Archaeology Resistivity Meter

Just came across this thread and it brought back happy memories of about 10 years ago when I did a resistivity survey near a Roman bath in East Sussex.  I too thought about building my own design but work and life got in the way.

The location aspect was the main thing that hit me as I was walking up and down the lines as has been mentioned ealier in the thread.   Both GPS accuracy, update rate and the point when the frame is vertical have been touched upon.   You can now buy cheap GPS modules for £4.50 from AliExpress (e.g. Quectel L96) which, according to the spec sheet, can update at 10Hz.   Assuming a walking rate with frame in hand of about 1m/sec (2.4mph) that would give a relative accuracy of 10cm, the reading being taken when an accelerometer chip indicates the frame is vertical.  GPS of course in low cost modules is only accurate to a few meters but if a 2nd GPS receiver were placed at a fixed known location in the field and able to communicate back to the mobile unit would that give you a more accurate differential reading?

Auto transfer at the end of a grid of the accumulated readings to a laptop would was another feature that sprang to mind those many years ago.  Meaning the minions can carry on trudging backwards and forwards while the boss looks at the results.

Hi Rob

robbarter  wrote:

 

Just came across this thread and it brought back happy memories of about 10 years ago when I did a resistivity survey near a Roman bath in East Sussex.  I too thought about building my own design but work and life got in the way.

 

The location aspect was the main thing that hit me as I was walking up and down the lines as has been mentioned ealier in the thread.   Both GPS accuracy, update rate and the point when the frame is vertical have been touched upon.   You can now buy cheap GPS modules for £4.50 from AliExpress (e.g. Quectel L96) which, according to the spec sheet, can update at 10Hz.   Assuming a walking rate with frame in hand of about 1m/sec (2.4mph) that would give a relative accuracy of 10cm, the reading being taken when an accelerometer chip indicates the frame is vertical.  GPS of course in low cost modules is only accurate to a few meters but if a 2nd GPS receiver were placed at a fixed known location in the field and able to communicate back to the mobile unit would that give you a more accurate differential reading?

 

Auto transfer at the end of a grid of the accumulated readings to a laptop would was another feature that sprang to mind those many years ago.  Meaning the minions can carry on trudging backwards and forwards while the boss looks at the results.

Re GPS, as Shabaz has mentioned above, unfortunately I don't think that's a goer in the way you suggest.

Two such modules in DGPS combination could be used the old days to partially negate the Selective Availability whereby the dithering for non-military-code users degraded the precision even further. The result that you're describing really comes from an RTK or PPK setup, but that has cost, complexity and practicality issues for the hand-held traverse-a-grid survey which will be the main use case.

More importantly though, in the main 'grid' use case, even if accurate positions are acquired (and my Trimble RTK kit can certainly achieve that) - post-facto positions, no matter how accurate, will deliver an inferior result as all the analysis requires points on a regular spacing - so if the measurements weren't taken at grid nodes, an extra interpolation step is needed, and that will inevitably degrade the result. I have spent hundreds of hours re-processing aerial LiDAR data to try and get better results that the simple averaging normally used to convert the for all practical purposes randomly located points in LAS/LAZ files into a regular grid. Most LiDAR that is delivered gridded is typically just a basic interpolation of a TIN surface, but that both smears everything and can introduce artefacts. There are a host of gridding interpolation strategies from nearest neighbour to inverse-distance and various other weighted strategies, to what I have found best is a certain Krigging setup - but all of those lose information.  If you are in the quick survey on something towed behind, say, a quad or a Land Rover use case, then post-processing with gridding is inevitable but expectations are (or should be!) lower.

Having said all of that, as I said in an earlier post, GPS tagging using a single such consumer grade module would be a good idea as it un-ambiguously identifies which grid is which, and the traverse start point / direction, so will assist when grids come to be stitched together for analysis.

Re auto transfer to a laptop - certainly as system it would be worth looking at ease of transfer. I wouldn't necessarily think every grid though, the appropriate occasion would probably be prior to relocating P1C1, so with typical 50m cable, that would be every third 20x20 grid - and the results would have some first-look validity as they wouldn't have the edge discontinuity that follows P1C1 relocation. Of course grids may be done in different orders or less at the end of a line, so probably a UI-initiated transfer (or maybe a pull from the laptop).

Dave

Dave,

Thanks for the info on GPS.   Saves me building a test setup. And that being the case then I agree that it can still be used to confirm the grid.  Possibly taking an average position of gps coords for each of the resistive readings to give a grid centre along with an approx start position and direction.

Rob

cheers Rob.

To be honest I wouldn't be against recording even a crude GPS position for each data point, so long as it is adequately qualified (e.g. Fix state and PDOP as well).

My thoughts are:

1)     If you record for each point, even if some are occluded then there should still be ample redundancy to identify the grid and direction of travel even if only a few points per line actually acquired.

2)     It would then be easy to substitute a high-precision (RTK) stream if the instrument ever makes it to, for example, a towed unit.

3)     It looks like the instrument may well need an initial post-processor to strip out data not required for existing processing (such as perhaps complex impedance, temporal data, etc.) so when exporting for processing the GPS data for each point in the grid, if acquired, could just be omitted from the file to be processed by the current upstream software.

Dave

davemartin there are three ways to determine travel you have gone.

• use an accelerometer and calculate how far you have gone.  Distance = time * velocity.
• use a rotary encoder (continuous) on your wheel.
• or in the DIY spirit use something else to follow wheel rotation. Hall Sensor, IR you pick.

If I were doing this (never done before of course), I would probably use GPS to give me a geo reference to my 20m by 20m site, then use some twisted braided rope (usually get 50m or so for about $15) and then mark out my 1 metre intervals on the rope and then mark out a length across the site using the rope, which will then used as a guide to ensure I walk in a straight line and to ensure I take measurements every metre.... for once, a solution with little electronics involved...

phoenixcomm  wrote:

 

davemartin  there are three ways to determine travel you have gone.

• use an accelerometer and calculate how far you have gone.  Distance = time * velocity.
• use a rotary encoder (continuous) on your wheel.
• or in the DIY spirit use something else to follow wheel rotation. Hall Sensor, IR you pick.

Agree - but in the main use case you pick up the frame, move it and plunge it down into the earth at the next sample point. The priority is to take a reading at the right point on an interval node (target is each point within 10cm); otherwise you have to interpolate back to the regular node points and as mentioned above that loses information.

With a towed rig, yes, you can use odometry or inertial, but with the towed rig you need as precise a record of your path across the field or wherever, not just distance along the track, hence typically use RTK if available.

Dave

BigG  wrote:

 

If I were doing this (never done before of course), I would probably use GPS to give me a geo reference to my 20m by 20m site, then use some twisted braided rope (usually get 50m or so for about $15) and then mark out my 1 metre intervals on the rope and then mark out a length across the site using the rope, which will then used as a guide to ensure I walk in a straight line and to ensure I take measurements every metre.... for once, a solution with little electronics involved...

You're pretty darn close!

Typical procedure:

1) Set up a base line, can be just in the most obvious by eye to an experienced surveyor, which if possible runs on as long an axis as possible to span the site.

2) (at some point, pick up co-ordinates of either end of the baseline by RTK or Total Station or measurement to suitable features)

3) Baseline can be maybe say a 200m tape

4) Set a right angle out from the baseline at either end and put another tape at either end. Right angle can be achieved anything from RTK, to a surveyor's hand-held twin prism, to a total station, to just a 3:4:5 triangle.

5) At 20m out on each of the end tapes, stretch another 200m tape - that will then be parallel to the baseline and 20m offset

6) Mark every 20m along both baseline and second long tape - that's your grid corners.

When it comes to doing the actual sampling, along the baseline and opposing edge of the grid, rather than try to read the tape (which invariably twists) you usually use a pair of marked marked lines. Sometimes use rope and splice-in markers just like those on an old coastal-navigation sounding lead line. Rope can stretch though, so favourite is plastic-coated washing line, marks being short lengths of glue-lined heatshrink. If your traverse lines are 1m apart, the marks are at 1, 3, 5, 7, .. metres. You then put a guide line between the two 1m marks on either side of the square (guide line will have marks every 0.5 or 1 m to match sampling interval) then you walk up the grid with the line on, say, your right hand side; then when you get to the far end you come back down the other side of the guide line. Whilst you're doing that, your colleagues will have placed another guide line between the 3m marks; and once you've started on that they pull the line joining 1m and leapfrog to lay it between the 5m marks. Rinse and repeat ... (takes longer to describe than do!).

Dave

davemartin

Agree - but in the main use case you pick up the frame, move it and plunge it down into the earth at the next sample point. The priority is to take a reading at the right point on an interval node (target is each point within 10cm); otherwise you have to interpolate back to the regular node points and as mentioned above that loses information.

Sounds like a freak'n workout! If wanted to do this. I would automate the frame "plunging into the earth"! I reality I most likely let it be autonomous!

~~Cris

And when I did it there were still crops in the ground!  It's not as easy as wheeling over a bowling green some of the time.

phoenixcomm  wrote:

 

davemartin

Agree - but in the main use case you pick up the frame, move it and plunge it down into the earth at the next sample point. The priority is to take a reading at the right point on an interval node (target is each point within 10cm); otherwise you have to interpolate back to the regular node points and as mentioned above that loses information.

Sounds like a freak'n workout! If wanted to do this. I would automate the frame "plunging into the earth"! I reality I most likely let it be autonomous!

~~Cris

You're absolutely right Cris, it is a good workout ..... but attempts at wheeled-probing are actually more of an issue and have probably caused more angst than the electrickery!

Just to elaborate a little on the (current) main use case for twin-probe archaeological resistivity survey.

The mobile probes (C2/P2) are mounted on a frame, which the operator locates at each sample point. The frame has a handle across the top and the operator holds it with a grip akin to holding bicycle handlebars. The frame is dropped at a sample location so that the two probes plunge into the soil. Once the reading has been satisfactorily taken, the operator lifts the frame, advances to, and drops it in the next location.

Points to note with this are that the probes penetrate ground (distance can vary but perhaps 1 - 2 cm). On the way down to the ground, the probes and frame may have come down through anything up to, say, 15 cm of vegetation. The probes don't always make good contact with ground / complete the circuits; there is both haptic feedback (your wrists know you hit a stone!), or there is no beep so you wait or re-plunge the frame maybe a cm or two displaced. Then, when you are ready to advance, you have to lift the frame and probes clear of not only the ground but also the vegetation. So, the locus of the probe tips is (using X for line of advance and Z=0 for earth surface), say Z = -1, take reading at X = 0; then raise Z = +10, advance to X = 50, drop frame so Z = -1. So, the vertical stroke of the probes is 11 cm at each sampling point.

The issue with trying to use a wheeled cart is achieving good contact for the probes. On a bowling-green type flat, very short turf, you can use spikes on the wheels and - stones and voids permitting, you'll get a reasonable set of readings.

If you have any height of grass or other vegetation, or say stones on the surface, the wheels ride righer and the probes need more 'stroke' to get to/into the ground, and it is increasingly difficult to get reliable contact. Issues with wheels riding on top of the vegetation can be addressed in part by both making the wheels as narrow as possible, and also by adding ballast. Radially mounted spike probes, if long enough to make contact, also have more of a swathe path, rather than the up&down when used with a frame, so tended to drag one wheel or the other. Overall experience with a hand-propelled wheeled setup in lieu of the frame for twin-probe (grid-based) res was not good.

Have also spent quite a lot of time trying to come up with a workable towed rig, with all four probes on the towed rig (i.e. no tether to P1C1) but all four probes in a travelling 'square array'. Problem with a four-wheeled cart in some respects even worse that with just two wheels/probes - difficult to get good contact, and need effectively independent suspension to avoid one wheel being off or further away from the ground; best results found with a three-wheel cart, two for contacts and one non-contact 'jockey' wheel, plus two more independently trailing ballasted wheels for the other contacts. Have also experimented with using cranks to throw the probes more downward, and even long-stroke solenoids. Overall though, results not worth it as towed = constrained to areas it can be used in.

So, 'workout' for the team still!

Dave

phoenixcomm  wrote:

 

davemartin

Agree - but in the main use case you pick up the frame, move it and plunge it down into the earth at the next sample point. The priority is to take a reading at the right point on an interval node (target is each point within 10cm); otherwise you have to interpolate back to the regular node points and as mentioned above that loses information.

Sounds like a freak'n workout! If wanted to do this. I would automate the frame "plunging into the earth"! I reality I most likely let it be autonomous!

~~Cris

You're absolutely right Cris, it is a good workout ..... but attempts at wheeled-probing are actually more of an issue and have probably caused more angst than the electrickery!

Just to elaborate a little on the (current) main use case for twin-probe archaeological resistivity survey.

The mobile probes (C2/P2) are mounted on a frame, which the operator locates at each sample point. The frame has a handle across the top and the operator holds it with a grip akin to holding bicycle handlebars. The frame is dropped at a sample location so that the two probes plunge into the soil. Once the reading has been satisfactorily taken, the operator lifts the frame, advances to, and drops it in the next location.

Points to note with this are that the probes penetrate ground (distance can vary but perhaps 1 - 2 cm). On the way down to the ground, the probes and frame may have come down through anything up to, say, 15 cm of vegetation. The probes don't always make good contact with ground / complete the circuits; there is both haptic feedback (your wrists know you hit a stone!), or there is no beep so you wait or re-plunge the frame maybe a cm or two displaced. Then, when you are ready to advance, you have to lift the frame and probes clear of not only the ground but also the vegetation. So, the locus of the probe tips is (using X for line of advance and Z=0 for earth surface), say Z = -1, take reading at X = 0; then raise Z = +10, advance to X = 50, drop frame so Z = -1. So, the vertical stroke of the probes is 11 cm at each sampling point.

The issue with trying to use a wheeled cart is achieving good contact for the probes. On a bowling-green type flat, very short turf, you can use spikes on the wheels and - stones and voids permitting, you'll get a reasonable set of readings.

If you have any height of grass or other vegetation, or say stones on the surface, the wheels ride righer and the probes need more 'stroke' to get to/into the ground, and it is increasingly difficult to get reliable contact. Issues with wheels riding on top of the vegetation can be addressed in part by both making the wheels as narrow as possible, and also by adding ballast. Radially mounted spike probes, if long enough to make contact, also have more of a swathe path, rather than the up&down when used with a frame, so tended to drag one wheel or the other. Overall experience with a hand-propelled wheeled setup in lieu of the frame for twin-probe (grid-based) res was not good.

Have also spent quite a lot of time trying to come up with a workable towed rig, with all four probes on the towed rig (i.e. no tether to P1C1) but all four probes in a travelling 'square array'. Problem with a four-wheeled cart in some respects even worse that with just two wheels/probes - difficult to get good contact, and need effectively independent suspension to avoid one wheel being off or further away from the ground; best results found with a three-wheel cart, two for contacts and one non-contact 'jockey' wheel, plus two more independently trailing ballasted wheels for the other contacts. Have also experimented with using cranks to throw the probes more downward, and even long-stroke solenoids. Overall though, results not worth it as towed = constrained to areas it can be used in.

So, 'workout' for the team still!

Dave