Contents lists available at ScienceDirect
Journal of Archaeological Science: Reports
journal homepage: www.elsevier.com/locate/jasrep
The application of GIS and satellite imagery in archaeological land-use
reconstruction: A predictive model?
Michael Kempf
Archaeological Institute, Department Early Medieval Archaeology, University of Freiburg, Germany
ARTICLE INFO
Keywords:
Remote sensing
Landscape monitoring
Environmental afordance
Cultural heritage
Predictive modelling
ABSTRACT
Landscape reconstructions are subject to multiple flters. Nevertheless, they are the basis for scientifc inter-
pretations of the archaeological record. The resulting distribution maps in turn serve to create archaeological
maps for the prediction of areas with a high archaeological vulnerability potential. This predictive modelling
also leads to the classifcation of archaeological-free areas. However, the fact that this spatial classifcation was
created on the basis of an archaeological database, which in turn is strongly biased by fnd densities in modern
agglomerations, leads to the potential destruction of archaeological monuments that are located in the in-
vulnerable periphery. Such potential maps can only be sharpened by integrating a large number of diferent
methodical concepts. This article presents the comprehensive combination of satellite remote sensing data, GIS-
based landscape analyses and environmental data sets for the construction of a potentially holistic landscape
scenario. Only with the highest possible information density of a large-scale study area can interactions in the
geographical control factors be identifed and conclusions be drawn on archaeological land-use and settlement
concepts.
1. Introduction
In recent years, satellite image-supported remote sensing applica-
tions have made a signifcant contribution to establishing new metho-
dical approaches in digital, quantitative archaeology (Keay et al., 2014;
Lasaponara and Masini, 2007, 2011). This includes multispectral image
analyses of medium- and high-resolution remote sensing satellite ima-
gery as well as data extraction from Synthetic Aperture Radar (SAR)
(Lasaponara and Masini, 2013). By the composition and recalculation of
the individual spectral channels, rapid estimates of the physical con-
dition of the (vegetation-covered) land surfaces can be made with little
fnancial efort or via open-source data packages from ESA, NASA and
third party missions (TPM) (Parcak, 2009). Evaluating so-called crop
and soil marks, non-invasive prospection methods of buried or shallow
archaeological remains can be carried out (Lasaponara and Masini,
2007; Masini and Lasaponara, 2006). Crop marks, however, are parti-
cularly dependent on the local soil conditions, water permeability,
depth and type of archaeological material to be determined, as well as
the type, degree and modern use of the overlying vegetation cover
(Doneus, 2013; Lasaponara and Masini, 2007). Structures in the re-
corded images that difer from their surroundings therefore pass
through two parallel flters that are methodically difcult to separate:
Are they archaeologically relevant material or are they naturally
formed fuctuations in plant physiological growth behaviour that point
to locally heterogeneous environmental conditions? This is very
strongly connected to another infuencing factor, which shows the
sensitivity of the archaeological fnding to remote sensing data: the
resolution of the images used in relation to the size of the fndings.
Tapete et al., 2013 already made clear that even medium resolution
imagery can serve to identify archaeological remains (Tapete et al.,
2013). Here, however, the direct distinction of the observed objects
from their immediate surroundings (e.g. soil, geology, vegetation) plays
a decisive role (Siart et al., 2008). Not all archaeological research areas
and thus chronological periods behave the same. Burial mounds, for
instance, which clearly stand out as landmarks in a less detailed en-
vironment, can easily be prospected by remote sensing applications
(Lasaponara and Masini, 2011). This is changing rapidly, if, for instance
small-scale wooden settlement structures or burial grounds which ex-
perienced massive surface restructuring are attempted to be identifed
by growth characteristics using medium resolution imagery. Even
though particularly deep-reaching structures that are protected from
mechanical rearrangement can be captured by aerial photo prospection
(Doneus, 2013; Doneus and Scharrer-Liška, 2003; Scharrer-Liška et al.,
2015), surface near structures can only be visualized with some dif-
culty. However, in cases where there is no direct visual access to ar-
chaeological remains, multispectral remote sensing data should be
https://doi.org/10.1016/j.jasrep.2019.03.035
Received 7 February 2019; Received in revised form 20 March 2019; Accepted 30 March 2019
E-mail address: Michael.kempf@archaeologie.uni-freiburg.de.
Journal of Archaeological Science: Reports 25 (2019) 116–128
2352-409X/ © 2019 Elsevier Ltd. All rights reserved.
T