by Sarah Parcak
Breaking into the Map Room
I had my first look at Tanis from space in 2010, a decade after I had visited it for the first time as an undergraduate. I remember being impressed at the site’s size and the number of statues out in the open, reused from the time of Ramesses II, but that’s all. In other words, I did not expect to find much; maybe I’d find additional rooms in one of the temples.
At that time, high-resolution satellite imagery had not come down in price. The cost-prohibitive nature of the data meant few archaeologists had applied it to locating smaller features beneath the surface of sites. I got a great deal on DigitalGlobe data as an educational user, and that made all the difference.
However, although DigitalGlobe’s database held many images, the company did not have the range and number of satellites in orbit that it does today. I couldn’t be picky about my data, so I just had to use what there was.
Work I had done previously using lower-resolution data suggested that imagery taken during the winter months allowed easier detection of entire sites. I thought the same theory would apply to finding features as well, and I got lucky. Two images appeared in the database taken in January 2010 from the WorldView-1 and WorldView-2 satellites. The first usefully had a .5-meter resolution in the panchromatic range—which, counterintuitively, is black and white—while the second was multispectral, with eight bands going into the near-infrared range, at a lower resolution of 1.84 meters. The multispectral data could show more, but only if the feature was bigger than a double bed.
I downloaded the data in my university lab office, and it took forever. While waiting, I opened Tanis maps in my archaeology books, which I hadn’t done in years. Right away, the large blank areas in the central and southern parts of the site jumped out. The temples in the north overwhelmed the site. If you did not know better, you’d say the site was a sacred space for worship and nothing else.
When the data was finally good to go, I loaded the multispectral WorldView-2 image into my main processing software, ER Mapper, and opened up the site and the fields surrounding it. As I zoomed in, the large Amun-Re temple walls showed up, a bit blurred, with limestone blocks in the center glowing a bright white. I scrolled south, toward a known Horus temple.
About 100 meters south of its outline, I could make out a cluster of fuzzy lines, suggestive of subsurface architecture. Playing around with the data, trying different combinations of the light spectrum bands, showed a little bit more detail, but the image remained mostly ambiguous. For me, though, this was already a success. The data had shown some potentially buried walls across a 600-by-800-meter area I knew, from on-the-ground experience, to be unremarkable-looking brown silt.
Next up, the WorldView-1 JPEG image. Instead of manipulating the black-and-white data, I decided to use a technique called “pansharpening.”15 It may sound like a death-match cooking competition, but here’s how it works: 1.84-meter-resolution data will not pick up smaller features, but you can supercharge it by merging it with higher-resolution panchromatic data. The result is higher-resolution multispectral data. I know, it sounds like magic. It kind of is.
Think of it this way: a lower-resolution color image shows the vegetation in an agricultural field, while a higher-resolution black-and-white image shows the internal agricultural field divisions. By merging them, you get the vegetation information at the same time and with the same resolution as the important field subdivision data. Lower and higher data sets do not even need to match perfectly in terms of geographic area: areas that match will be pansharpened automatically.
I waited while the imagery merged, not expecting much, then zoomed in on the new imagery, starting in the north. Temple walls appeared with slightly more clarity, but the limestone still glowed brightly. Satellites take these images in the late morning, so you cannot do much about bright stone reflecting light.
Tanis Strips Off
I scrolled down. And then I almost fell off my seat. I thought I was hallucinating: an entire ancient city leapt off the screen. Ambiguous, faint streaks that had appeared in the multispectral image now emerged as clear buildings, streets, suburbs … everything.
If you get one discovery like this in your lifetime as an archaeologist, you have led a blessed life indeed.
But that wasn’t all the tricks up my sleeve. After the pansharpening came the fiddling, like tweaking the radio to get the best possible signal. You can take so many different approaches, it is overwhelming. But armed with the season of the imagery, its resolution, the geology and soil type of the site, and the materials and size of the structures you seek, you can narrow things down.
A thousand data points affect what buttons you push. Some techniques enhance the subtle differences in the brightness of pixels next to one another, and others broaden the total range of values a pixel can have.16 Also, what works well in one part of your image might not work well across the entire thing.
After trying dozens of different processing techniques during the afternoon and into the evening, I ended up with a crisp image showing the outline of the main city of Tanis, almost like the map room in the Raiders of the Lost Ark scene, with only a little more imagination needed. My headpiece of Ra was a multispectral satellite, and my expression was the same as Indy’s, wonder that a great ancient secret had revealed itself to a patient archaeologist.
Today, looking at Google Earth, anyone can see outlines of the structures in central and south Tanis on satellite images, which currently have a resolution of .3 meter. That’s smaller than your average laptop. But in 2010, that was not the case (a resolution of .5 meter, remember), which shows how dramatically the technology has developed.
At home, when I went to show the images to Greg, I just about dropped my computer as I took it out of my bag with shaking hands.
“What’s going on?” Greg asked. I opened the laptop and pulled up the imagery.
He stared. And then said, “What is this?”
“What do you think? It’s Tanis! All of it!”
He grabbed my mouse and scrolled around the processed satellite image, zooming in and out. I pointed out all the details and their implications for Egyptology, at which point he informed me that he had worked in Egypt since 1988. Sigh.
We discussed the next steps for how to proceed. First, we had to digitize the data, drawing every individual building on the computer to see them more clearly. I thought I could draw the entire map of the city, which measured 800 square meters, using my mouse and the line tools in a program called ArcGIS, which allows the user to store layers of maps and connected information like census data.
Well, I tried. And failed miserably. The lines appeared straight when I zoomed out, but when I zoomed in and looked close up, they had blobby edges of differing thicknesses. After six hours of everything going wrong, I gave up. There had to be an easier way to capture the varying details.
Greg had the brilliantly simple idea of drawing the town plan by hand, the old-fashioned way. We would print out a massive poster of the satellite imagery for the entire central city, and then cover it with transparent plastic sheeting, to draw in every detail in pen. It’s one of the ways Egyptologists record scenes from blocks or temple walls. I took the image to a print shop and asked them to expand it as much as possible, on one piece of paper. I got looks.
The final map measured 2 meters by 1 meter and covered our entire dining room table. It took us more than two months to draw it all, a little at a time. Greg let me fill in the wall lines. He’s not sexist—just protective of his artwork. As you might have gathered from my ArcGIS wobble, I manipulate technology, but I can’t draw.
We estimate it took more than 50 hours of work during those eight weeks. After discussing the tiny details and making sure we drew the clear features and dotted the ambiguous ones, we could do no more. It gave us a far deeper insight into the distinct buildings and the three phases of the city’s occupation than we’d seen in publications before.
Better Than Ground Survey—or at Least C
heaper
For fun, Greg and I decided to compare the efficiency of satellite imagery versus ground surveys. We chose magnetometry as an example; that’s the same subsurface mapping technology we applied in Newfoundland. A good magnetometry surveyor, with an assistant, can map 80 square meters in a normal workday, if the site is flat and there’s no vegetation to get in the way. For every five days of mapping, specialists need a day to process their survey data. At Tanis, the central city covers an area of 640,000 square meters, and there’s an additional area of 20,000 square meters to the south with clear structures, for a total of 660,000 square meters.
So that’s 103 days to survey all that. Assuming a standard month of work—longer periods are generally not possible for technical specialists, since they often have other job commitments—103 days is five full seasons. With an average cost of about $1,000 per day for a magnetometry survey, including the expert team’s airfare, room, board, and in-country travel, surveying the entire settlement of Tanis carries a $200,000 price tag.
This assumes that you have permission from the Egyptian government to conduct the survey, cooperative customs agents for importing the necessary equipment, functional equipment once it arrives, and survey specialists who manage to stay healthy. None of these is ever guaranteed.
Compare that to $2,000 for the satellite imagery, and only, say, 60 hours total of our time. It was an enthusiastic thumbs-up for the satellite data, even if magnetometer results might have had a bit more detail. But oh, the things we could see on the dining room table from 600 miles in space.
In the distinct occupation phases of Tanis, there were roads, the full economic range of houses, and large administrative complexes. Some structures have to have been elite homes or palaces, bigger and more elaborate than other buildings at the site. And there appear to be at least three of them.
The satellite map allows us to travel to the past, and for the first time, to really understand how the city might have functioned. Its implications are wide ranging, primarily because Tanis represents one of the largest and most well-known capital cities in antiquity.
Digging Down
When we find things from space, we must always ground-truth them, either via excavation or survey. Even when the maps looked great, and “showed” things clearly, we had to verify if what we saw on the surface of Tanis represented what lay beneath. Thanks to a collaboration with the French team that worked at Tanis through 2014, led by archaeologist Philippe Brissaud,17 we had a chance to put the data to the test.
I got in touch with Philippe right away. To ground-truth, Philippe would have to apply to the Egyptian government for permission to do a test excavation in the south, though having never used satellite imagery before, he was skeptical about our results.
I arrived on-site during the full swing of the fall excavation season. History hangs heavy at Tanis, inside and out. The French dig house goes back almost a hundred years, with photographs of every archaeological team that has ever worked there on the walls. Philippe, an enthusiastic and gregarious dig director, welcomed me with proper French hospitality. Delicious as lunch was, I did not, however, partake from the bowl of chicken feet on offer, especially after he shoved one of them in my face, to peals of laughter from his team.
Philippe gave me a fantastic tour, showing me ongoing work in the temple of Mut, where they had uncovered a series of gorgeous carved blocks. My favorite bit was visiting the storeroom on-site, full of intact and reconstructed pottery vessels from the last 100 years of French excavations. The pots told the history of excavation at the site as much as the site’s history—had Montet himself held some of them in his hands?
In the central area, where a clear 20-by-20-meter house had appeared just south of the Horus temple, the team had worked for several days to uncover walls up to 2 meters thick. A smaller room appeared in the middle of the house, only a meter wide—a storage area perhaps—which the satellite imagery had not picked up. Curious. I asked Philippe for his impression. He looked at me with a big smile.
“It works! With 80 percent accuracy, I think. The imagery just did not pick up the smaller rooms, and there is a 20-to-30-centimeter offset between the edge of a wall and its true corner.”
The team had dug down about a meter into each room, finding different phases of house construction, some of which matched the other building phases around the house that we had seen on our dining room table.
We also examined the edges of the excavation unit and the depth of the surface silt, and tried to determine what quality had made the buildings pop so clearly. At Tanis, the silt covering the site is slightly sandier than at most other settlement mounds in the Delta. That sandy soil contrasts with the crumbling mud brick of the building foundations, and the mud bricks gain an even more distinct color when they absorb water during the winter rainy season. Although I am aware that what worked so well here may not give the same results at other sites, I felt much more confident after that visit about the use of satellite imagery in general, and about beginning to reconstruct this ancient city in particular.
Digging at Tanis with Philippe Brissaud’s team [PHOTO BY THE AUTHOR]
Daily Life at Tanis
People of every kind lived at Tanis: kings and queens, priests, administrators, artisans, architects, soldiers, and a large working class to support the temples, palaces, and main town.18 Like every major city today, it had a bustling central section. The Tanitic branch of the Nile curved around the northeast part of the site, facilitating the transportation of carved blocks for the temples and artisan workshops. Given that geography, there must have been more than one harbor area,19 with marketplaces dotting the riverside. There, ship captains and traders hawked their goods from across the empire, from Israel in the north to Nubia in the south. We know of similar marketplaces at Luxor, the southern city associated with the Valley of the Kings.20
The priests lived in housing located in the northern section, close to the temples.21 The temple’s size suggests a staff of hundreds, from high priests and their assistants to the cleaners. People brought food and offerings into the temples all day; and during festivals, thousands of Tanites would have crowded the exterior court of the temple enclosure, hoping for a glimpse of the king or some good fortune from the gods.22
Moving just south of the Horus temple, in the central part of the site, we found an area full of homes, each measuring 20 square meters, lined up along streets. The houses seem to have four to eight internal rooms, with at least one larger in size than all the others. They seem similar to the large homes of royal officials at Amarna, which was Egypt’s capital 300 years earlier.
If you had business in the home of a Tanite official, you were ushered into a central room or the public area of the building. Alongside the official, scribes were there to prepare any necessary letters following your conversation. The walls were probably whitewashed and painted, with stone or wooden columns to hold up the ceiling.23 Servants awaited orders, ready to pour imported wine from Palestine to quench your thirst.24
The back of the house was private and included the kitchen, bedrooms, and even a separate bathroom area25—not bad, for 3,000 years ago. As at Amarna, these houses may have had upper stories. Best of all, given the central location of the homes, the owners had a five-minute stroll up the street to get to the temples and administrative buildings, or a two-minute hop to the palaces. The Nile brought cooling breezes to their homes, giving this part of Tanis a more pleasant smell than the poorer neighborhoods. It was prime real estate, if you could afford it.
Just south of this desirable neighborhood were the 20-to-30-room villas and palaces occupied by the elite and two dynasties of Egypt’s kings and queens.26 Nearby, artisans toiled away in their workshops, producing stunning jewelry and fine objects,27 and chefs prepared delicacies with spices and goods from across the Mediterranean, always ready for a feast.28 Emissaries from foreign lands waited for the city’s top officials, hoping for the king’s ear on pressing diplomatic matters. The
quarters of the royal family, located away from the prying eyes of the court, housed the queens and royal family.
In the throne room, maybe with a beautiful painted floor depicting birds and wildlife along the Nile and symbols of Egypt’s dominion over foreign lands, the king held court, surrounded by his vizier, head of the treasury, chief general, and countless scribes.29 Even in Egypt’s last Intermediate Period, the pharaoh lived as a god on Earth as he attempted to balance the forces of the universe in Egypt’s favor.30
In the spit of land forming the tail of the mound, the southernmost part of the site, we can see several dozen 8-by-8-meter homes, built next to one another in more organic fashion. The area immediately appears less affluent.31 Each house seems to have one or two rooms, so perhaps families lived in adjacent one-room dwellings. These could have been homes for the people who worked in the palaces. Without excavation we cannot tell, but we can fairly confidently say that this area was the poorest zip code of Tanis, farthest removed from all the amenities on the doorstep of the wealthier district. And it does appear to be a distinct neighborhood, aligning with the second phase of construction observable in the satellite data.
Reconstructing the Landscape and Population
Figuring out the population of this bustling community based on the satellite imagery is tricky. We know that major cities like Alexandria had an estimated population of up to 500,000 some 700 years later,32 and these numbers could be a clue. Here at Tanis we can see dozens of houses, and we have the total area of the central city. The Delta has experienced great archaeological loss due to the growth of modern cities and the mining out of soil by farmers, so we have to wonder just how big the city was.