Kite Science: Assessing Water Pollution from the Sky

(This article is derived from a European Researchers' Night 2021 lecture. Theoretical parts and fundamental research by Maša Mramor, Gimnazija Vič; kite flying, kite aerial photography, and photo processing by KAP Jasa kite society.)

As we have said many times before, kites are spectacular, affordable and versatile aerial science platforms. A large Rokkaku can easily lift a couple of kilograms of scientific equipment, it can fly for hours, and flying kites is less restricted than flying other aerial devices. We have employed kites for science before, mostly for archaeological research - discovering a Roman villa rustica, surveying medieval field partitions and the ridge-and-burrow remains at Castle Kalc while searching for an Iron Age settlement, researching the prehistoric landscape near Divača, finding a "Roman" road that doesn't lead to Rome., and exploring a real Roman road - using our two infrared cameras, a modified Pentax WG-10 that was gifted to us by the inimitable Dr Wells, and an AgroCam Geo NDVI.


We were invited to participate in the largest scientific event of the year, the European Researchers' Night 2021 in Ljubljana (we did ERN twice before, in Banja Luka). This year's ERN theme in Slovenia was water, and while we have a ton of very pretty kite aerial photos of pretty lakes, dreamy rivers, mysterious swamps and the like, we felt showing pretty pictures is not sciency enough - we've got to do some real stuff.

We were thinking - can we use our near-infrared cameras in the sky and the power of the NDVI index to see how much (and why) a body of water is polluted?


The NDVI index is a powerful tool used mainly to assess the health of plants. Healthy plants absorb most of blue and red parts of the sunlight (well, the chlorophyll in their chloroplasts does), and reflect most of green and near-infrared. By calculating the ratio of near-infrared (that is mostly reflected) and red light (that is mostly absorbed) we can see how well a plant is doing. In healthy plants the NDVI index approaches +1, stressed or diseased plants reflect less NIR light so the NDVI value drops towards zero, and dead plants reflect almost none of the incoming NIR light, so the index drops down to -1.

By taking a multispectral photo of say a field of crops one can make a nice NDVI map out of it, calculating the index for every pixel and representing the values with nice colours; red for +1, then yellow and green and cyan and deep blue down to -1, thus revealing places where the soil is poor, or where some pest started causing mayhem among the plants, or possibly the buried remains of a Roman villa.

The differences in plants' NIR reflectance are extremely subtle - can't see them with your eyes; the NDVI enhances them and makes them visible. For example, you can't see anything in these two photos of a wheat field near Kašelj - NIR on the left, multispectral on the right. But make a NDVI map from them and ... look!

The buried remains of a Roman villa rustica are clearly visible.


Photosynthesising plants living in water are mostly algae and cyanobacteria, and they too do this thing reflecting near-infrared light according to their health, enabling us to do the NDVI map of a body of water. A particular NDVI value can show how well the algae are doing, and in turn can show what kind of pollutants mess with their well-being.

Heavy metals - cadmium, mercury, lead, chromium and the like - wreak havoc upon them, so if a river is getting polluted by them the NDVI values will decrease, and keep decreasing downstream. On the other hand, nitrogen compounds and phosphorus are pollutants as well, but the algae love them, so they thrive if the influx of nitrogen compounds increases - this is the cause of algae and cyanobacterial blooms in lakes and the sea. In this case, the NDVI values increase when these pollutants are present in the water. The third type of water pollution that can be detected with a NDVI map is mechanical, usually associated with illegal removing of river or lake bed, for sand and pebbles used in construction. The disturbed, muddy water reflects less near-infrared light and can also be made visible with the NDVI index.


Armed with a kite and a near-infrared camera we headed to Ljubljana Marshes landscape park to test a hypothesis. We were looking at the River Ljubljanica and chose two points to take NIR kite aerial photos of - one near the village of Bevke, just before the confluence with Borovniščica stream, and the other downstream of the town of Podpeč.

The hypothesis was simple: Borovniščica runs through an agricultural land and by a couple of cattle ranches, while in Podpeč and the villages near it there are mostly individual houses with no connection to a sewer system - therefore the influx of nitrogen (fertilizers and cattle ranches are notorious nitrogen polluters, and even the best septic tanks do leak) into Ljubljanica should be substantial, and as the algae love additional nitrogen they should thrive more at point 2 (downriver of Podpeč) than at point 1 (near Bevke). The NDVI index value at point 1 should therefore be lower than at point 2.

Multispectral image of Ljubljanica near Podpeč; AgroCam Goe NDVI on The Spark Rokkaku kite

We lifted the camera at point 1, got the multispectral photo, cut a part of it showing the water surface, averaged the colour values so things like floating detritus and shadows wouldn't mess with calculations, and computed the NDVI index.

Note: the NDVI index is an index. It doesn't show absolute values of pollutants in the water. It could - if the camera could be totally calibrated and all the photos could be taken at exactly the same condition (time of day, cloudiness, wind disturbing the water surface etc.). The NDVI index shows relative differences in pollution of two or more spots, nothing more.

We repeated the proces at point 2 - fly the camera, take the picture, cut, average, do the NDVI - and we got this:

See the difference? The NDVI value is higher downstream, where Ljubljanica got the stuff from cattle ranches and villages. But is the higher NDVI value at Podpeč really due to algae enjoying themselves in all the excess nitrogen? Are the nitrogen compounds from fertilizers and cattle and humans responsible for this?


Luckily the House of Experiments of Ljubjana, a co-organiser of the European Researchers' Night 2021, gave us a "water research package" with a couple of water pollution measuring strips, so we could do a control test for our kite flying NDVI experiment. These strips have some little active squares on them that are sensitive to different pollutants and water properties, and change colour accordingly (similar to the pH measuring litmus test).

We took water samples from both locations - it was not easy, as the banks of Ljubljanica are muddy and steep and guarded by thorny bushes - and tested them. Here are the results:

It's hard to take a clear photo of a tiny pollution measuring strip, but trust us - the differences in water quality are noticeable. From Bevke to downstream of Podpeč Ljubljanica gets a bit more alkaline, its pH value increases. The quantity of heavy metals in the river is low and pretty much the same at both locations. But the quantity of nitrates and nitrites - them nitrogen compounds that algae love so much - increases noticeably!. So there is an influx of nitrogen between Bevke and Podpeč, and agriculture (fertilizers and cattle) is definitely responsible for that.

But what about the individual septic tanks in villages? One element gave them up: chlorine. As chlorine is the active ingredient in various septic tank cleaning compounds, its presence is telling: septic tanks leak into the water table and into the river (or people are dumping the contents directly into the environment), thus contributing to the excess nitrogen influx and the growth of algae in Ljubljanica.


Encouraged by these results we headed off to Zalog in the eastern part of Ljubljana basin. If a little town like Podpeč and a couple of cattle ranches visibly (for our near-infrared flying camera) pollutes the river, how much and what kind of pollutants a large city like Ljubljana throws into Ljubljanica.

Ljubljana has a supermodern Central Water Waste Treatment Plant and in general the city's waste management is very very good. But such a populous town still pollutes - people throw stuff into the river (from bikes to batteries and everything in between), vehicle exhaust settles on the ground and gets flushed into Ljubljanica, and a few parts of Ljubljana aren't yet connected to the central sewage system.

Lifting a kite near Zalog proved tricky - the hills just beyond Ljubljanica happily made some heavy turbulence for us, and huge trees on the banks made it impossible to see what are we going to shoot - so we flew an ordinary camera first. This is what we saw, and it made us happy:

Ljubljanica - coming from the right - does a turn here and goes over some small rapids. But the interesting thing is that stream flowing into Ljubljanica from the top: it's a braided stream, a dead arm of the river. This coincidence made it possible for us to test the ability of the NDVI index in assessing change in water pollution on a very small scale. Rotting detritus (and whatever people are throwing into the thick bushes there) in this anabranch is leaching into the river, algae and (cyano)bacteria love it and thrive.

We lifted our near-infrared camera ...

... and did the NDVI magic on two points on the river, one just before the rapids, and the other after the confluence with the dead arm.:

Not only there is a substantial difference in levels of pollution compared with those on Ljubljana Marshes - the river is visibly more polluted after the confluence, and it changes in jus twenty meters downstream! This is important: it shows it is possible to use a kite and a near-infrared camera to discover an acute, topical water pollution. One can discover an exact point of entry of pollutants; usually a farmer - or a group of criminals - dump slurry or some toxic stuff into the river in the cover of the night. This is especially useful when considering a karst river like Krka - Krka is the only surface river in Suha krajina; most of its tributaries flow underground and empty into it underwater, so it is hard to see where exactly is the source of the pollution.


River Sava - the largest river in Slovenia - flows just nearby, so when we did Ljubljanica, we headed a hundred meters north to do Sava too. We lifted the camera ...

... and did the NDVI - which surprised us quite a bit:

The value of the NDVI index of Sava is lower than that of Ljubljanica. Does it mean Sava - flowing through towns and industrial sites for miles and miles - is less dirty than Ljubljanica?

Well, it's possible. But there are other, clearer reasons for this unexpected result. One is that since Sava is much larger, deeper, and faster than Ljubljanica, our camera simply could not detect the algae at the bottom. The other possibility is - heavy metal.

Sava flows through Upper Carniola, the industrial heart of Slovenia. From iron smelters and steelworks of Jesenice to chemical plants of Kranj all the toxic stuff used to be dumped straight into the river. The industry - the part that survived the encounter with capitalism - is now very clean, but the pollutants of yesteryear are still lurking in the deep: hidden in the mud at the bottom of hydroelectric plants.

Medvode hydroelectric plant was built over 50 years ago, and all this time heavy metals from industrial processes accumulated in the mud behind the dam, from where they slowly leach downstream, poisoning the life in the river. So the low NDVI value of Sava near Zalog - some 20 km downstream from the plant - could be a result of heavy metals impeding the growth of algae. (Incidentally, we are acutely aware of the ticking ecological bomb behind the hydroelectric dams; the cleaning of that is a huge undertaking, as one can't just dig the mud up, since the disturbed toxic stuff would go over the dam and kill everything in its path.)

But this is another type of pollution a kite with a near-infrared camera can detect: is the value of the NDVI index is surprisingly low - or it even decreases downstream - the pollutants are most likely heavy metals like cadmium, mercury, chromium and lead.


The last data collecting flight was over the river Kolpa. Kolpa is with Soča, Idrijca, upper Sava, and upper Savinja rivers one of the cleanest in Slovenia. It springs up deep in the uninhabited forests of Gorski kotar in Croatia, from where it flows in a canyon up to the town of Vinica, making the border between Croatia and Slovenia. There are few settlements by the Kolpa, the agriculture on its banks is negligible and industy nonexistent, so the river is really clean. A perfect one to test our NDVI stuff, as we already imagined the result: a deep blue, negative value of NDVI approaching -1.0 of a crystal clean water.

Well, it wasn't to be so easy. As we lifted the kite and the camera and were just about to steer it over the water, completely oblivious that the river is actually a heavily guarded Schengen border, we heard a distinct noise of a helicopter rotor coming in behind us.

The chopper was flying low over the Slovenian bank of the river, less than fifty meters above ground. Way lower than the kite and the camera - and heading straight into the kite line. Suddenly we were faced with a potential disaster - the rotor blades cutting the line and the kite and the camera lost - and a potential catastrophe: the kite line tangled into the rotor and the helicopter crashing. Sure, we like to be in the news, but not in that way ...

Luckily, the pilot noticed something in front of the chopper and turned away, the blades missing the kite line by 10 meters. When we were able to breathe again we quickly pulled the kite down and left the place - happy that nothing had happened.

We thought the near-infrared kite aerial session was lost, but in all the commotion the kite managed to take one photo of Kolpa:

Enough to do the NDVI!

We confirmed what we and hundreds of happy swimmers already knew: Kolpa is a very clean river. Safe to bathe in, and even if one gulps a bit of it, it's ok.


We - well, our kite! - did quite a bit here. We saw how agriculture - fertilizers and cattle - and sewage tanks increase nitrogen levels that cause algal and cyanobacterial blooms. We saw what a town or a city does to a river, and how a small dead arm of a river can visibly raise the levels of pollution on a very short distance. We were surprised by river Sava and its unusually low NDVI values, and we confirmed the status of Kolpa as one of the cleanest rivers in Slovenia.

A near-infrared camera on a kite is a powerful scientific tool. It is cheap, easy to operate, clean, quiet, versatile and useful. With a bit of math it can be used to assess the level of water pollution, to discern different types of pollutants, to discover the source of an acute pollution, to check differences in pollution levels over the course of a river or between different bodies of water, and to monitor water pollution seasonally or through the years.

Kite science, yeah!

Big thanks to Damjana of the House of Experiments who invited us to participate in European Researchers' Night, to Mojca and all the staff of Library of Šiška for kindly letting us use their conference room to deliver the lecture, to Irena and Igor who hosted us in Vinica, to Dr. John Wells who generously gifted us our first near-infrared camera, to Maša who did the theoretical part of this lecture and proved herself as a superb co-presenter, to Viktor who expertly piloted the Rokkaku kite over the field in Kašelj, discovering the Roman villa rustica in the process, - and to all of us having fun doing kite stuff!

Kite aerial photos shot with Canon A810 (visual), near-infrared modified Pentax WG-10 (720 nm NIR), and AgroCam Geo NDVI (multispectral) on The Spark Rokkaku.