bottom and ignore the depths.
Figure 17: First twelve time-slices from the second GPR block. North is to the bottom, depths are incorrect.
The most interesting thing about this block are the linear features which show clearly from about slice 7 onwards. We have two clear parallel lines running alongside the river, and one which runs at a right angle to the southern (upper) line into the river bank. Figure 18 shows the eighth time-slice on Google Earth.
Figure 18: time-slice 8 from the second GPR block.
As can be seen, the northern linear feature follows the one seen in the mag data. This feature is, therefore, something that (a) is more magnetic than the background; (b) a good radar reflector and (c) in places remains green when other areas are parched (see Figure 19).
Figure 19: photo taken using a UAV during the drought in 2018. Courtesy of CVAHS.
Comparing Figures 18 and 19, we can see the southernmost linear feature shows well as a green mark in the field and as a strong reflector. This feature, however, does not show in the magnetic data. We can look at the radargram and the time slice at the same time using GPR Slice (Figure 20).
Figure 20: a radargram and time slice seen in pseudo-3D.
Looking at the radargram we can see that the broad reflection layer at the top is continuous over the northern linear feature, but is interrupted over the southern one. This suggests to me that the southern feature might be relatively recent.
So what have we discovered? Largely, I feel, most of the features are to do with geology and the river. The two main mystery features for me at the long linear feature and the shorter one with right-angles. Both show clearly on the aerial and the GPR. The longer linear one also shows on the mag. I’m not sure what either of these are, and only “ground truthing” might give us an answer. My instinct, however, is that neither are all that old, but this is only an instinct.
As always, many thanks to everyone who helped both during the survey and with moving the gear. Also, many thanks to the Institute of Archaeology, UCL and SEAHA for the loan of the equipment.
For those awaiting the first results from Verulamium 2019, they will be posted in the next day or so.