‘Spying’ on the Amu Darya

how people changed the river course and destroyed archaeological heritage

CORONA satellite imagery is an amazing resource which allows us to virtually ‘time travel’ and see what the landscape was like and what archaeological monuments were still there 40 to 50 years ago.

  • Turkmenistan/Uzbekistan
  • Coordinates of fort still visible: 41.22689173°, 61.73932628°
  • Coordinates of fort now underwater: 41.18286941°, 61.64853225°
  • Site type: fort

Operational from the late 1950s until 1972, the CORONA satellite programme conducted photographic surveillance of the Soviet Union, People’s Republic of China and other key areas. The programme was officially classified until 1992 when the images were considered obsolete for military purposes; hence the US government started to release them to the public (Ruffner 1995).

The images are black and white panchromatic frames covering 200 km length by 15 km width with a spatial resolution of 2-4 m. This has resulted in an enormous amount of historical aerial images, allowing archaeologists to investigate vast territories in pre-urbanisation and pre-agricultural scenarios, where the visibility of archaeological sites and their contextual landscapes are now lost.

Using this extraordinary resource CAAL has begun to explore the changes in the water course of the Amu Darya (also known as the Oxus river) and the effects that this has had on local archaeological heritage. Specifically, in this case, we focused on the area around Gazadzhak at the border between Uzbekistan and Turkmenistan. The area is particularly interesting as the landscape shows substantial transformations during the last 50 years. After georeferencing a number of CORONA images dating from June-July 1970 it is possible to appreciate how the riverine area has been drastically transformed through processes of soil deposition to the upstream and soil erosion towards the downstream part of the image (Figure 1). Draining activities resulted in the appearance of new arable lands around Hazorasp in areas previously covered in marshland or simply part of the Amu Darya. Downstream of Gazadzhak, the opposite phenomenon of soil erosion destroyed and flooded vast areas of cultivated fields and villages. Remarkable is the case of a pair of fortified structures located on either bank of the river that are clearly visible on the CORONA image but have been heavily affected by artificial flooding of the river. In Figure 2 the structure is now completely underwater, whereas in Figure 3, the twin structure is highly threatened by the ongoing bank erosion process.

Figure 1. Georeferenced CORONA datasets that cover a stretch of the Amu Darya along which runs the border between Uzbekistan and Turkmenistan. @Digital Globe, via Google.
Fig. 2. Comparison between a fortified structure how it appears in the CORONA image from 1970 (left) and in the Google Satellite from 2019 (right). The structure is completely underwater in present days showing how the area has been flooded over the last 50 years. @Digital Globe, via Google
Fig 3. Comparison between a fortified structure how it appears in the CORONA image from 1970 (left) and in the Google Satellite from 2019 (right). The structure is highly endangered by the bank erosion caused by the artificial flooding of the Amu Darya river. @Digital Globe, via Google.

Relying on modern methods it is also possible to automatically map the present course of the river, from freely available satellite images like Landsat, and compare it to the old course observable n historical imagery. A Random Forest algorithm (Breiman 2001) can be used to classify Landsat 8 imagery from September 2019, thus providing 12 classes of different land uses (Figure 4). Subsequently, water bodies can be extracted and converted into polygons (Figure 5), thus allowing a superimposition with the historical imagery and to observe how the landscape has significantly changed in a half century.

Thanks to the US government spying beyond the Iron Curtain we can now rediscover precious stories about the past undertakings to control a significant communication route and a critical border between empires.

By Marco Nebbia

Fig. 4. Example of Landsat 8 (September 2019) image supervised classification using Random Forest algorithm.
Figure 5. Water bodies classified in the Landsat 8 image superimposed on the CORONA 1970. This image shows the magnitude of changes along this particular stretch of the Amu Darya.

Breiman, L. 2001. “Random Forests.” Machine Learning 45: 5–32. https://doi.org/10.1201/9780367816377-11

Ruffner, Kevin, ed. 1995. Corona: America’s First Satellite Program. Washington DC: History Staff, Center for the Study of Intelligence, Central Intelligence Agency. https://www.cia.gov/library/center-for-the-study-of-intelligence/csi-publications/books-and-monographs/corona.pdf

Leave a Reply