Best practice approaches for applying satellite imagery for landscape archaeological applications: a case study from the world heritage site of Sanchi, India

A ‘Reflexive’ Multi-Stage Survey Methodology for Historical Landscape Research in Central India:Field-Walking, Local Knowledge, and Satellite Imagery as Archaeological Site Prospection and Mapping Tools in the Sanchi Survey Project

“…Particular emphasis was placed on diachronic human : non-human : environment interaction through ground-based mapping of waterresource systems, geological dating of dam and reservoir deposits, and hydrological and climate data analysis 64 , while database redesign, remapping at selected sites, and satellite remote sensing were other key foci of enquiry. The latter aided the mapping process, ameliorated the GIS data quality, and enabled ground-based site representativeness during stage I to be tested against visibility levels within a subset of different satellite imageries 41,55 . The study area covered approximately 750 km 2 , extending to a 10-15 km radius around Sanchi hill (Figure 1).…”

“…The Ikonos imagery did independently reveal settlement mounds, verified against the original stage I survey data. The large size of these sites meant that they were potentially detectable in the LANDSAT ETM imagery also 55 . A selection of co-registered pan and multispectral (MS) Quickbird imagery was purchased primarily for site reconnaissance and mapping in the hilly regions, with variable but generally positive results.…”

“…However, the simple existence of remote sensing resources does not always make them the best starting point for primary reconnaissance, especially as the free Google Earth data are of inconsistent quality across any one study area 4 . As discussed below in relation to the SSP, dense forest coverage across much of the study area meant that satellite imagery only became useful after an initial stage of ground-based survey 55 .…”

“…The revised dataset included many previously undocumented sites, and revealed major inconsistencies and errors with regards to location (sometimes up to several kilometres) and quantification, thus shedding new light on models of Late and Post-Harappan urban transformation. The use of a range of commercial datasets by the SSP prior to these developments represented an early attempt to assess the usefulness of satellite imagery as a site prospection and mapping tool 1,55 . However, the simple existence of remote sensing resources does not always make them the best starting point for primary reconnaissance, especially as the free Google Earth data are of inconsistent quality across any one study area 4 .…”

“…A selection of co-registered pan and multispectral (MS) Quickbird imagery was purchased primarily for site reconnaissance and mapping in the hilly regions, with variable but generally positive results. The sensor characteristics of the SPOT imagery were simulated from the Quickbird imagery using two techniques: degradation of 0.6 metre pan imagery to 2.4 metres, and extraction of the first principal component of the MS bands 55 . The spatial resolution of the pan imagery was high enough for identification, in shadow, of wires between pylons.…”