By Victoria Pagano
Hello again, it’s Victoria here to tell you that I’m excited. Excited about the work we’re doing out here in Eagle Cave and with the ASWT project as a whole. Now this is not to say that I wasn’t enthusiastic when I first found I would get a chance to intern in an amazing place, with knowledgeable people, learning and doing great new things; but, I’m writing now with a little training under my belt as to the way things work and and a better understanding of how absolutely fantastic it really is.
First, I would like to tell you a bit about my first and only field work in archaeology…just to offer a bit of perspective. I was unbelievably lucky to work in Belize. A beautiful country full of cultural and ecological diversity– not to mention the incredible historical and archaeological richness it holds as well. The project was based in the Mopan River Valley, studying the ancient Maya sites of San Lorenzo, Xunantunich, and Buenavista del Cayo. My work was focused at Buenavista, a mid-level city center with plazas and stone structures that had been reclaimed by the jungle.
It was in Belize that I learned basic excavation procedures:
Step 1: Find somewhere you want to excavate and establish an excavation unit. This includes (for most) establishing a permanent datum, laying out the unit, and taking starting measurements.
Step 2: Establish your excavation protocol. Are you going to follow natural breaks in the stratigraphy, or are you going to use an arbitrary measurement to create your strata, lots, layers, etc. You’ll probably want to sketch and photograph the starting and ending surfaces, too, as you work your way down.
Step 3: You find something really cool in the floor or wall of your unit… a hearth, post-hole, projectile point, a body, etc. — you decide you want to make sure this is in your notes (hopefully you are taking notes, good notes), so you need to take additional steps.
Step 4-6: You need to 4) Take photographs– with a scale and some indication of direction 5) Map it i.e. create a drawing by measuring to and from objects in your unit to an established point or points, like a sub-datum. This will yield a plan or profile map with detail as to where your find is and where everything else in your unit is in relation to your find. Detail, detail, detail! Depending on how precise you want your map to be, if you have help, and your level of OCD, drawing a map can take anywhere from minutes to hours. 6) Take more notes of the object’s location, this may include a GPS point that you tie to your datum later, or measurements that you will use to associate the object’s location relative to the datum.
Step 7: You’re probably pretty tired from all those steps you took to draw your unit. You need a nap, but chin up, you established your unit today AND you found something! Hopefully your notes are good, you read that compass properly, and you’re mapping skills are adequate enough that your map doesn’t simply look like a box with a few misshapen circles, squiggly lines, and a triangle.
Now I have nothing against all those steps (the old fashioned paper and pen method works), but there is always room for improvement. So why is it I am so ecstatic to work in a place where there isn’t monumental architecture, elaborate burials, mysterious mythology and codices? Structure from Motion.
What is Structure from Motion?
Structure from Motion (SfM) is a surprisingly simple technique that is easy to learn, quick to do in the field, and potentially available to archaeologists wherever they work, or at least those with access to modern technology. SfM uses still-motion photography to rebuild real-world, dimensional objects. Using a digital camera you take a series of overlapping, sequential photographs of your desired target and run them through a software program, such as Agisoft Photoscan. The software is able to match up all the different photographs and build a virtual 3D model of the target (for more info on what Structure from Motion is, see our blog post from last spring: Structure from Motion).
For the ASWT project, we are using SfM to document and record everything from entire sites to small excavation layers. In other words, a digital camera and a computer take the place of the traditional pencil and sketch map technique that I became familiar with in Belize. Creating sketch maps is somewhat fallible in terms of reliability due to human error; we can only record and note what we see or notice at the site. Often, having only a single chance to record something before we move on to the next layer. Even more often when we sketch we focus on the big things, the obvious things, not necessarily because we think the rest inconsequential, but because we cannot physically draw every detail. SfM captures all of the visual detail that we can’t see or maybe don’t even think to include at the time because we’re so focused on recording our super cool projectile point or rock alignment.
When it comes down to it, many of the steps and methods are the same (we’re still completing forms and taking notes and we aren’t taking any shortcuts), but what really changes is the end product: our results. Our notes, drawings, photos, and forms are all we have left after an excavation. SfM offers us a permanent, virtual record that preserves and offers accessibility to our excavation data for years to come (and dozens more eyes). Nothing gets skipped over, nothing forgotten.
Our Work So Far
As I mentioned in my introductory blog post, I am interested in archaeological applications of GIS. I also mentioned that I was intrigued by the SfM technique that I myself first learned about from this blog [Eagle Nest Canyon at the Texas Archeological Society Annual Meeting]. Now I come to you one month in, with a bit more knowledge on the project and the technique to present another perspective.
I spent the January session re-exposing a profile face, PS005, that was initially exposed in 2014. This profile sits in what we now call Strip 4, almost smack dab in the center, top section of the South Trench wall of Eagle Cave.
At first it was a mess. After removing the backfill and geo-cloth, we discovered that the profile face had suffered damage from continued erosion and rodent burrowing since it was originally exposed. In 2014, the investigators assigned strat numbers based on their original profile exposure –i.e. each visible stratum received a unique number. However, they then excavated a small sampling column and did their best to follow the layer seen in profile across the unit. The presence of numerous rodent burrows, especially through the ashy layers, made strat definition challenging.
I should add one more factor, at the end of the 2014 excavations the PS005 profile was sampled by the geoarchaeologists who removed micromorphology samples. Although done carefully, the wall was no longer pristine.
This helps explain why when we re-exposed the PS005 profile we could not easily match what we were seeing in the field to the original profile illustration. So, we decided to excavate a sampling column through a portion of the jumbled profile, using the 2014 strat numbers for our layer designations . This was done with two goals in mind:
1) Cut back eroded face (profile) and re-expose the stratigraphy.
2) Collect high-resolution samples of the matrix and artifacts within the profile.
Excavating a sampling column involves collecting the matrix of each layer (along with things like Spot Samples, Geo-matrix Samples, and samples for radiocarbon dating) that can be further analyzed in a lab. We are not only collecting samples of each strat, but using the TDS shots of each sample and the strat location, we will add them all to the SfM model. So whoever processes and analyzes the samples can have a virtually exact geospatial reference of its origin. This will help us build an assemblage of associated artifacts, radiocarbon dates, and deposition event, aiding in our understanding of the shelter and the canyon: how it was used, when it was used, what they were doing there.
Rather than draw a standard paper and pen illustration of each layer as we excavated, we instead used SfM to document the top surface of each strat. This not only gives us an idea of what we were looking at, but it allows us to use GIS to calculate volume of matrix removed.
Once I finished with the sampling column, attempting to follow the strats that were assigned the previous year, the profile face that was exposed was extraordinarily rich. In other words, by cutting back the wall we found better preserved and more complex stratigraphy. The newly exposed profile exposure is called PS010.
Previously, only about 10-12 strats were identified in this area. We have now defined 22 individual strata from the “same” exposure. I re-photographed the profile giving us three sets of 3D data: TDS shots, 3D models of all the excavation layers, and now the model of newly exposed PS010. We now have a new high resolution 3D model to overlay all of the excavation layers and samples onto – all of which can be manipulated to aid in analysis.
Where it All Comes Together
Our field lab is where all the sets of photographs are processed. Using Photoscan we align and georeference all the images for each individual layer. The photographs, GCPs, TDS, and notes are all combined to digitally rebuild the excavation. A 3-dimensional, manipulable dataset that works hand-in-hand with all of the physical data—matrix, artifacts, etc.—and the recorded data i.e. notes, photos, etc. In order to have these georeferenced for GIS or used in photogrammetry, no less than six GCPs, ground control points, are included in each excavation exposure. Ground control points are geospatial reference points that you place on your object or in your unit, shoot in with a TDS or GPS, so that photographs and models can not only be more accurately aligned with each other, but linked to a geographic grid. This becomes incredibly handy when you are working in say, a canyon with multiple sites carrying on extensive excavations that you would like to map and relate to one another. Then, not only can you reference all of your units and sites among the canyon, but you can reference and cross-analyze your work with other sites across the region or the world. Once we have our models we can then export all or parts of the model into many different formats; GeoTIFF, TIFF, JPEG, KMZ, etc. Our models are ready to imported into GIS software where we can further manipulate and analyze them.
A goal of the ASWT project is to not only excavate and collect, but to gather the best data we can – or best representation of that data –backing it all up with SfM and GIS. Structure from Motion gives us the opportunity to not only georeference our units, finds, and strata, but we can literally rebuild them, at least digitally speaking. No longer are we relying upon the traditional mapping, measuring, and sketching techniques of years past that result in rather dimensionless visualizations of excavations.
SfM also easily provides a new solution to an old problem: excavation vs. preservation. The basis of archaeology is essentially destroying material history in the name of research and discovery, so that we can preserve and record it as best we can. Granted we have gotten much, much better at recording and excavating than back in the early days of the field, there is still room for improvement and innovation. In 2014, Bryan Heisinger (2014 ASWT Intern; 2015 ASWT Staff Archaeologist) presented at the Texas Archeological Society annual meeting, on the uses of SfM and GIS for not only modeling, but extrapolating volumes of material removed and creating digital elevation models (DEMs). These can be used to study stratigraphy and depositional events of floods, people, and even animals– as Emily and Larsen can attest to. Our documentation of profiles, like PS005 and PS010 helps us build a database of all the excavations and the shelters to aid in the analysis of what is to some a rather abstract concept of time. Our work becomes more dimensional, more visible. You aren’t just looking at the profile of a wall or structure or shelter. You can virtually walk around that wall, walk into that structure, and around that shelter, without ever being there. The outreach potential is exponential.
The ASWT project personnel and many of our colleagues believe that SfM is that next step in improving archaeological documentation. Incorporating SfM and GIS technology we can model excavations with millimeter level precision recording finer detail in stratigraphy and location than ever before. Physical 3-D models that can be pieced together or pulled a part. High resolution, detailed, and accurate data that can be manipulated, viewed, and analyzed virtually any way we desire. Even better we can share our results in a brand new ways: 3D printing, virtual tours, etc., we could and can literally print pieces of art, artifacts, even a scale model of the canyon if we wanted to! This project, this technique isn’t just for the archaeologists and researchers understand the shelters better, our goal is to be able to help everyone understand the shelters better because the shelters are a part of all our histories.
To elaborate on what Charles has said numerous times and will likely say many more, “50 years ago they were using completely different techniques..50 years from now they’ll be using completely different techniques…but right now we have the technology and the opportunity to set those standards for the next 50 years. We are doing something amazing here with SfM, and sure, we’re not the only people using this method, but there could be a dang lot more of us using it.”
If you haven’t already, you should click on over to our older posts on the subject, and I highly encourage you to visit the Mark Willis Blog http://palentier.blogspot.com/, where you can see some of the other extraordinary uses of SfM 3-D modeling.