00:00.00 archpodnet Welcome back to the architect podcast episode 82 and we're catching up with Paul after his couple months of field work over in the Middle East in Iraq more specifically and Paul you mentioned a few times we talked about it before you went that you you took some magnetomery training and I'm really interested in this because I haven't used a magnetometer for. 12 years since I was in grad school right? So I'm sure the well the principle hasn't really changed obviously but I'm sure the technology has changed for. You know how all this works and you mentioned having a process the results and all that and we'll talk about that. But I'm curious as to how you think your? um. 00:20.65 Paul M. 00:37.23 archpodnet Your survey methodology came out I remember seeing ah because the magnetomery the magnetometer has um, ah different like stabilizing poles and stuff like that right and would and they make these little holes I remember seeing a picture you put on Twitter of like the grid of holes like from a drone shot from doing the magnetomemetry survey. Which was kind of cool. Yeah, oh okay, okay. 00:56.53 Paul Ah, those holes those holes are my footprints and that whole pinstriing video ah image is ah the 2 tracks are the ah the wheels on the cart for the magnetometer and those holes in the middle are my footprints. 01:09.34 archpodnet Ah. 01:12.57 Paul Ah, so basically what I had done is I divided up our site into forty five Meter square grid and I positioned it so that it would line up with our trenches in a way that meant that I had as many so grid squares as possible without having to you know. 01:21.40 archpodnet Ah. 01:31.63 Paul To to work around our our existing trenches. Um, and the reason why I did forty five meters and I mentioned this before but I'll mention it again because I think it's worthwhile. Ah for people to hear is that we had 3 sensors on our cart. So. It's a one meter white cart and there's a sensor each fifty centimeters so one on the left one the middle of one on the right each of them fifty centimeters from the next. Um, it turns out that on a zigzag bustrafidon style pattern going across these grid units. That the width of the grid unit has to be evenly divisible by the number of sensors that you have so I have 3 sensors I want to do a fifty meter grid but I can't do that. So my options were forty five forty Eight or well. 02:14.51 archpodnet Okay, yeah. 02:27.83 Paul 60 was to be the next sensible one up 48 doesn't make a lot of sense because if I go to a 5 set of 5 sensor grid 48 isn't even lead devisible by five forty five worked and 60 would work and I went with 45 because I have fifty meter tapes 02:29.60 archpodnet Yeah, yeah. 02:47.80 Paul And I didn't want to have to lug along one hundred meter tapes or something else to ah to do it so it was limited by that it doesn't really matter a whole lot. But um, but that's why I went with forty five meter grid which is kind of weird sounding but it makes sense once it's explained. 02:49.87 archpodnet Nice. 02:58.37 archpodnet Yeah. 03:03.70 Paul And I'm going to put another plugin on my github page I wrote a little script that anybody can download and look at that will do those calculations for you I'll tell you what your center line should be what your first and last sensor should be if you input the width of the grid and the number of sensors in the width of the ah the car that you're using. It'll calculate all that for you so you can help plan. Um, so yeah so I would just choose a grid unit and then with the total station. We would lay out the corners of the grid and I would lay the tapes out a baseline east to west on the south and another one east to west on the north. 03:24.31 archpodnet Okay. 03:35.70 archpodnet Ah. 03:41.86 Paul And then the way that I was taught I mean I've done gpr magome tree where you pull a rope or a um or a chain or a tape northouth from the baseline to the other line to the endline and then you. Follow that to you know to guide you to keep you going straight and then you know if you're using um a system that has a trigger that you have to press manually you press at every one meter or every whatever the interval is um, the system that we had had an odometer on the cart. 04:00.59 archpodnet And. 04:11.40 archpodnet Okay. 04:15.80 Paul So I didn't have to worry about pressing a trigger the ah the odometer would do that automatically but I did still have to worry about going straight so instead of drawing that that chain North South as a guide what I was taught and this work beautifully is I could. 04:34.30 archpodnet Um, replie. 04:34.50 Paul Place a traffic cone center line of the target on the on the north line and then one on the south line and each time you know so I'd aim at the traffic cone and why I got there I would move the traffic cone a couple meters over and turn around and come back and move that traffic cone a couple meters over and I would just do that. Back and forth moving the traffic cone and it worked great for me because what this meant was normally you have to have 2 or more people for a survey like this because you've got people that are moving the chains aside from laying out the grid units the corners. Ah i. 05:05.75 archpodnet Ah. 05:11.18 Paul Where I needed somebody runninging the total station I was running the the poll and staking um I was doing it all by myself I mean there were a couple of times that people help me and I certainly it it sped things up a little when they help me but it was not necessary to do this work and so that made it a 1 personson job for the most part and. 05:18.58 archpodnet Yeah. 05:27.46 archpodnet That's cool. Yeah yeah. 05:30.62 Paul We got toward the end. We got a yeah ah dgps with the trimble corrections and that meant that I could actually lay out those corners in a couple areas so we did some tests around the site all by myself took longer than it does with the total station. But. You know it was then entirely start to finish a 1 person job. 05:51.74 archpodnet Yeah, nice. Well then that's really cool when you've got ah obviously a limited crew and you can't just like call somebody and say hey can you come help with this so you know making that even if it takes you just a little bit longer. It's still way more efficient in this scenario. So um, you know I'm also wondering. So you mentioned that they did a ah magnetomemetry survey, um, back in 2019 on this area. Did you survey the exact same area on lagosh. 06:13.50 Paul M. 06:21.40 Paul No I planned mine so that I mean the grid that I drew up covers the entire site including the areas that previously been surveyed but I planned it so that I would do once that were complimentary that hadn't overlapped with the other ones. Um it turned out that that 2019 stuff 06:26.67 archpodnet Okay. 06:33.93 archpodnet And did they oh really? okay. 06:39.10 Paul Had a significant spatial error and I found the error. Yeah and I corrected it I corrected all that 2019 data and once I did so it it lined up beautifully with not just my data because you know there I'm using my as a reference and and. 06:53.23 archpodnet Like. 06:58.67 Paul Well who's to say which is more accurate but it also lined up with what we were finding in our excavations which we weren't finding in the spring which was baffling to us why we had this beautiful magnetomery that looked like we could see features. But then when we dug they weren't showing up in the same place. Well. 07:03.17 archpodnet M. 07:13.42 archpodnet Sure, ok, ok, yeah, so did they end up I mean even though they had this spatial error. Did they find anything in 2019 of any significance. 07:15.51 Paul Turns out because they were twenty meters over 07:29.30 Paul Ah, they did and there is a ah, an article that's making the rounds right now about worked done at Lagosh that relies in part on this really badly botched um magdometry from 2019 I rectified it. Um, and so basically. 07:40.26 archpodnet Yeah, like. 07:47.80 Paul Doubled the footprint of what we had here. Um, you know the 2019 data plus mine and things line up beautifully I mean so ah, you know streets and walls that you can see in the 2019 data go off the edge of the 2019 data and write into my. 07:48.82 archpodnet Okay, okay. 07:58.69 archpodnet Yeah. 08:05.91 Paul 2022 data perfectly in alignment and then there's a trench and you can see where the street is in the trench we excavate it and we found it in excavation. So um, it's for our purposes. It's really proving to be incredibly valuable for predicting. What's what we're going to find subsurface. Um, we. 08:09.28 archpodnet Yeah, nice. 08:25.10 Paul Yeah, yeah, actually I'll to tell you a little example so we had a student an undergrad that was on the project and she had very little field experience. She would actually and this is something else I'm going to want to talk about at some point is ah my program that I've hinted at for a long long time about my total station program I Finally got it up and running. 08:26.50 archpodnet Yeah, yeah. 08:44.59 archpodnet Um. 08:44.90 Paul Well enough that I dogfooed it for both these projects for or and for loggosh and made a whole bunch of little like interface tweaks as I went to make it easier to use in the field but I taught this undergrad how to use it and she became our total station person. 08:52.49 archpodnet Okay. 09:01.19 archpodnet Nice. 09:02.84 Paul And so she would ah you know while I was training her one day I was like well let's let's go. We're going to open a trench in this area. So why don't you go with the the till station we'll we'll have you and I was running the total station there because I was still I was still you know tweaking some of the interface things. But I gave her the Pollle and again very little field experience. I said find everything that you think looks interesting on the surface here. So if you see a bunch of pottery if you see a ring of something and tell me what you think it might be the rings are going to be drains the piles of pottery are just going to be positive and so she went and she tagged all these different areas. She's like. Here's a ah concentration of pottery here looks like a drain here is some some dark soil that looks kind of ashy and then we ran the magnetomery over it and what she picked out on the surface and what we found on the magnetomery were matched exactly. It was beautiful. 09:44.78 archpodnet Um. 09:55.83 archpodnet Nice, Nice. Yeah yeah. 09:59.71 Paul And then when we excavated it matched even better. So um, you know we could find oh you found that that ashy bit Guess what we found here. We've got an oven. 10:10.26 archpodnet Nice, Well, that's really cool, especially when you get that anytime you get that kind of correlation. It's just ah, you know, magical vindication of everything you've been trying to do so I'm curious as well. Again, it's been a long time since I saw um. 10:22.16 Paul Yeah, really, it's very cool. 10:27.28 archpodnet Been a long time since I saw magnetomemetry data is there and and you you mentioned at the beginning of this conversation that you know you still have to process all the results you know and and really figure out what you had going on but out in the field. Do you get? what? what kind of indication. Do you get just from the results you're getting in the field with um, well first off. Can you tell anything as you're out there actually running the equipment or do you start maybe getting a hint at some stuff once you bring it back in toss it all on the computer without like full processing but maybe minimal processing How much can you tell out there is what I'm kind of getting in. 11:04.62 Paul All right? So the ah the processing the um I think these conflating 2 different things a survey that that stillney's processing is my my systematic surface collection. Um that I started before that was on the um. 11:15.68 archpodnet Ah, okay. 11:20.14 Paul The data that we have for the magnetomemetry were processed nightly when we're in the field and actually the the data collection happens now with the system we're using on I'm holding it in front of me a Panasonic tough pad f z g one. So it's basically a windows computer in a tablet form and. 11:22.62 archpodnet Oh okay. 11:34.90 archpodnet Yeah, yeah. 11:37.99 Paul Um, the data collection Software does not show you what? yeah and for this particular system. The Census system does not show you what you're seeing under the ground but as soon as you're done with any one of the units you hit done and. 11:50.34 archpodnet Ah. 11:54.89 Paul Then just open up a different program and process the data right? There takes just a second to import it change the um you know, just change. Ah do do a little bit of um to a Median correction on it and then and then adjust the. 11:57.41 archpodnet Oh wow. 12:10.67 Paul The values that you're showing and boom. You could see it right there. So within minutes of me finishing a unit I could look at the at the data that I just collected and so sometimes we had done that you know there wasn't any real point for me doing that for the most part but there were a couple days that we had visitors on site and they saw me out there. 12:21.64 archpodnet Ah. 12:29.16 archpodnet Ah. 12:31.10 Paul With Bessie Bessie's what I'm calling the cart because it's black and white and I'm plowing this field. It reminds me of a cow. Um, ah I'm out there with Bessie and then I finish up and I can quickly process it as I'm walking back towards the visitors and I can show them exactly what I had just picked up. You know subsurface you know. 12:34.62 archpodnet Nice. 12:47.76 archpodnet So Wow that's really cool. Nice nice, all right? Well any. Ah yeah, sorry keep going. 12:49.52 Paul 15 minutes ago 12:54.80 Paul Yeah, so it's not. It's not real time but but. 13:06.76 Paul So It's not real time but it's ah but it's close enough that for our purposes you know we could definitely use it in the field and and see the results right there you know, maybe during break or something. Ah, the only caveat to that is the software that we're using runs with a hardware Dongle. So if I wanted to do that on any particular day I'd have to remember to bring the hardware dongle with me and um, yeah, the fear then is losing the dongle somewhere out in the field. Um. 13:19.66 archpodnet Ah. 13:31.98 archpodnet Yeah, okay. 13:35.69 Paul So I don't know that I'd always recommend that but but it it worked in those cases where I had to do so. 13:41.27 archpodnet Nice all right? Well I think that's ah good enough for now on this episode. We just wanted to have a quick catch up with you and see what was going on. We do have ah an interview coming up and and with actually somebody you worked with out on one of these projects and we've got some more. Good episodes and some stuff planned. We're going to get back on track. We got off track for a few reasons a Paul being gone and b we did the October shutdown on new written new episodes and we've you know we've had some. Some you know times getting back in plus we moved the apn onto a completely different hosting service in the last couple of weeks that's been really fun. So aside from all that I think we're we're getting back into it and yeah, it's just been a lot so it's been a fall for sure and the apn turned 8 on December first so we are now. 14:21.19 Paul Oh boy. 14:32.34 archpodnet Ah, eight years old officially as a network which is pretty cool. So anyway, ah paul any ah, any last parting thoughts on your on your logosh fieldwork to to kind of sum up your time out there. 14:46.60 Paul So it was excellent I'm glad I went I'm glad to be part of this project I can't wait to go back again next year we have a lot of interesting ideas that we're working on things are starting to gel in terms of how we understand. 14:55.56 archpodnet Ah. 15:01.43 Paul History of the site What we understand about the topography How we've matched up different data sets between the the soil cores and the thermal imaging and the excavation and the magnetitude tree and the aerial survey. Um, even yeah, just like I was noticing doing the Magnetome tree that. Some of the gullies on the side of the the mounds on the tell seem to follow the um, the ancient roads subsurface which I guess makes sense to some extent. Um, but we didn't expect that. Um and it's not everything not every gully follows a road but it's. 15:25.30 archpodnet Oh yeah. 15:37.63 Paul Common enough and we didn't know that before you know we we realized that as we were doing it and that's one of the things that I really missed about um about doing the field work is that you know we're there talking about this stuff immersed in it with our colleagues day in and day out and you get a bit of a feel for something and you. 15:38.69 archpodnet Yeah, like. 15:55.23 Paul You know chew the fat with them and maul some ideas over and some of them end up in the circular bin but some of them actually seem to bear fruit some of them bear a little more looking into oh that's the other thing about the magnetomery data. So the magnetomemetry data we have. 15:57.87 archpodnet Ah. 16:06.60 archpodnet Yeah, yeah. 16:12.78 Paul Distinct signatures that we can eyeball so I can look at something I can tell you it's a wall that's a mud brick wall I can look at something and tell you it's a Kiln I can look at something and tell you it's a street. Um, what I want to do is actually formalize that and so we're starting conversations with somebody that can do a little bit of machine learning a little bit of a computer vision. 16:18.76 archpodnet Ah. 16:31.43 archpodnet Ah. 16:31.65 Paul To help define these patterns. The things that we're picking up intuitively but then look at the difference in the levels and shapes and juxtapositions of highs and lows to try to to formalize and regularize that and then let the computers do it for us as maybe a first pass through to say hey this looks like it's a road. 16:39.23 archpodnet M. 16:49.41 archpodnet Cool. 16:50.84 Paul Hey this looks like it's a wall hey this looks like it's a Kiln hey this looks like it's an oven This looks like it's a drain this you know and then maybe other features that we haven't yet been able to identify So we're you know we're going to be exploring other post processing towards sorts of techniques like that that I think will be pretty exciting. 17:09.65 archpodnet Okay, cool. Well that is pretty exciting. Definitely interested in hearing more about that. So all right Paul? Well with that I think ah, we'll go ahead and head out for this episode and thanks for thanks for coming back and and bringing back all kinds of good information and we will. Come back next time with some some more great stuff I think that episode's coming out right before Christmas so at least here in the United States and if you're I don't know celebrate Christmas if not it comes out right before any other weekend. So there you go all right? Well thanks everybody and again we'll see you next time. Thanks Paul. 17:44.68 Paul Thanks Take care chris.