Dr. David C. Thomas is an honorary research associate with the Department of Archaeology and History at La Trobe University in Victoria, Australia. Between 2003 to 2007, Thomas assessed sites at the early Islamic capital of Jam in Afghanistan. After security concerns prevented fieldwork, he began remote image analysis via Google Earth, joining a field that’s expanded rapidly since 2011.

Thomas helped uncover over 850 new archaeological sites in the country and is the author of The Ebb and Flow of the Ghurid Empire published by Sydney University Press that covers the ancient landscapes, rise, expansion and fall of the Eastern Iranian dynasty over 800 years ago.

When did you first get involved with remote sensing analysis?

I started using satellite imagery of the UNESCO World Heritage site of Jam, in central Afghanistan in 2005 – I bought an expensive, high resolution (1 pixel = 60cm on the ground) image because I wanted to build on our fieldwork at Jam which was documenting the extent of looting at the site. Jam is thought to be the summer capital of the 12^th-13^th century Ghurid dynasty and is most famous for its magnificent ~65m tall minaret. The harsh mountainous terrain limited the area we could survey on the ground. By studying the satellite imagery and applying what we knew from fieldwork, I mapped how much of the site had been looted.

After we were unable to return to Jam in 2006, I started using satellite imagery available through Google Earth. The big advantage of Google Earth is that it’s free, so I was able to study other known sites from the early Islamic period across Afghanistan and to explore areas such as the Registan Desert which had not previously been surveyed. We discovered 850 new archaeological sites in the Registan Desert which nobody knew about previously, nor imagined would be there. 

How has remote sensing analysis changed how sites in conflict zones are monitored?

Remote sensing analysis has totally changed how archaeologists monitor sites in conflict zones. Most importantly, it’s safe. Sites can be studied systematically over years, without needing to visit the conflict zone. Remote sensing analysis also integrates well with digital mapping and allows you to create a base-line against which to measure looting – not just in the future, but also looking backwards. If you have available imagery, you can study when looting starts, how it expands and when it stops. If you understand the process, you have a better chance of stopping or reducing it.

How has remote sensing shaped cultural heritage management in countries like Afghanistan?

It already has, both through the work I have done over the past decade, and others.

In 2008, for example, I presented a poster about the 10^th-12^th Ghaznavid capital of Bust / Lashkar-i Bazar at the World Archaeological Congress, which illustrated how satellite imagery could add to the discoveries of French archaeologists at the site from 1949-1952. More recent satellite images show how badly urban expansion and agriculture have encroached upon the archaeological remains.

Another study of mine focussed on the Ghaznavid capital of Ghazni, following its nomination as ‘Cultural City of the Islamic Civilisation’ for 2013. I used satellite imagery to revisit and re-record known sites. I also discovered 31 new sites around the Ghazni, including a possible mausoleum and large garden, and in another study I used satellite imagery to help locate the ‘lost’ minaret of Sakhar, based on information from local heritage sources.

Other, more sophisticated research is being carried out by the Afghan Heritage Mapping Partnership at the Oriental Institute in Chicago. Emily Hammer and her colleagues are about to publish a major study of 1000 archaeological sites using a variety of satellite imagery, some stretching back to the 1960s. They are using these to challenge some of the myths and misconceptions about when the looting started and who is doing it. They have also used artificial intelligence to teach computers how to recognise robber holes, and quantify the amount of looting that has taken place at the Hellenistic city of Ai Khanoum.

What goes into the analysis of satellite imagery—mainly —what’s the timeline on evaluating images and how frequent images are provided?

It varies a lot. The analysis is very time consuming – you have to be methodical and focused. Your eyes get tired very quickly! We estimated that it took 10 hours to survey one of the 17-km square (275 km²) study areas for the Registan desert. Then, the places marked have to be catalogued and cross-checked – about 60% of the original placemarks were discarded at this stage as being natural features or too ephemeral.

The actual images themselves are collected quite regularly but can be difficult to access. Many of the old satellite images are either difficult to track down or too low resolution to be useful. I mainly used Google Earth because the images were free, but resolution varies and sometimes the imagery changes overnight without warning. The Afghan Heritage Mapping Partnership has funding from the US State Department and access to many more high resolution images, some with 10cm pixel resolution, which is exceptionally good. This has enabled them to do much more.

Do you see any major changes to the remote sensing field in the near future?

Technology is always advancing. More and more images are being taken, with higher resolution. The cost of the imagery is coming down and the way we analyse images is becoming more sophisticated. Automating this process is key, because manually inspecting images is very time consuming. Now that we can teach computers to recognise the ‘signatures’ of looting, we can apply these techniques to many more sites and larger areas. Of course you have to cross-check what the computer identifies and ideally follow up with inspections on the ground, but that is not always possible, particularly in conflict and post-conflict areas. While fighting may have stopped, the risk of mines is a real concern.

In future, authorities and heritage experts will be able to set up automated monitoring of sites, particularly in remote areas, and create alerts when looting takes place. The imagery can also be used by urban planners to protect sites and to survey large areas prior to major infrastructure projects, to identify where they will have least impact.

The use of multispectral imagery also has great potential. It uses different bandwidths of light to identify sites not visible to the naked eye – for example, buried sites, which give off a slightly different ‘signature’ in the imagery.  Radar imagery (LiDAR) can identify slight changes in the topography of an area which might reveal a buried site.

It can also see through vegetation and has been used in jungles in South East Asia and central America. It’s an exciting time – the more we look, the more we find; as technology advances, we can ask different and more challenging questions.