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.

01 September 2013

Latest Results: Barcombe (and Roman agricultural planning)

Now the crop is coming of at Barcombe, I have returned for some more magnetometer survey at Bridge Farm, extending the settlement eastwards. The process was speeded up with the aid of a second Bartington, which was on loan to the Culver Project from AOC Archaeology following the recent excavations. Members of the Culver Project and Ringmer Group aided with the survey, enabling the area to be finished in a day. Here are the results.

Latest results. Click for a larger image.

The roadside settlement continued east, with strong evidence of settlement close to the road. The extent of the settlement either side of the road is less than to the west, but it is still substantial. The strong pit features cluster within 50 metres of the road, with a very strong outlier to the north-west, which may be some sort of kiln. One puzzling element are the two linear features heading north and south, at a different alignment to the road. They are parallel to eachother, but their alignment is strange. Eventually, I worked out that they are exactly parallel to the Barcombe to London Roman road, which is further to the west, as seen in the image below.

Click for larger image.

What I had been expecting is more of the carefully laid out ditches near the centre left of the image, where a square block of land had been marked out with side of 4 actus. An actus is a unit of Roman measurement used for setting out land in an ordered way, set up by Roman surveyors in a repeating pattern of squares known as a Centuriate. I was hoping to see this continue to the east, but my theory was wrong. Instead we found these linear features extending out at the same angle as a Roman road about 300m to the west. Was there some link between the two?

Click for larger image.

Rather fortuitously, a further example of exactly this had been found recently by members of the Culver project using their borrowed magnetometer. They had been looking at more of the north-south Roman road that the project had found to the west of the river. These fields have been more heavily ploughed, so the results are not as clear. As well as the road, they have found a series of ditches parallel to the road. One of the measurements in between ditches is 2 actus, but the further ditch is 5 metres above the 2 actus mark.

Is this an example of Roman planned cultivation? The results in the image above seem sparse. Either the more shallow ditches have been ploughed away, or only part of the system was surveyed and never used. Hopefully this is something further geophysics will answer.

29 August 2013

Latest Results: Oaklands Park

I did buy the Groundvue 3A ground penetrating radar cart that I discussed in my earlier post, and have been giving it a go at Oaklands Park, where I did a large magnetometer survey with IHRG a while back. There were a few areas where we thought that there might be a possibility of buildings. The results have been very positive. The two survey areas we covered, outlined in red in the image below, are 40x40m and overlap as we found that the building was on the edge of the first grid we did. The first area, to the north-east, was recorded down to 70ns. The second area, to the south-west, was recorded down to 50ns.

GPR survey areas. Click to enlarge.

There are interesting features at various levels, and to get a full idea of them all, you need to look at different depths. Below are features from 10ns and 13ns followed by an interpretation. To get a better picture of what is going on, you can look at the whole lot on youtube. The video for the first survey is here and for the second survey is here.

Features at 10ns. Click to enlarge.

Features at 13ns. Click to enlarge.

Interpretation over magnetometry. Click to enlarge.

You can see in the above image the location of features on the GPR surveys overlaid on the magnetometry results. The most important feature is the building in dark blue. This shows up fully on the magnetometry, but the west wall and a second north wall are missing on the GPR results. This may be due to robbing. The green feature seem to be metalling on a trackway, most likely iron slag. It doesn't extend all of the way along the track, but seems to have been placed outside of the building, which may indicate that this section of the track was busy and in need of repair. The two light blue features are very strong both magnetically and very dense on the GPR results. They may be where the iron was processed. The yellow lines are boundary features, which mark out an enclosure around the building and a further enclosure to the north-east, the other side of the track. The orange areas are large pits. On the GPR results, these can be seen to reduce in size as they go deeper. Light purple areas seem to be dumps of material with no pit. Finally, the small dark purple feature to the north-east seems to be a well. It is circular, with a small L shaped feature attached at higher levels, which disappears further down. While all other features disappear by 30ns, the well keeps going all the way down to the 70ns recording limit.

These GPR results are great. I'm looking forward to using it on a lot more sites in the future.

19 August 2013

The FM256 Problem

A while back, I attempted to write an import for the Geoscan FM256 in Snuffler based on the documentation alone, as I didn't have a machine to test with. Unfortunately, I had a number of my users reporting that the import didn't work. Despite extensive testing with a number of them, for which I am very grateful for their time, I couldn't resolve the problem. An opportunity arose recently to borrow a machine from the Surrey Archaeological Society (many thanks to them), so I was able to do some testing with a physical machine. I tried the download myself and all appeared fine. I couldn't work out what the problem was. I then realised that I do something slightly different to most people. I use a PCI serial port card rather than a USB to serial converter. I tried a couple of these converters and the results came out as garbage. It wasn't just Snuffler either, it looked like junk in Hyperterminal. What I can't understand is how people are using Geoplot and getting it to work with this data. It is a mystery which I have little inclination to solve. The upshot of all this is that if you want to download from an FM256, get rid of the USB to serial converter and get a PCI serial card. That would be no use to you if you have a laptop of course. You would either need to buy a tower computer, or better yet, a Bartington.

14 August 2013

Digging up the Geophysics at Barcombe

Despite what I said in my last post, I always enjoy it when somebody else digs up a site based on my geophysics. I get to learn so much by comparing the geophys results to what is excavated, yet someone else has to deal with running the thing and paying for it all. This happened at Barcombe this year, on the site of the fortified Roman settlement I had previously surveyed. I even took a week off work to have a dig for a change, which is not something the local archaeologists are used to seeing.

The Culver Project was of course running the dig, with the help of AOC Archaeology in a giant community dig funded by the Heritage Lottery Fund. The excavation was carried out in a pasture at the southern end of the site. The Culver guys also did an earth resistance survey to the west of the excavations, which shows the trackway and the old river bank nicely. Here is an annotated zoomed in image (click to enlarge) with descriptions of the various features described below.

A) This is a very strong feature on the survey, and it was expected to be industrial. The feature didn't disappoint, as it turned out to be a large circular pit with strong in-situ burning. There was no pottery wasters, but there was some tile wasters, suggesting that it might have been used as a tile kiln. This is not certain though, as Roman tile kilns are usually rectangular.

B) This curving ditch feature surrounding feature A contained a lot of tile wasters, reinforcing the idea that it was a tile kiln.

C) This small feature on the geophysics turned out to be very interesting indeed. It was a rectangular pit lined with tiles. A large blob of concrete was in the bottom, seemingly put in wet, as it had stuck to some of the tiles at the bottom. Nobody knows what this feature was used for.

D) This is one of the many tracks going through the site. The side ditches were quite substantial, being nearly as large as you would expect on a normal Roman road. The surface, where it showed, was river gravel, and had cart ruts running through it. Being a Roman road nerd, this feature is the one that got me most excited.

E) Half of a large pit was excavated here, containing occupation debris. These sort of features are common across the site, especially in the centre of the settlement, where there hardly seems to be a spot free of them.

F) This trench was put in to examine the relationship between the double ditched enclosure and the trackway as they crossed, and did not disappoint. The later enclosure clearly cut the earlier trackway. An almost complete quern stone was found towards the eastern end.

G) This trench was put in to see how the archaeology was affected by ploughing in the arable field to the north. Fortunately, the archaeology seems to be mostly below plough depth. The western ditch of the trackway was found as expected.

I) The eastern ditch of this trackway could not be found. Instead, the two features shown on the geophysics turned out to be two large pits. The southern one was square with vertical sides.

J) The enclosure was examined in greater detail here, by means of a hole cut through the hedge. The layers were not as clear as at F, with three ditches seemingly crossing the trench. The ditches were very deep, heading well down into the water table.

K) The ditch crossing here was again found to be cut by the main enclosure ditch. An intact (before being clipped by the bucket) cremation burial was found here and block lifted to be excavated elsewhere.

If you want to know more details about what happened, you can visit the excavation diary.

16 July 2013

Landscape Geophysics

It is the time of year when all of the fields are full of that annoying stuff called crops, so my options for survey are limited. I've been spending the time writing my book and being sociable for a change. Here is a small extract, musings on the subject of Landscape Geophysics.

Landscape archaeology is the study of not just one site, but the way in which peoples used and modified the environment around them. It uses a multitude of techniques, such as remote sensing, geophysics, environmental archaeology, excavation and bioarchaeology. This gives a broad picture of land use rather than focusing on one particular site. The usual course of events in archaeology usually involves three steps.

1) Exploration. Finding a site by methods such as remote sensing, metal detecting or looking for pottery scatters.
2) Survey. Study of the site by fieldwalking, geophysics and earthwork survey.
3) Excavation. Answering specific questions about the site by targeting excavations to answer them.

The problem I have found is with the third part. It takes an awful lot of time and effort to excavate and publish a site, even if the excavated areas are quite small. There are pros and cons to excavation. A site that would take a couple of weeks to geophys might take about 200 years to fully excavate if the same area was covered (not that you would want to). Of course, excavation can provide a huge amount of information about dating, phasing and the use of the site, whereas geophysics will only give you a basic layout. The pay-off comes when you survey a site and move on to the next. Rather than digging that site for the next 20 years, you could survey hundreds of sites in that time, building up basic information on a huge area. This enables the study of a landscape that the time restrictions of excavation would not allow, and this is the method I have chosen to study the Roman road network.  

Roman roads and their associated settlements show up well using geophysics. It is a landscape that is being ploughed away. It will not be there forever. By studying as wide an area as possible, not only will the limited information that geophysics provides be available long after the site is destroyed, but it offers the chance to save such sites using schemes such as Higher Level Stewardship. The study of the road network itself gives a better understanding of what the Romans were trying to achieve in their occupation, giving the basic layout of their entire infrastructure. It is archaeology on a large scale. Where you lose the finer detail, you gain a much broader picture of what is going on. That is what I like to call Landscape Geophysics.

I am not for a moment suggesting that excavation is bad. I am saying that the two methods are complementary, and a greater balance between the two would benefit archaeology as a whole.

22 May 2013

Version 1.11 of Snuffler Released

It's only a small release time, with a couple of new features.

If you happen to have a Geoscan multiplexer, MPX15 or RM85, you can now import from 2 to 4 parallel readings. There are limitations. The probe spacing is assumed to be 0.5 metres and the size of the grid is limited by the number of readings taken at one time, so for instance, you couldn't have a 20 metre grid with 3 parallel readings, as that just wouldn't fit. As for taking readings at different depths at the same time, not in this version of the software. Perhaps in a later one.

Snuffler will now also import files exported by the EPE magnetometer download software. These are DIY devices constructed from schematics in an old issue of the EPE magazine. They lack any means of balancing, and don't have a fixed number of readings per line, so what Snuffler will do with these is allow you to set the size of the grid, chop off any readings in lines greater than that size and insert dummy readings for lines less than that size. It's not ideal, but then neither is the hardware unfortunately.

I had hoped to get the import for the Geoscan FM256 working for this version, but I have given up for now. It has been made more difficult because I don't have access to one of these machines for testing. I have been doing some testing with a number of my users to try and sort this out, but to no avail. I thank them all for the time they have given to help me. I think there is some sort of issue with the comms handshaking, but it is not entirely clear, so until I can physically get my hands on one of these, the download option for this machine will not work I'm afraid.

You can download the new version at the usual place.

18 May 2013

Trying Out Ground Penetrating Radar

I've been wanting to get into Ground Penetrating Radar for some time now, but they are expensive beasts. Rather than buy a loft conversion for my lovely and long suffering wife, Merryn, I have been saving up for one of these, and now have enough to make a purchase. Aren't I mean.

With so many choices, which one to buy? It doesn't help that there is little comparative information out there, making it difficult to choose between them. The mostly commonly used GPR cart systems in archaeology (as far as I can tell) are the GSSI SIR-3000, the Mala X3M and the S&S Noggin. Fortunately, a friend of mine (Hi Dave) had used all of these and had some words of wisdom to impart. Apparently there were two areas where these systems varied significantly. Firstly, there are variations in the way that data is stored. Whilst the Noggin will store raw data, the other two systems will record with the gain applied when surveying, which causes problems with processing the data later on. Secondly, there are variations with the way that the systems handle GPS data, with the GSSI having problems with it, the Mala performing somewhat better and the S&S Noggin dealing with GPS in the best way. So the Noggin is the clear winner right? Well apparently it is also the most expensive, and there is no local support here in the UK, so if something goes wrong, you have to send it overseas.

It was then that I met Erica Utsi at the NSGG conference at the end of last year. Utsi Electronics produce a cart system based on the Groundvue 3, which apparently stores data raw like the Noggin does, but doesn't cost a fortune and is produced here in the UK! I decided to try hire one for a day to see how I got on with it, hence this monster of a blog post. I must stress at this point that I only know enough about GPRs to be dangerous, so everything I say should be taken with a pinch of salt. I had two test sites on which I wanted to try out the machine, sites which I already knew a lot about, so I knew what I was expecting to find. Here is how I got on.

The first site is part of the Culver Project, one of my favourite sites in Sussex. The site is on clay, which is one of the geologies that radar has trouble with, especially when it is wet. Whilst the ground at the time of the survey wasn't sodden, it was far from dry, so this was a real stress test for the machine. The other thing that I wanted to test, which this site allowed, was how the radar performed with different surfaces, in this case tarmac, grass (shaggy but not long) and a ploughed field (flat). As you can see from the image below, there is a Roman road crossing the site, which is know to consist of a single layer of flint. The side ditches can be seen on the magnetometer survey to the north, and the road surface can be seen in the earth resistance survey to the south. The road is known to be in good condition to the south, but there was a crop in the field, so only a single traverse was done at the edge of the field. A further survey area, 40 metres wide, covered the road and the grass verge to the south of it.

The Roman road we are investigating

Despite the harsh conditions for the GPR, the road was certainly visible in the results, as you can see below. This is the best vertical traverse of the group, and you can see the road surface in the middle at the bottom of the image.
40 metre traverse along the tarmac road and across the Roman road

Creating a horizontal time slice shows the road surface heading diagonally across the image, but not the side ditches.

Time slice showing the road surface

Whilst the road was certainly visible in the results, there were issues. The road did not show very well under the grass verge, with a lot of messy readings in the results. The single traverse in the field to the south did not show the road at all. On the grass, ground coupling may have been an issue because of the shaggy grass, but that wasn't a problem in the field to the south. It may be that the road has protected the area below from the worst of the damp weather, leaving the signal in the grassed and field areas subject to higher levels of attenuation from the water content. It may be that GPR in clay fields such as this is best left to just after harvest time, leaving a rather small window of opportunity for survey.

Whilst we were at Culver Farm, I took the opportunity to test out the encoder wheel on the cart, to see how accurately it performed. One of the things I am hoping to do with the GPR is to do isolated long traverses across fields to locate Roman roads crossing through, without having to spend time doing a full survey. We laid out a 100 metre tape and recorded a traverse. On download, the traverse was found to be 102 metres, which is not disastrous, but not ideal. It may just need recalibrating, but traverses can be stretched to the right length, so no great issue.

The second site is on chalk, a geology that GPR is more at home with, looking for walls, which GPR is good at looking for, so this site is nowhere near as harsh as the last one. The site is a deserted village on the downs at Exceat. There was a church here, but other than the lumps and bumps on the surface, the only sign that anything was here is a large stone in the middle of where the church stood, which was erected after excavations in the early 20th century. Greg Chuter had organised an earth resistance survey, to try to find out what was around the church, and the results clearly showed the church and the churchyard wall. Part of the site, outlined in red in the image below, was then surveyed with the GPR.

Survey area shown over the earth resistance survey

As you can see from the three time slices below, the GPR performed very well here, highlighting features much clearer than the res survey, with some interesting phasing information as well. The expected graves in the churchyard did not show up, but it may be that I did not record deep enough. I don't think I hit the chalk geology with the depth I had the radar set to. I am very happy with the results from this site.

Time slice at 9ns

Time slice at 15ns

Time slice at 18ns

Finally, here is a youtube video of the slices, from top to bottom, and a picture of the intrepid John Kane using the machine. Thanks John.

Intrepid John Kane with Groundvue 3 cart

So those are the test sites. What about the machine itself? There weren't any real problems in the field. The data logger is simple to use and does the job, though being based on MS-DOS, it isn't particularly pretty. Downloading the data to a computer was a bit of a pain though. Downloading is via a network cable, which means unplugging yourself from whatever network you are plugged into to plug in the datalogger. You then need to change some setting on your computer to make the connection work. Unfortunately for me, I run Windows XP in user mode, for security reasons, and I wasn't able to change those settings, so I had to log in as admin, change the permissions on my account before changing the network settings. After that, the download is simply a matter of cutting and pasting the files as if from a network drive. Adding another network card to my computer would save the hassle of changing all that each time, but it would have been better to have the data on a removable memory card, or have a USB connection to the data logger that would act as a USB mass storage device.

There are two software options for processing the data from a Groundvue, ReflexW and GPR Slice. The latter is nice, but hideously expensive, so I borrowed a copy (version 6) of the former, which is somewhat cheaper, to try it out. ReflexW is certainly fully featured. There are an extraordinary number of things you can do with it, most of which you will never need to use. There wasn't anything that I found missing. The user interface is somewhat clunky though, as a lot of scientific software can be. It also didn't help that English is not the first language of the person who wrote it, so the manual can be a bit difficult to follow at times.

So in summary, though I haven't tried out other systems to compare, I was certainly happy with how the day went, and I will be buying a Groundvue 3 for myself.

31 March 2013

Surveying for Archaeological Geophysicists

Something that archaeological geophysicists have to grapple with constantly is location. It's no good having a pretty geophysics image if you can't put it on a map or tell someone where to dig.   Not only do you have to set out your survey grids, but you have to record where they are. The whole process can be time consuming, taking away from your actual geophysics survey time, so it is best kept to a minimum if at all possible.


Tapes are where it all started. Old hands at geophysics will remember marking out 20 metre grids off of a baseline using two tapes at 20m and 28.28m (yes, Pythagoras was good for something after all). Anyone doing this over any great distance will notice that their lines of grid markers start to curve off into the distance. It's even worse if you are surveying on a hilly area. Even then, how do you get your survey onto a map? Certain points in your gridded area can be measured to known markers in the landscape, at least two per point, so that the grid can be re-established. All in all, it's far from ideal, but fortunately, technology is here to save the day.

Total Stations

Total stations are the next step up. Like tapes, they have no idea where they are in the world besides what you tell them, so you will be using arbitrary grids. They do have the advantage of being quicker and more accurate over long distances. Grids can be set out by setting an arbitrary position and direction for the total station, then communication with the person holding the prism pole until they are in the correct position for the next grid marker. This part is quicker if you have a robotic total station, which can be used by a single person, but is more expensive. A previous grid can be re-established with resection points. These are two or more points in the landscape which you can describe with great accuracy, e.g. "NE corner of W gatepost of gate in SW corner of field". Alternatively, survey pegs can be placed somewhere where they are unlikely to be removed to act as resection point. When returning to re-establish the grid, you can then place your total station wherever you want, then measure your resection points, telling the total station their grid locations, as previously surveyed, and the total station will calculate where it is within your original survey area. If you pick the right resection points, you can re-establish your grid with an accuracy of 1cm. It is best to keep these points a long way away from each other, to reduce rotational error. As for getting your geophysics on a map using a total station, I have discussed this in a previous post.

When looking to buy a total station, there are a number of things to look for. The first is the accuracy. Each total station will be rated as accurate (for rotational angle) to a number of arc seconds, so 1" is better than 3" is better than 5". In addition to this, some basic models will only deal with X and Y coordinates, so if you are interested in height information (Z), you need to check for this. Robotic total stations (single user) will be more expensive than a normal total station (two user).


Now to the real reason for this post. For some time, I have been looking at buying GNSS, to replace the total station I use currently, but there is a lot more to consider when buying one. The main advantage to using GNSS rather than a total station is that you will have absolute coordinates, rather than using an arbitrary grid. The subject can be a bit of a technical minefield, so I wanted to share something of what I have found so far. The field is changing rapidly, so no doubt this information will be out of date in quite a short amount of time. I will assume a basic understanding of how GPS works, to avoid this guide becoming too verbose.

GNSS (Global Navigation Satellite System) is the modern term for what most people understand as GPS. GPS is actually just one of four constellations of satellites, GPS (US), GLONASS (Russia), Galileo (EU) and COMPASS (China). The first two systems are mature, with a full compliment of satellites. The other two only have a few satellites in place, and can't reliably be used for positioning yet. These separate systems will give you different locations, your receiver may not use all of the constellations at once.

Many people will be familiar with the GPS functions in devices such as smartphones. These are consumer level devices, accurate to a few metres, which is far too inaccurate for our purposes. Ideally, we want 1cm accuracy. For that, we need survey grade equipment. To understand the benefits of survey grade equipment, we need to know a bit about the signals that the satellites use.

Each signal is composed of the code phase and the carrier phase. The code phase is a string of data, a pseudo random number (PRN) that allows the receiver to work out roughly where a particular satellite is. Because this string of data is quite long. you can only use this to work out where an individual satellite is in relation to you to within about 20m. Having a large number of satellites at your disposal will allow you to improve the accuracy to an extent, but only so far. Consumer grade devices only use the code phase, which is one of the reasons they have poor accuracy. The carrier phase is the carrier wave used to carry the data in the code phase. The wavelength is much shorter than the time it takes to transmit one PRN in the code phase, which makes it possible to use this for greater accuracy. Receivers that use the carrier phase are RTK (Real Time Kinematic), which at the moment, is only survey grade equipment.

Each constellation many have several signals, consumer grade equipment will currently only receive one of these. The more signals that your receiver can pick up, the more accurate it will be.

GPS : L1 (Consumer), L2 (code phase encrypted, carrier only), L2C (not yet active), L5 (not yet active)
GLONASS : L1 (Some consumer devices), L2
Galileo : E1, E5a, E5b
COMPASS : B1, B2, B3

To pick up the signals from satellites, your receiver will need an antenna. Consumer grade equipment tends to have quite poor patch antennas compared to the survey grade equipment. Also on survey grade equipment, the antenna is mounted above your head, so you aren't getting in the way of a weak signal, and will have a base plate, to stop multipath from the ground. Multipath is one form of inaccuracy that a receiver will need to deal with. A signal can bounce off of something before hitting the antenna, making the receiver think that the satellite is further away. As well as physically stopping multipath from the ground, survey grade equipment can reject multipath from other sources such as buildings using statistical processing.

Another source of error is distortions in the ionosphere and troposphere altering the speed of signals on the way down. The solution to this is corrections, which come in different forms. SBAS (Space Based Augmentation System) is the most basic, and is often available on consumer grade equipment. These are regional services, such as WAAS (US) and EGNOS (EU). The way it works is that base stations, which have known locations, will pick up the signals from satellites, and work out how the ionosphere and troposphere are distorting that signal. They will then broadcast those corrections to a satellite, which then broadcasts to your receiver, which applies those corrections to make your location more accurate. Unfortunately, the base stations are very widely spaced, so the corrections will not give us the 1cm accuracy we desire.

Traditionally, the way that survey grade equipment deals with the problem of corrections is to have a separate base station of its own set up nearby, on a known point. Corrections will then be sent by radio to the surveyor (rover), giving the desired 1cm accuracy. The longer the distance between the base station and the rover, the less accurate the corrections will be. The downside to this is having to find a known point, or waiting a long time to let the base station properly work out where it is. Also, if the base station is any distance away, you will need a second person to look after it, in case it gets stolen. The current way to deal with this problem is to have a closely spaced network of ground stations, which create a decent model of atmospheric distortions. The corrections are sent over the internet rather than via a satellite, with the receiver having a mobile network SIM card installed to pick them up. This is known as 'Network RTK'. Unlike SBAS, these corrections are not free, costing around £1000 a year in the UK, which has various network RTK services, all based on OS Net.

So... did you really need to know all that? Well yes, because if you are spending a large amount of money on a new piece of kit, you better understand exactly what it is doing, otherwise you might end up with a lemon that only receives GPS L1. There are a large number of options when buying survey grade equipment. Survey grade has traditionally been very expensive, but prices are starting to drop with increased competition. The satellite constellations are also improving, with GPS L2C and L5 due to be activated in 2016 and Galileo and COMPASS are reaching maturity, so I am going to be waiting a bit to see how receivers develop before making a purchase.

Coordinate Systems

Another thing you will need to understand if you are using GNSS is coordinate systems. I will discuss those relevant to the UK. There are two types of coordinate systems, geographic coordinates and projections. Geographic coordinates are based on a latitude and longitude, a number of degrees from a certain point. It is the measurement of the globe of the earth in 3D, and you can't put that on a flat map, hence projections. Projections are used to present a curved area of the earth on a flat map, and generally have an easting and northing rather than latitude and longitude. There are many different types of projections for different purposes, for example, one might preserve the size of land masses and another might preserve distances.

As for geographic coordinates, you will likely come across two in the UK. WGS84 is one of the most commonly used around the world. Consumer grade GPS will generally default to this, and it is the coordinate system used by Google Earth. The problem with WGS84 is continental drift. Different plates will move in different directions, so the same point on the ground will slowly change coordinates over time. The solution to this is to have a geographic coordinate system tied to each plate, so any given point on the ground wont change coordinates over time. the UK is on the Eurasian plate, which is covered by ETRS89. This system was created in 1989, and at the time was the same as WGS84, but the two diverge by 2.5cm a year, so the difference is currently about 60cm.

The most commonly used projection in the UK is OSGB36. This is the projection created and used by Ordnance Survey way back in 1936 for their mapping. Theoretically, a transformation from geographical coordinates to a projection like OSGB36 should be a simple mathematical calculation, but there is a problem. Back when they started surveying using OSGB36, the technology was not as advanced as today. Over long distances, slight errors crept in to the maps. When this was discovered, because of improving technology, OS had the opportunity to correct their maps, making the transformation from geographic coordinates to projection work again, but they didn't. What they actually did was to keep their maps the same, and create a database of local variations to the basic transformation (14.5Mb in size), which can be up to 20m horizontally. This set of corrections is called OSTN02 for horizontal corrections and OSGM02 for vertical corrections. Why is this a problem? There are variations in which pieces of hardware and software support it. Consumer GPS receivers almost certainly wont. Survey grade GNSS may do, but it is something worth checking, especially if you are buying old equipment second hand. As for software, Proj.4, the geographic transformation library used by pretty much all open source GIS software doesn't support it as standard, and neither does the standard commercial option, ArcView, though they do have an optional download which will allow you to use it. All of the digital maps for GIS supplied by Ordnance Survey are based on the OSTN02 corrections, so if you are surveying using a device that doesn't support it, you will be in the wrong place on the map. Most people in the UK will understand and expect OSGB36 coordinates, so a surveyor will have to use them, despite these problems. For the benefit of other surveyors, please also give ETRS89 coordinates in your geophysics reports. They will thank you for it.

18 March 2013

Latest Results: Alfoldean

About a year and a half ago, I started working with the Horsham District Archaeology Group at Alfoldean. I had intended to return last Autumn, but things didn't work out, so the task of continuing where I left off was pushed back to this (almost) spring. It certainly hasn't felt very spring like, but the signs are there, and I am slowly emerging from my winter geophysical hibernation.

(Click for larger image)

Unfortunately, we lost one of our total station resection points, always a risk, so we had to start with a fresh grid layout. You can see a small gap between the old survey to the north and the new survey to the south.

The part we were most interested in was the small enclosure next to the floodplain on the western side. The sort of tile associated with a bath house had been found here, and we wanted to find out more. What we found was quite intriguing. A new, thinner track appeared, heading south-west, directly for Dedisham Manor. Apparently, Roman material had been found there, and there was a theory that there was a villa at the site of what is now the manor. This was further proof that there was indeed something going on in that direction.

There was also a much more substantial track heading south-east, towards Stane Street, which seemed to continue on north-west, into the floodplain. It is possible that rather than just a bath-house, we have a river crossing, or a port, with the building material found being associated with that. Is this the start of a new Roman road? Where would it go? Neatham? I hope to find out.

Another interesting feature appearing in this years results was an enclosure, in the south-east corner of the survey area. The track from the north decreases in width as it approaches the enclosure, like a funnel. There doesn't seem to be much sign of occupation inside. Perhaps the funnel effect of the track is for funnelling animals, and the enclosure is some sort of animal pound. Were animals driven along Stane Street, and was this a place to store them on the journey?

As a final note, it seems that the trade in looted antiquities is alive and well in Sussex. We found several metal detecting holes across the site, including the scheduled area, which is of course illegal. More appeared over night between two days of surveying. Yes, the site is being looted by night hawks. Those people for whom our collective heritage means nothing more that a few quid on ebay. I am not against metal detecting. Having worked with many metal detectorists, I know that responsible metal detecting can provide a wealth of information on a site. Unfortunately, that is not the sort of wealth that these people are after.

14 March 2013

Talking, for a change

Occasionally, I will stop walking around with a machine that goes beep for long enough to actually talk about what I have been doing. If you live in the South-East of England, you might like to hear me waffle on about geophysics and Romans at two events this year.

Sussex Archaeological Symposium 2013

Date: Saturday 27th April 2013
Venue: The Fulton Building, Lecture Theatre A, University of Sussex, Falmer
Programme: Here
Application Form: Here

This will be a catch up talk, from when I last talked at the Symposium, two years ago. I've discovered a lot since then. Obviously, they have put me in the graveyard slot, after lunch, to wake up the weary troops with my superb oratory prowess (no laughing). Many of the other talks are on the Roman period, my favourite, so I'm very much looking forward to the Symposium this year.


Date: Saturday 16th November 2013
Venue: The Assembly Rooms, Preston Street, Faversham, Kent
Programme: Here

This conference is right up my street, as a lot of what I do is a combination of geophysics and landscape archaeology. A lot of what I will be talking about will be the same as at the Symposium, but it wont be a catch up, so I will be taking a more general view of things, plus of course talking about what I find over the course of this year.

06 February 2013

Latest Results: Barcombe

This is the big one. I actually did the bulk of this survey a couple of years back, but it has been kept under wraps until the fields had been cleared of coins, to avoid the place being looted by nighthawks. Now, finally, I can talk about it. This survey was done as part of the Culver Project and I thank the members for their help. We originally started this survey on the site of Ivan Margary's section 14 on the London-Lewes road  (See his book, "Roman Ways in the Weald"). Margary had mentioned that he had found pottery in the trench, by the side of the road he had exposed. Occupation is is good for showing up the Roman road ditches using magnetometry on clay, which are invisible on that geology otherwise, so we thought it would be good to look at what was there, and try to find out what was going on with the road further to the south. What Margary didn't know was that he was only a few metres short of the end of the road, and in the middle of a large fortified settlement, making this road the Barcombe to London road instead.

The northern road stops at an east-west road, most likely the Greensand Way, which is somewhat further to the south than Margary anticipated, and also continues to the east into the weald, most likely towards the settlement at Arlington. The settlement itself is quite substantial, with a defensive enclosure 165m square, which seems to cut a lot of other features. Many settlements were enclosed in such a way in the late second century, against a threat not well understood, apparently cutting through existing buildings and roads in the process, suggesting the locals weren't responsible for their construction. In Sussex, the roadside settlements at Alfoldean and Halland have defences dated to this period, though Chichester is a notable exception in Britain, with its walls being built at a later date.

The settlement and road network are surrounded by a number of plots of land enclosed by ditches. They may be fields, or small-holdings, or purely used for occupation, for which there is plenty of evidence in many of them on the geophysics results. One of these enclosures next to the east-west road seems to have been partially replaced by the northern road up to London, suggesting that the east-west road and at least part of the settlement were established before the construction of the road up to London.

Part of this settlement is being excavated this year, and is open to all for excavation, so if you wish to dig on this fantastic site, you can. I shall certainly be spending some time there. It is always nice to see my geophysics excavated, as I learn so much each time. I hope to see you there.

21 January 2013

Version 1.1 of Snuffler Released

The latest version of my geophysics software, Snuffler,  has made an appearance. What wonders are there to be seen this time around?

The main new feature is channel merging. This is where you can display multiple plots on the same image, for example a magnetometry and resistivity plot. The one restriction is that the grid layouts of the various surveys must be on the same alignment, but apart from that, the grids needn't be the same size, shape, be in the same place, or be at the same resolution. Here is an example. The survey is of a  medieval farmstead that was in use until quite recently, and is now in a woodland, under some trees. The magnetometry is in red, and the resistivity is in green. As you can see from the top-right corner of the image, there is a high resistance feature that has a large magnetic halo around it. This is the whole point of the channel merging idea, comparing the location of features on different plots.

Also added is manual destriping. If you have areas of particularly high magnetism in your magnetometry survey, the normal destriping tool may struggle to get it right. You could always use the Modify Selection tool, but that isn't particularly fast. Now you can simply select a line to modify using Ctrl  + mouse drag, and then use the two new buttons in the toolbar to add or subtract 0.5. The selection of the line will be hidden, though not de-selected, to aid visual comparison with adjacent lines.

Finally, I wanted to speed up the screen drawing a bit, as Snuffler can be a bit sluggish at times, especially with large images. One of the things that was slowing it down was having to support the dot density plot. I was thinking of how to speed up this part of the software, when I suddenly realised, 'Why on earth would you want to use the dot density plot?'. Let me give you a bit of background on that. When archaeological geophysics first started out, computer equipment was a lot less sophisticated. Everything was in monochrome, unlike today, where you can display and print in shades of grey. A 4x4 square of pixels, with various pixels black or white, will give you 16 'shades'. This restricts you to having a single reading a minimum of 4 pixels wide, and it still wont look anywhere near as good as a proper grey scale. The dot density plot is an anachronism, and needs to go. No-one should be using it any more, so the choice to me was clear, dot density had to go. But where I taketh away, I also giveth. I had been asked if I could provide a greater number of grey shades for the display. Snuffler had supported 16 and 32 shades of grey, now there is an option for 64. When I first wrote Snuffler, 16 shades of grey was 'browser safe', i.e. you could be sure that any browser could display all the shades. I'm sure browser technology has moved on since then. Not all surveys will benefit greater from 64 shades of grey, but here is part of an image which shows that off to an extent. The graduation of colours shows up nicely in the alluvium next to the Roman port.

The difference between the 16 shades at the top and the 32 shades in the middle is obvious, but I must admit that I struggle to see much difference between the 32 shades and the 64 shades at the bottom. I am assured that some people can. On the subject of shading, I found some interesting blog posts about why colour shading is a really bad idea, but don't worry, I wont be removing those.

You can of course download the new version of Snuffler at the usual place.

06 January 2013

What to look for in a magnetometer

I have been looking at buying a GPR recently, and trying to get decent comparative information between makes and models is difficult to say the least. It occurred to me that other people must have the same problem when buying a magnetometer, so I thought I would write a quick guide detailing what knowledge I have gained over the years. I wont be discussing anything too expensive, like alkali-vapour magnetometers, as they are outside most peoples budgets. I will stick to fluxgate magnetometers. There are three makes I will discuss.


Geoscan make the FM256, a fluxgate gradiometer with 0.5 metre sensor spacing. I have personally used its predecessor, the FM36. I am not entirely sure of the differences between the two models, though I gather than the main points I will touch on have not changed.


Bartington make the GRAD601, a fluxgate gradiometer with 1 metre sensor spacing. I personally have a GRAD601-2, the version with two sensor columns.


Foerster make the Ferex. I personally have no experience of these devices, and what I have to say on the matter is merely what I have learned from other people and the internet.

With all that in mind, I will discuss the differences between the devices in various categories.


One of the most important things to consider is the sensitivity of the instrument, and how good it is at picking up the slight changes we are looking for in archaeology. In my personal experience, the Bartington is more sensitive than the Geoscan, perhaps helped by the longer sensor columns. Apparently, the Foerster is not very sensitive. The Bartington wins here, not sure who comes second, but I would guess the Geoscan.

Setup & Stability

Fluxgate instruments, being directional in nature, need to be balanced before use. The Geoscan instrument has a manual process, where physical knobs are turned to align the sensors. The process is somewhat time consuming, and is not helped by the device suffering a lot from thermal drift, so you may find yourself realigning the sensors after each grid. The Bartington is much better here. It has an electronic balancing process, which calculates the differences in sensor alignments and compensates electronically. It is also helped by being very temperature stable. I only tend to balance it a couple of times in a day. The Foerster is apparently set up in such a way that it doesn't need balancing. I'm not quite sure how this works, but it seems to do so, thus the Foerster wins here, with Bartington second.

Array Options

The Geoscan instrument has an option to carry two separate devices on a carrying frame, with a single button to start recording. Each device has to be balanced and downloaded separately. The Bartington comes in the single column GRAD601-1 variety or the dual column GRAD601-2 variety. You do not need  separate balancing and downloading for each column with the GRAD601-2. Because of the lack of setup needed with the Foerster instrument, it is easy to have a large array of devices, perhaps towed behind a vehicle even. Foerster wins this one with the Bartington second.

GNSS Integration

The geoscan instrument has no GNSS integration. The Bartington has an option of a separate data recorder that uses GNSS, but that will not do normal gridded recording. The Foerster has full GNSS integration, which helps with its cart and vehicle towed setups. The Foerster wins this one with the Bartington second.


The Geoscan has a very good reputation for reliability, these things never seem to go wrong. The Bartington has a poor reputation for reliability. Personally I've had to have my machine repaired twice. Once to replace the motherboard in the data recorder, and once to have a sensor column rebuilt after water got in, causing thermal drift. They can be damaged by rain, especially after seals have perished. Other people I have talked to have had a similar experience. I don't have any information on the reliability of the Foerster instruments, but I would guess somewhere between the other two, so Geoscan wins this one with Foerster second.


I am somewhat lacking in information here. All I can say that is concrete is that my GRAD601-2 cost me £10,500 a few years back. I gather that the GRAD601-1 and FM256 are roughly the same price, but for two sensor columns, the GRAD601-2 is much better value than buying two FM256's. I know nothing about the Foerster prices. I can't call who wins this one, get some quotes.


What I would recommend going for depends on how you will be using it. If you are surveying using a gridless GNSS technique, then the Foerster is probably your best bet. If you are doing a gridded survey, I would recommend the Bartington. If anyone out there has further information to contribute to this guide, especially regarding the Foerster instruments, please leave a comment below.