Long-term land use and water management strategies in arid margin landscapes
​MarginScapes aims to bring together large-scale, multi-temporal and multi-source geospatial analysis to re-evaluate the cultural landscapes of South Asia. The project focuses on the Cholistan Desert (Pakistan) and the northern margins of the Thar desert (north-western India). Those areas were core regions for the development of the Indus Civilisation (ca. 3300-2500 BC).
​
By using a novel combination of Earth Observation data in petabyte-scale cloud computing environments and machine learning geostatistics, MarginScapes provides methodological tools and open access data to:
​
-
Understand the relationships between palaeohydrology, relict palaeosoils and the distribution of ancient sites;
-
Identify historical and modern landscape transformations;
-
Foresee the mechanisms of how populations coped and adapted to climate change, water scarcity and desertification.
People
Francesc C. Conesa
Cameron Petrie
MSCA-IF Fellow (2018-2020)
University of Cambridge
IJC Fellow (from April 2020)
Catalan Institute of Classical Archaeology
MSCA-IF Supervisor
Department of Archaeology
University of Cambridge
Project collaborators
​
-
Arnau Garcia-Molsosa, Catalan Institute of Classical Archaeology
-
Adam S. Green, McDonald Institute for Archaeological Research
-
Agustin Lobo, National Spanish Research Council
-
Marco Madella, Pompeu Fabra University
-
Tworains Team (ERC CoG n. 648609), McDonald Institute for Archaeological Research
Hector A. Orengo
IJC Supervisor
Catalan Institute
of Classical Archaeology
Research themes
Automated site detection
Endangered cultural heritage
Land use & change
​We explore the use of satellite virtual constellations (eg. Sentinel 1 & 2) and classification workflows to automatically detect and characterise soil signatures from archaeological features.
​We combine Earth Observation data with historical geospatial data, such as archaeological legacy data, topographical maps & journey narratives to map and re-evaluate historical landscape features.
​We monitor land cover seasonal trends, with a specific focus on vegetation dynamics and water availability, to link long-term landscape dynamics with past land use and archaeological site distribution.
Methods & Results
Archaeological mounds
​
-
A Google Earth Engine workflow that combines archived data with large scale analysis capabilities
-
Synergetic use of multispectral (Sentinel 2) and radar data (Sentinel 1)
-
Integrated Random Forest classifier
-
Validation with LDA and Montecarlo simulation in R
-
Detection of 504 potential archaeological mounds in the Cholistan Desert, Pakistan
Endangered cultural heritage
​
-
Medieval to modern military forts (UNESCO Tentative List)
-
Combined use of high-res satellite imagery (worldView 2-3), historical topographical maps and XIX century narratives
-
Reconstruction of historical trade routes and past semi-nomadic mobility
-
Detection and assessment of c. 40 forts in the Cholistan Desert, Pakistan
Publications and open data
"Detection of 504 potential archaeological mounds in the Cholistan Desert, Pakistán"
Automated detection of archaeological mounds using machine learning classification of multi-sensor and multi-temporal satellite data
Orengo, Conesa et al.
Submitted
Supplementary Information
&
Google Earth Engine code
(JavaScript)
Orengo, Conesa et al.
Coming soon!
Endangered margins: historical forts, caravan routes and fading heritage in the Cholistan Desert
Conesa et al.
Coming soon!
The cultural landscapes of Cholistan: geospatial archaeological & historical database
Conesa et al.
coming soon!
Contact
Francesc C. Conesa
Landscape Archaeology Research Group
Catalan Institute of Classical Archaeology
Plaça d’en Rovellat, s/n
43003 Tarragona
Funding
MSCA-IF Host Institution
IJC Host Institution
H2020-MSCA-IF-2017, n.794711
Special thanks to
