Recent Posts



Not All Roads Lead to Rome

A chance look at the right part of a LIDAR map led to a two-part KAP session, over 500 multispectral kite aerial photos, hours of image processing, NDVI calculations and contrast enhancing, poring over maps plotting directions and trying to disentangle everything - and when it finally started to make sense a real archaeologist deflated us with a laconic "probably not Roman" ... Well - probably my ass. Here is the story. It all began with this:

From a beautiful and free online map repository called Atlas okolja - the Atlas of Environment -. by Slovenian Environment Agency ARSO. You can find the LIDAR map of the whole territory of Slovenia by clicking the "Podlage" (base maps) drop-down menu on the upper left.

This is a LIDAR map of a part of Ljubljana Marshes north of the village of Ig. River Ižica meanders on the right, two roads and two large channels run through in a roughly north-south direction, and an intricate pattern of drainage canals cross this typical marsh landscape. But what caught our eye are two roughly parallel tracks going from lower left to upper right, disappearing in the boggy soil.

The western track as seen from the ground, standing on it looking southwest towards Ig. So inconspicuous it's hard to take a clear and informative photo of it. Those tracks are pretty unremarkable when you see them from the ground level - a nondescript lump, a couple of meters across and maybe 30 cm high. But it's interesting how they cross roads and drainage canals without paying much attention to them .... so the tracks were here first and are the oldest feature in the landscape. Does that mean they are ancient Roman roads?

Well, people started to make inroads into the wet heart of Ljubljana Marshes only in the late 18th century. The big road - Ižanka - that connects Ig with Ljubljana was constructed in 1848, and the drainage canals probably not long before that, so the tracks being "older" than them does not necessarily mean they are "ancient". But when our eyes see a thing like this, our brains go "Archaeology! Romans!" ...

The thing is Ljubljana Marshes are a treasure trove of archaeology. From the Mesolithic period onward people were camping and hunting here, their evolving culture culminating in famous Neolithic and Bronze Age Pile dwellings that are today a UNESCO World Heritage site. The oldest wooden wheel was found here, the Celtic Taurisci were guarding the approaches from the Mediterranean, the Romans were establishing vici and emporia and building roads between them and maybe even moved a river closer to a marble quarry ... there are so many sites and finds - and there are many many more just waiting to be unearthed.

Map of known archaeological sites of Ljubljana Marshes.


One well-known Roman road - the first ever that crossed the treacherous Marshes - is still visible between Lavrica and Ig, and it pretty much looks like these tracks, just a bit wider and higher. We were there last year and we did some multispectral kite aerial photography to calibrate the camera and see if we could "see" the road using a method called Normalized Difference Vegetation Index (NDVI) mapping. It's based on a pretty cool idea that plants grow poorly if they grow over some buried remains like walls and roads, and grow stronger if they find themselves enjoying a deep ditch full of soil beneath them. One can enhance these subtle differences by taking a multispectral photograph - visible and near-infrared - and calculate the NDVI index: healthy plants reflect a bit more near-infrared light, withering ones a bit less, and by mapping the differences one can bring to light what lies hidden beneath in glorious chaos of psychedelic colours. Like this one:

A NDVI map of a part of the known Roman road near Ig - see how the plants reflect less NIR light when they grow on the road where the soil is thinner?


Looking at "our" tracks we wondered why on Earth would anyone build two nearly parallel roads through an infernally difficult terrain where the Devil himself - as the stories from the Marshes tell - devours overnight everything that was built during the day? After the Romans, no road dared to cross the Marshes until 1848! Constructing one road here is half crazy - but two?


The other pressing question was where do these roads lead - as the extant structures simply and quietly disappear into the marsh. The other Roman road was leading from the vicus of Ig across the bog to Babna gorica hill where an outcrop of limestone offered the builders some respite, and on to Lavrica on the main road from Emona (Ljubljana) to Neviodunum (Drnovo) and Siscia (Sisak). But these two?

Extrapolating from their directions just before they disappear it seemed the roads aimed at Grmez hill, another solitary outcrop in the Marshes. Were the builders looking for an anchor on the hard ground - as they did building the other road - before soldiering on to the main road at the edge of the Marshes?

Were the roads going to Grmez hill (top right)?

The last question we were cocky enough to pose is where exactly did the roads cross Ižica river. Bridges are usually sturdy constructions and their foundations often stand up to the ravages of time for much longer than the roads, Plus, river crossings were important places in the Roman world; they were very inclined to throw votive offerings to the gods of rivers, for good luck and safe travel and whatnot, and many of those offerings now tell us stories about the world passed long ago. Wouldn't it be nice to have more stories from that time? It was time to fly a kite and a camera!

Kite aerial video, shot just for fun with a regular action camera.


With our heads full and our excitement growing, we packed the gear and went on site. A strong breeze compelled us to use the Box Delta kite, and in no time the AgroCam NDVI multispectral camera was aloft, taking photos that were to reveal all the mysteries; the roads, the bridges, the votive treasures, and all.

Did a Roman bridge once span Ižica river somewhere around here? Uncalibrated and unprocessed kite aerial photo, AgroCam NDVI on a Royal 69 sled kite. See us holding the kite line on the right!

Almost immediately after coming back and unloading some 200 photos onto the computer, our first assumption was shattered. The thing is the Marshes are a vibrant place, literally so - everything is subsiding and growing and faulting and uplifting, and a river crossing them is no exception. While Ižica is a Karst river, springing in its almost full form from underneath a rock in the center of Ig and therefore carries very little abrasive sediment that could erode its banks, it still meanders and floods and is prone to changing course. So when - and if - the Romans were building their roads around here, the river probably wasn't flowing where it flows now! Of course it wasn't.

A meander in the middle of a meadow.

Photos revealed a meandering path crossing the marsh, hiding under the straight drainage canals. This is the old, long-abandoned course of Ižica, about 1.500 m long 20-30 m wide and almost unnoticeable in the wet meadows.

Old Ižica (marked in blue) crossing the meadows, and new Ižica on the right.


Now we had a massive 3GB+ folder of multispectral photos, but most of them were shot above a wrong place ... partly because the river-crossing was not where we supposed it to be, partly because who would walk for hours around a soggy marsh with a crazy kite trying to crush your fingers and tear your hand off, and partly because of the wind's strength (way too much and growing) and direction (less-and-less-than-ideal). We needed another KAP session.

A couple of days later we were back; the wind was quite nice (if gusty) and we even had time to fly other kites before we went to work ;-)

A kite in action caught on a multispectral photo by another kite.

This time it all went well. We got the part where the roads disappear, we got the old river course where the bridges should be, we got hundreds of photos. and even enjoyed a couple of beers, Then it was processing time!

One of the tracks ("Road A"; vertical line on the right) disappears into the marsh.

One thing with kite aerial photos is that no matter how well the Picavet rig is constructed and attached to the kite line, the camera sways in the wind. The other is that the kite is stable alright, but the wind strength varies and so does the flight level. Different perspective and camera distances conspire to make stitching the photos together a true nightmare. After much coaxing we managed to persuade Photoshop to produce this - the main stitched image, showing most of the 'area of interest':

The area of interest. The old meandering river course is clearly visible, the point where Road A disappears is circled in red.

So, where are the roads hiding? We found a great online tool for processing infrared photos called Infragram, but when we put the image through their NDVI processor, we got this:

NDVI map produced by Infragram.

See anything? No? Well, pareidolia does help sometimes, and there was a hint of a ghost of a straight line. Actually of two of them: