Journal of Archaeological Science 38 (2011) 89e100

Contents lists available at ScienceDirect

Journal of Archaeological Science

journal homepage: http://www.elsevier.com/locate/jas

Integrated GIS, remote sensing and geomorphologic approaches for the

reconstruction of the landscape habitation of Thessaly during the neolithic period
Dimitrios Alexakis a, *, Apostolos Sarris b,1, Theodoros Astaras a, Konstantinos Albanakis a
a
Remote Sensing and GIS Applications Laboratory, Department of Physical and Environmental Geography, School of Geology, Aristotle University of Thessaloniki,
Thessaloniki 54124, Greece
b
Laboratory of Geophysical, Satellite Remote Sensing & Archaeo-environment, Institute for Mediterranean Studies, Foundation for Research & Technology,
Hellas (F.O.R.T.H.), Rethymno 74100, Crete, Greece

a r t i c l e i n f o a b s t r a c t

Article history: The aim of this study is to manifest the contribution of integrated approaches such as GIS, geomor-
Received 15 January 2010 phology, remote sensing and DEM analysis for the detection of neolithic settlements, the modeling of
Received in revised form habitation and the reconstruction of neolithic landscape in the area of Thessaly.
2 August 2010
Thessaly is a region of low relief in central Greece where hundreds of neolithic settlements called
Accepted 2 August 2010
magoules were established from Early Neolithic period until Bronze Age. In this study field survey,
statistical analysis of coring data, spatial analysis of environmental parameters in GIS and image pro-
Keywords:
cessing techniques of satellite images and DEMs were carried out to contribute to the detection of the
GIS
Remote sensing
Neolithic settlements and the reconstruction of Neolithic landscape.
DEM Ó 2010 Elsevier Ltd. All rights reserved.
Thessaly
Settlements
Neolithic

1. Introduction survey, analysis of stratigraphic coring data, spatial analysis of envi-

The three main applications of GIS in archaeology are visuali-
zation, spatial analysis and modeling (Church et al., 2000). Addi-
tionally, archaeology has become a recent application field for
satellite remote sensing and nowadays ancient settlements are able
to be detected with high resolution remote sensing techniques
(Menze and Sherratt, 2006).
Neolithic Thessaly has been studied for understanding human
partitioning and territoriality of the landscape by non-hierarchical
human groups. The distinct natural features of the Thessalian
landscape are ideal for reconstructing the major habitation patterns
of the first Neolithic farming groups of Greece. Thessaly is a region
of low relief with great alluvial plain, where hundreds of Neolithic
settlements/tells called magoules were established from Early
Neolithic period since Bronze Age (Alexakis et al., 2008).
In order to proceed with an integrated geo-archaeological
research, multiple methodological approaches were employed. Field
* Corresponding author.
E-mail addresses: alexakisdimitris@hotmail.com, alexakis@chania.teicrete.gr
(D. Alexakis), asaris@ret.forthnet.gr (A. Sarris), astaras@geo.auth.gr (T. Astaras).
1
www.ims.forth.gr.
ronmental parameters in GIS environment and satellite image pro-
cessing were carried out to cover the extensive study area and
contributed to the detection of the low relief Neolithic settlements.
The fusion of all these data with the digital elevation model (DEM) is
capable of providing detailed information for archaeological purposes
(Siart et al., 2008). Thus various digital elevation models were
reconstructed for each period of the Neolithic, which were ultimately
used in the corresponding spatial analysis of the landscape.
At the final stage, predictive models were created to contribute
to the detection of unknown settlements. The models that are
frequently used to forecast the location of archaeological sites
belong to the methodologies of archaeological prospection and are
frequently used in archaeological studies (Espa et al., 2006).
2. Study area, data and methodology
Thessaly is an extensive plain region located in Central Greece. It
is surrounded by the mountains Antichasia, Ossa, Maurovouni,
Pelion, Othris and Pindus. The Neolithic settlement mounds called
magoules are low hills of 1e5 m height and mean diameter 300 m.
The mounds mainly consist of loam and mud based materials. The
vast majority of magoules are laid on Larisa basin and with
a smaller number distributed in west Thessaly (Karditsa basin).

0305-4403/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2010.08.013
90 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100
Both of these plains consist of Quaternary alluvial deposits
(Alexakis et al. 2009).
Initially the study involved detailed modeling of the Thessalian
landscape by incorporating 3 different modules:

Topographic mapping of the settlements through the use of
Global Positioning Systems (GPS). The particular task was
carried out to map a large percentage (more than 342 settle-
ments) of the existing archaeological sites (Fig. 1).

Digitization of 1:50,000 scale topographic and geological
maps (Fig. 2) of the Geographic Service of the Hellenic Army
and of the Institute of Geological and Mineral Exploration
respectively. The original Digital Elevation Model (DEM) of
the study area with a cell size of 20 m was based on the
digitized 20 m elevation lines of the topographic maps. The
geological formations of the geological maps were reclassi-
fied to form a unified set of formations and all the further
details of the above maps, such as rivers, lakes, faults and
modern villages were also superimposed on the particular
background layers.
Fig. 1. (a) Thessaly is located in central Greece and constitutes the larger agricultural plains of it. (b) Spatial distribution of magoules in Larisa (east) and Karditsa (west) basins. (c)
magoula Rini. (d) magoula Aerino.

Construction of a detailed archaeological and environmental
database in SQL environment (type of site, chronological pha-
ses, type of raw materials present in the sites, height, diameter,
geology, vegetation, etc).
3. Landscape reconstruction
Before proceeding to the landscape reconstruction of Thessaly
during the various phases of the Neolithic period, the reliability and
accuracy of the existing terrain model was evaluated compared to
other digital elevation models, such as the 90 m pixel size DEM
from the Shuttle Radar Topography Mission (SRTM) and the 30 m
pixel size DEMs provided by ASTER images or constructed by the
L1-A stereoscopic products (3N and 3B) of ASTER satellite. The
equation of Schumann et al. (2008) was used to check the validity of
DEMs by comparing a number of reference points (datum points):

qX
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2
RMSEDEM ¼ ER  EDEM =n
 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100 91

Fig. 2. The Geological map of Thessaly.

where EDEM stands for the heights of each DEM, and n is the total
number of reference points.
The results indicated that the RMSE for the DEM created through
the digitization of the contour lines of the topographic maps was
the lowest (0.246), followed by the RMSE of the SRTM DEM
(0.263). In contrast, the RMSE of ASTER products was the highest
(RMSEL1A ¼ 0.318 & RMSEASTER ¼ 3.490). Based on the above, the
DEM that was produced through the digitization of the topographic
maps became the basic layer of the terrain for the analyses that
followed. Reconstruction of the terrain addressed specific attributes
of the landscape, such as the alluvial basins, the coastline and the
regime close to Lake Karla, each one based on different processes.
3.1. Alluvial basins
Thessaly consists of two major basins, Larisa Plain with an area
of 1020 Km2 and Karditsa Plain with an area of 2210 Km2. The two
plains contain 181 out of the 342 known registered “magoules”.
This fact proves the important role of reconstruction of the relief of
each basin during each Neolithic sub-period. Both geological
(stratigraphic data from boreholes and past geomorphologic
studies) and archaeological data were placed under consideration
for achieving this task.
Geologically, Thessaly belongs to Pelagonian massif to the east
and the Pindos range to the west and due to their different tectonic
activity each alluvial basin was reconstructed in a different manner. A
geological database of 50 selected drill cores (based on the quality
and details of data presented) was constructed out of a total of more
than 400 drill cores that had been collected and reported by the
Prefectures of Karditsa, Larissa, Volos, and Trikala. The depth to the
alluvial deposits was estimated based on the most reliable drill cores,
whereas provision of the local relief was taken into account for a more
accurate and absolute estimation of the alluvium deposits depths.
Estimations of the alluvial deposits for the three main Neolithic
sub-periods were obtained following Demitrack’s assumption (for
Larisa basin) about deposition rates (Demitrack, 1986): 5.5 m for
Late Neolithic, 6.5 m for Middle Neolithic and 8.5 m for Early
Neolithic.
Based on the above, the reduction of the drill core data con-
cerning the alluvial deposits for the particular periods was based on
the following algorithm:
X ¼ ðdy  ymin =ðymax  ymin ÞÞ þ ymin0
where X is the reduced depth of deposits (for the particular drill
holes), dy is the current depth of deposits (from drill cores), ymin is

Table 1
Equations estimating the reconstructed depths to the alluvial deposits of the three Neolithic periods (early, middle and late) for each one of the major basins of Thessaly
(Karditsa and Larisa). In the above formulas, Y is the current elevation and X is the reconstructed alluvial deposits’ depth.

Early Neolithic Middle Neolithic Late Neolithic

Karditsa Y ¼ 31.88  X  171.62 Y ¼ 31.88  X  107.85 Y ¼ 31.88  X  66.59
Larisa Y ¼ 17.78  X þ 219.38 Y ¼ 17.78  X þ 183.82 Y ¼ 10.18  X þ 120.08
92 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100

Table 2 settlements and the excavation information provided for “platia

Statistical results of the location of the Neolithic settlements (for the three major magoula Zarkou” (Van Andel et al., 1992) were statistically pro-
periods of Neolithic) in terms of the reconstructed depth of alluvial deposits for each
period of interest. The classification of the depths was based on the concentration of
cessed following a classification of magoules to 14 categories
the data within certain depth limits by taking in account the standard deviation of according to their duration of habitation. The reduction of the
the samples. corresponding heights was based to the cross-section plan of
Neolithic sub-periods Number Percentage % for Reconstructed depths
“platia magoula Zarkou” provided by Van Andel et al. (1992). Three
of Sites each period to alluvial deposits (m) buffer zones of 50 m each around each magoula were created
Early Neolithic 67 37 <8 taking in account the mean diameter of magoules as it was esti-
Settlements 108 59.6 8e10 mated during field work. The first (central) buffer zone kept the
6 3.31 >10 height of each magoula as mentioned above, whereas in the second
Middle Neolithic 21 11.6 <5 (middle) and third (outer) zones the height was gradually reduced
Settlements 123 67.95 5e7
37 20.44 >7
by 50% and 25% of the initial height correspondingly (Fig. 3).
Late Neolithic 58 32 <4
Settlements 97 53.59 4e6
26 14.36 >6 3.2. Reconstruction of coastline

A rise of sea level by 100e120 m affected the Aegean Sea after

the minimum depth of deposits (within the basin from drill cores),
ymax is the maximum depth of deposits (within the basin from drill
cores) and ymin0 is the minimum depth of deposits according to
Demitrack (1986).
Although the Quaternary of TrikalaeKarditsa basin is less well
known than that of the Larisa basin, a similar procedure was fol-
lowed, due to the fact that depositional processes in the area reflect
a parallel history of floodplain deposition and incision, closely
related to that of Larisa basin.
The above estimates were used for calculating (through a best fit
line) a much more general equation for converting the current
elevation (namely current DEM of alluvial basins) to the corre-
sponding alluvial depths (reconstructed DEMs) for each one of the
three major Neolithic sub-periods. Six linear equations were
approximated to define the depth of the deposits for the three
periods of Neolithic for Karditsa and Larisa basins independently
(Alexakis et al., 2008) (Table 1).
The above approximations were applied to the two alluvial
basins and a reconstructed DEM for each basin and each Neolithic
period was created. Additionally, all the altitudes at the moun-
tainous areas remained unaltered. Then, using Boolean algebra, the
reconstructed DEM for the whole region of Thessaly was formed.
An overlay of the magoules on the reconstructed DEM provided
valuable information about the altitude of the settlements in the
specific locations (Table 2).
The next step regarded the reconstruction of the micro-topog-
raphy around the settlements. The data regarding the height of the
the last glacial period of Wurm. This fact affected also the coastline
of Thessaly, especially the plains of Holocene alluvium deposits. In
order to examine the changes of the coastline during the Early
Neolithic to the Late Neolithic period, the results of the study of
Kampouroglou (1990) and Vouvalidis et al. (2005) were adopted.
Kampouroglou made approximate plans of the Neolithic coastline
of Volos plain based on a number of drill holes within the Volos
basin. All these data were geo-referenced with the help of ASTER
images. According to the spatial measurements carried out in
different sections of the coastline, it was noticed that in the Early
Neolithic period the coastline was about 650 m seawards, whereas
during the Late Neolithic period the coastline moved 300 m inland,
compared to the current coastline.
3.3. Reconstruction of Lake Karla
The reconstruction of ancient Lake Karla was considered to be of
major importance for the imaging of Neolithic Thessaly landscape.
During the Neolithic period, Lake Karla was at the SE of Larisa plain
and its coastline continued to fluctuate during various time zones
until it was dried during the last century. The reconstructive model
was based on the outline of its prehistoric extent by Grundmann
(1937) and the spatial distribution of the settlements as they
were recorded by the GPS survey. The results suggested that during
the Neolithic period the lake’s level extended up to the 40 m
contour line in agreement to Halstead (1984), who considered the
lake’s levels to exist below the 44 m elevation contour (Fig. 4).

Fig. 3. (a) Reconstruction of the micro-topography around the Neolithic magoules - Point location of the magoula. (b) Reconstructed DEM around the magoula.
 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100 93

Fig. 4. The area of Lake Karla. The red dots represent the Neolithic magoules around the lake. The blue line is the suggested lake level according to Grundmann (1937). The purple
color is the suggested lake extent according to the recent study. The light green color represents the Larisa plain. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article).

4. Satellite image processing 4.1. Preprocessing of satellite images

Satellite remote sensing was also employed in the particular
project in order to investigate the capability of space and aerial
sensors and image processing techniques in the detection of
Neolithic settlements in Thessaly. Satellite data from ASTER, Landsat
ETM, HYPERION and Ikonos space platforms were employed,
together with aerial photos, having different area coverage and
resolution (Table 3).
Masking of the sea, clouds and snow areas in all images preceded
the processing of the images in order to focus to the mainland and the
areas that provided useful information. Image mosaics were created
accordingly depending on the types of sensors and both image
mosaics and isolated images were rectified to a common projection
system (EGSA87/HGSR87). Digital numbers of images were also
converted to reflectance values according to specific conversion

Table 3
Spectral, spatial and temporal attributes of the satellite sensors and airphotos that were used for the study of the Neolithic settlements of Thessaly.

Sensor Acquisition Date Spatial Resolution (m) Subsystem Spectral Bands Radiometric Resolution
Hyperspectral Sensors
1. HYPERION September 3, 2001 30 VNIR, SWIR 137 16-bit

Hyper Spatial Sensors

1. IKONOS June 1, 2006 1 VNIR 4 11-bit
December 12, 2005
March 1, 2007
December 12, 2005
June 16, 2006

Multi Spectral Sensors

1. Landsat e 7 July 28, 1999 30 VNIR, SWIR 7 8-bit
60 8-bit
15 Panchromatic 8-bit
2. Aster March 19, 2003 15 VNIR 8-bit
June 30, 2004 30 SWIR 8-bit
June 30, 2003 90 TIR 12-bit
March 19, 2003
3. Airphotos January 3, 1998 1,5
94 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100

Fig. 5. (a) RGB / 2,3,7 e Farsalitis settlement. (b) Appearance of Melia 2 settlement after the application of 23 bands RGB composite to HYPERION image.

equations. The last step was necessary in order to have a uniformity
in the values of images originating from different sensors.
4.2. Spectral enhancement techniques
Several RGB composites were constructed in an effort to
examine their efficiency in the detection of the Neolithic settle-
ments. For the ASTER image with acquisition date 19-03-2003,
where most of the magoules were registered, the RGB / 1,2,3,
RGB / 3,2,5 and RGB / 2,3,7 composites were the most successful
for the visual detection of the Neolithic settlements (39 out of 239
settlements were highly visible, 49 average visible and 151 poorly
visible). All these composites appeared to have the highest
Optimum Index Factor (Alexakis et al., 2009). Similarly, RGB
composites of IKONOS images were able to detect 27 out of 48
settlements within their area of coverage. It is worth mentioning
that 19 of the detectable magoules, namely the highest of all cor-
responding to an average altitude of 4.6 m, were highly visible in all
RGB composites. On the other hand, RGB composites of Landsat and
HYPERION images were not very promising (for HYPERION
composites only 5 out of 21 settlements were detected). Due to
their high spatial resolution, all the 5 settlements that felt within
the spatial limits of the airphoto mosaic were easily detectable. As
a general conclusion however, the acquisition date of the images
proved to be the most crucial factor for the detection of magoules
mainly due to the intensive cultivation (mainly soft and shallow
cultivation) of the landscape both on the top in the surroundings of
magoules.
Principal Component Analysis (PCA) was applied to ASTER,
Landsat and HYPERION images, being especially effective for ASTER
images where 39 and 47 out of 247 settlements were highly or
medium discriminated correspondingly. Image fusion techniques
through the combination of high spatial resolution images such as
IKONOS (1 m) and high spectral resolution images such as HYPE-
RION (30 m) concluded to very promising results. Finally, a spectral
mixer utility (Erdas Imagine 9.1 software) helped to exploit the
dynamic range of all the multispectral information of the HYPE-
RION image by combining more than three bands to an RGB
composite. Using the specific utility and assigning a weighting
coefficient for each band, an RGB composite of 23 bands (38, 42, 48,
49, 50, 51, 52), (85, 86, 87, 88, 89, 90, 91, 92,) & (93, 94, 108, 109, 110,
111, 113, 114) was constructed that enhanced the visual appearance
of the magoules (Fig. 5).
4.3. Radiometric enhancement
Of equal importance was the radiometric enhancement of the
images through manipulation of the histogram. After applying

Fig. 6. (a) Appearance of 3 settlements in the original IKONOS image (left) and the radiometrically enhanced image where three Neolithic settlements are highlighted. (b) To the
north of Galini-3 settlement, shown at the lower right of the image, another smaller candidate magoula is suggested.
 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100 95

Fig. 7. (a) Land classification of ASTER image through the use of Mahalanobis distance (fuzzy) algorithm. (b) Detail from the application of NDVI to Landsat image. (c) ASTER image
around Halki area after the application of Sobel Right Diagonal filter. Neolithic magoules are indicated within the ellipses.

radiometric enhancement to ASTER images (acquisition date of 19-
03-2003) we managed to detect 57 settlements. A non-linear
radiometric enhancement of the HYPERION PCA image, followed by
an inversion of brightness was able to highlight 8 out of 9 settle-
ments. Similar type of non-linear radiometric enhancement of the
high resolution Ikonos images through the modification of the
histogram was able to outline the round shape of known magoules,
as well as to identify 10 more targets of similar geometry that need
to be verified by the ground truthing activities that will follow
(Fig. 6) (Alexakis et al., 2009).
4.4. Land classification, vegetation indices, spatial enhancement
A spectral signature database was constructed to provide the
basic spectral information about tells, especially at the plain areas
of Thessaly. Several classification methods were applied to Landsat
and ASTER images in order to investigate the land use regime
around the magoules. Examination of the overall accuracy of the
various algorithms tested (based on the error matrix), proved that
the Mahalanobis algorithm was the most efficient for the exact
classification of the images (Fig. 7a). Additionally, object based
segmentation techniques were applied to ASTER images and 15
settlements in total of 234 were detected easily (Alexakis et al.,
2009). The computation of the Normalized Difference Vegetation
Index (NDVI) was used to highlight the vegetation differences
during different periods of time, in an effort to pinpoint any vague
indications for the detection of magoules. As expected, the NDVI of
the spring ASTER image was higher than the summer Landsat
image, but still the vegetation differences of the spring time favored
the detection of magoules mainly due to the differentiations of the
soil’s humidity (Fig. 7b).
Application of certain spatial high pass filters contributed
further to the spatial enhancement of smaller features such as the
magoules. The most reliable of them proved to be Sobel Right
Diagonal 3  3 and Laplace 3  3, both of which outlined clearly the
limits of the most prominent of them (Fig. 7c).
5. Analysis in GIS environment
An extensive spatial analysis of the magoules was carried out in
a GIS platform using both the current DEM and the reconstructed
DEMs (Fig. 8). The current 20 m pixel size DEM, produced through
the digitization of 1:50,000 scale maps of the Geographic Service of
Greek Army, was also used to understand the local characteristics of
the magoules on the current terrain. Extraction of statistics about
the Neolithic settlements in relation to the aspect, the slope (Table
4) and the relief height were carried out after a reclassification of
the above attributes. The vast majority of settlements (48, 11%)
were found to be laid on plain areas (slope of relief 0e1%) and more
than 50% of magoules have an E-NE aspect.
The distance of settlements from natural resources was also
explored by applying buffer zones around the quarries and the water
springs (mainly springs existing on the mountainous areas).
Watersheds were also constructed and the distance of each settle-
ment from the neighbor watershed was calculated. Three different
cumulative viewshed maps, corresponding to the 3 different
Neolithic periods, were constructed by taking in account a 1.64 m
average height of a neolithic man standing on a base of about 1e3 m
in height (Fig. 9). The cumulative viewshed maps are the algebraic
sum of the individual binary viewshed maps from each site,
displaying all regions that are visible from a cluster of sites and
thus providing an insight regarding the potential hierarchical

Fig. 8. (a) Digital elevation model of Thessaly for the Early Neolithic period. (b) Slope gradient map of Thessaly for the Early Neolithic period. (c) Aspect map of Thessaly for the Early
Neolithic period.
96 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100

Table 4
Percentages of settlements (total population of settlements) in relation to the slope
(current DEM).
Slope Slope of Number of % Percentage of
relief % Settlements settlements
Flat e Almost Flat 0e1% 166 48.11
Very gentle 1e3% 81 23.4
Gentle 3e5% 37 10.74
Modest 5e10% 30 8.69
Moderately steep 10e15% 9 2.6
Steep 15e30% 11 3.18
Very steep >30% 11 3.18
classification among the settlements (Lock and Stancic, 1995). By
using Kolmogorov e Smirnoff goodness-of-fit statistic test
(Kvamme, 1990), we concluded that in most cases sites are distrib-
uted respectively to their visibility connection with other sites
(Table 5). The particular test allows the comparison of an empirical
sample distribution against an expected referent distribution. The
test is performed by plotting the cumulative distributions of the
sample and population and obtaining the difference D between
these two curves. The maximum value, Dmax is compared with
a critical value d obtained from the sample size and the required
confidence interval. For alpha ¼ 0.05, for example, and in a two-
tailed test, the value of d can be calculated from equation d ¼ 1.36/
On. If Dmax exceeds this critical value d, then the sample distribution
is significantly different from the population distribution and one
can conclude a non-random distribution (Wheatley, 1995).
Density maps of the settlements were also created for each
Neolithic period. The calculation of the density of the settlements
was accomplished through the use of a non-parametric Kernel
technique. The kernel technique is based on the kernel function
described in Silverman (1986). The results indicated that the
settlements were not dispersed for all the three periods of the
Neolithic, in agreement with the conclusions of Halstead (1984)
and Perles (1999) (Fig. 10). With the help of density maps and the
geomorphologic regime of Neolithic Thessaly, effort was made to
recognize groups of habitation zones/clusters. For each of those
zones, a geographic centroid and a diachronic centroid were
identified by taking also in account the size and the temporality of
habitation. The spatial territorial limits of the settlements were
further explored using the Thiessen polygons analysis. The Thiessen
polygons are related to Delaynay Triangulation of a set of points,
which is defined as that in which the triangles are closest to
equilateral, and in which the circles whose circumferences pass
through the points of the triangles contain no other points
(Wheatley and Gillings, 2002). The classification of the size of the
Thiessen polygons was accomplished through Natural breaks
method that uses Jenks Optimal Statistic Method to create specific
Table 5
The results of KolmogoroveSmirnov test for the three periods of Neolithic.
d Dmax Distribution of sites in
relation with viewshed
Early Neolithic 0.18 0,29 (h ¼ 1 m) Not random
Middle Neolithic 0.21 0,17 (h ¼ 1 m) Random
0,36 (h ¼ 3 m) Not random
Late Neolithic 0.16 0,17 (h ¼ 1 m) Not random
categories with internal continuity (Jenks and Gaspall, 1971;
Conolly and Lake, 2006) and it was suggested that there is no
specific relationship between the size of the settlements and their
catchment area (Fig. 11). On the other hand it was noticed that the
high density of the settlements in the valleys of Larisa and Karditsa
corresponded to small catchment areas.
The site catchment of Neolithic settlements was also verified
through least cost surface analysis. Cost surfaces contributed also to
explore the regime of communication between the different
settlements. The cost surfaces can be regarded as modifications to
a continuous proximity product that takes into account the type of
the terrain over which the proximity is measured. Using the result
of a cost surface algorithm, it is possible to ascertain the ‘leastecost’
pathway between any two given locations in the landscape
(Wheatley and Gillings, 2002). Based on this, least cost paths were
also computed between the geographic and diachronic centroids of
each habitation zone (Fig. 12) in order to model the main
communication routes between the habitation zones.
Finally, GIS tools were employed to construct predictive models for
each phase of the Neolithic period in an effort to locate areas that
could host similar type of settlements. The specific predictive models
were based on the use of a multi-parametric spatial analysis method
of geographic elements and other information (statistical, archaeo-
logical, a.o.) in order to construct maps of archaeological interest
(Kohler and Parker, 1986; Judge and Sebastian, 1988; Mehler and
Wescott, 2006; Van Leusen and Kamermas, 2005; Jacoli and
Carrara, 1995). All the environmental factors (height, aspect, slope,
distance from watersheds, distance from water springs, distance from
quarries, geology, viewshed, distance from chert sources, least cost
paths, a.o.) that could have affected the choice of habitation in
Neolithic Thessaly were statistically examined and certain weight
factors were specified according to expert’s knowledge in relation to
the correlation of factors with the already known settlements and
their importance in terms of the location of the settlements (Alexakis
et al., 2009). The category of each factor was also rated (Table 6) and
a final equation was computed to account for the fusion of the various
layers: ([DEM]  1 þ [Slope]  1 þ Aspect]  0.6 þ [Distance from
Watershed]  0.7 þ [Distance from springs  0.7 þ [Distance from
quarries]  0.4 þ [Viewsheds]  0.8 þ [Geology]  0.9 þ [Cost

Fig. 9. (a) Cumulative viewshed map for Late Neolithic period. (b) Cumulative viewshed map for Middle Neolithic period. (c) Cumulative viewshed map for Early Neolithic period.
 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100 97

Fig. 10. (a) Density map for Late Neolithic period. (b) Density maps for Middle Neolithic period. (c) Density map for Early Neolithic period.

Fig. 11. (a) Thiessen polygons for Late Neolithic period. (b) Thiessen polygons for Midde Neolithic period. (c) Thiessen polygons for Early Neolithic period.

Fig. 12. (a) Least cost path analysis for Late Neolithic period. (b) Least cost path analysis for Middle Neolithic period. (c) Least cost path analysis for Early Neolithic period.

Distance from Coast]  0.4 þ [Distance from chert

Table 6
Example of weight factors and rating indices for two environmental parameters sources]  0.3 þ [Distance from least cost path]  0.7 þ [Distance from
corresponding to each phase of the Neolithic period. Lake Karla]). At a final stage, a fuzzy logic algorithm and a normali-
zation equation were also applied for a better tuning of the results and
Environmental Rating Early Middle Late
Factors Neolithic Neolithic Neolithic for rating the final probability from 0 to 1. Three different predictive
models were constructed for each phase of the Neolithic period
Weight Weight Weight
Factor Factor Factor
through the application of GIS Boolean algebra, making use of the
corresponding DEMs (Fig. 13).
1 Height (DEM) 1
(meters) The formulation of a predictive model does not guarantee the
<80 10 9 9 accuracy of its predictions (Conolly and Lake, 2006). The validity of
80e100 8 6 9 the model was tested by adopting the method of Kvamme (1988)
100e120 6 6 6 through the estimation of the predictive gain that is defined as:
120e150 6 4 6
150e200 6 4 5
G ¼ 1 e (% of total area where sites are predicted/% of observed
>200 3 3 5 sites within area where they are predicted),
where G is calculated for a specified probability of site occur-
2 Distance from 0.7
watershed (meters) rence, that ranges from 1 (high predictive utility) to 0 (no predictive
<100 4 4 4 utility). The high gain for the 3 phases of the Neolithic period (0.73,
100e250 9 9 9 0.78, 0.8 for Early, Middle and Late Neolithic periods respectively)
250e500 7 7 7 confirmed the reliability of the specific models. Indeed, about
500e1000 7 7 7
>1000 3 3 3
60e70% of the sites felt within the high and medium probability
zones in the corresponding predictive maps.
98 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100

Fig. 13. (a) Fuzzy logic prediction maps for Late Neolithic period. (b) Fuzzy logic prediction maps for Middle Neolithic period. (c) Fuzzy logic prediction maps for Early Neolithic
period.

Fig. 14. (a) magoula Tyrnavos 6 in the original DEM. (b) magoula Tyrnavos 6 after the application of matched filter.

6. Application of sophisticated filters to the digital elevation

models
The final approach of the particular project involved the detec-
tion of Neolithic settlements through the analysis of the Digital
Elevation Models (DEM) with the use of three different semi-
eautomated methodologies. Three different DEMs (90 m pixel size
SRTM DEM, 30 m ASTER DEM and a 20 m DEM from the digitization
of contours of topographic maps) were tested in all procedures to
attest for their potential in the detection of the magoules.
The first methodology involved the estimation of the index of
convexity (CI) to the three different DEMs according to Fry et al.
(2004):
CI ¼ ðx  xmed Þ=ðxmax  xmed Þ;
where x is the initial DEM, xmed is the DEM after the application
of median filter 7  7 and xmax is the DEM after the application of
maximum filter 7  7.
Although the index of convexity seemed to be ideal for the
detection of low hills such as magoules, only 35 (28%) of them at
Larisa plain and 28 (47%) at Karditsa plain were detected by this
method.
The second methodology is related with the design and appli-
cation of customized filters similar to those used by Menze and
Sherratt (2006). The optimal filter for the detection of a signal
with a well known shape is the matched filter. For the construction

Fig. 15. (a) Final detection map of magoules in the Larisa plain created following the approach of Iwahashi and Kamiya (1995). (b) Details of it showing the areas of high convexity in
orange. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100
of the matched filter, an area of 5  5 pixel was cut around the DEM
of each settlement for a total of 50 settlements. Then, the value of
the central pixel was subtracted by each pixel, followed by stacking
of all the 50 local DEMs (through the layer stack utility of Erdas
Imagine software) to form a final multilayer image. The particular
image was imposed to Principal Component Analysis and the first 5
principal components were summed. The negative sum replaced
the value of the central pixel of each of the above 5 filters, which
they were then applied individually to Larisa and Karditsa basins as
a detection filter (Fig. 14). The statistics for these filters proved that
the specific methodology is really promising especially for the
SRTM DEM in the area of Larisa. More specifically, about 60% of the
settlements were detected through the application of the first and
second filters.
The third methodology followed the approach of Iwahashi and
Kamiya (1995) for the estimation of the geometric signatures of
DEM through a combined study of slope gradient, surface convexity
and texture. Specifically, binary files were formed through estimation
of the mean value of slope gradient: all pixels with value above the
mean value took a value 1 and all the rest took a value of 0. The same
binary archives were created after the subtraction of the initial DEM
from the one that has been processed through the application of
a median filter. The last binary image was created after the applica-
tion of a Laplace 3  3 filter to the initial DEM. In the end, the three
binary archives were summed and the final map highlighted the
areas of high local convexity. Fuzzy logic algorithms have been
applied to the final results of the filtered DEM in order to produce
a better classification scheme (Pixel values equal to 0 formed the first
group, values from 0 to 3 formed the second group and pixels with
value equal to 3 formed the third group) (Fig. 15). The application of
this methodology to SRTM resulted to the detection of 35% of
magoules in the Larisa plain and only 15% of the magoules for Kar-
ditsa plain.
The results obtained through the Menze and Sherratt (2006)
approach to the SRTM DEM were also implemented to the predic-
tive modeling, together with other subproducts of the satellite
image analysis, such as the NDVI map, land use classification and
spectral signatures library of magoules. A similar methodology of
significant weights and factors was considered and results were
subjected to fuzzy logic and normalization techniques. Still, the
results of predictive modeling did not alter significantly from the
previous approach, signifying a state of saturation for the param-
eters considered.
7. Conclusions
Both geological studies and archaeological evidence contributed
to the reconstruction of the Neolithic Thessaly landscape. Each one
of the local environments of the Thessalian plain (alluvial plain,
coastline, Lake Karla) was approached in a different way to model
the macro-topography of the Neolithic period, which together with
the micro-scale relief changes around the magoules based on the
existing archaeological evidences, gave rise to a series of recon-
structed terrains for each phase of the Neolithic period. The
reconstructed DEMs proved to be much more effective than
the current DEM for carrying out the spatial analyses required for
the study of the settlement patterns of Neolithic Thessaly. Almost
all the specific analyses, including viewshed analysis, site catch-
ment, least lost surfaces and distances to raw resources, density
maps and clustering suggest that the particular settlement patterns
and distribution of sites was far from random. Satellite image
processing techniques and DEM analytical methods proved to be
also promising for the detection of Neolithic magoules, the latter
especially when applied to the SRTM DEM. In the end, a synergy of
all the above was required to participate in the design of a modular
99
significance weighted predictive model that could promote the
further detection of Neolithic magoules and their protection.
Acknowledgements
This research project is co-financed by INSTAP (Institute for
Aegean Prehistory), PENED (E.U.-European Social Fund (75%) and
the Greek Ministry of Development-GSRT (25%)) and IKY (States
Scholarship Foundation).
References
Alexakis, D., Astaras, Th., Sarris, A., Vouzaxakis, K., Karimali, L., 2008. Reconstructing
the Neolithic Landscape of Thessaly through a GIS and Geological Approach. In:
A. Polsluchn, K. Lambers and I. Herzog (eds.), Layers of Perception. Proceedings
of the 35th International Conference on Computer Applications and Quantita-
tive Methods in Archaeology/(CAA), Berlin, Germany. Kolloquien zur Vor und
Fruhgeschite, CD publication e p. 411, Bonn.
Alexakis, D., Sarris, A., Astaras, T., Albanakis, K., 2009. Detection of neolithic
settlements in Thessaly (Greece) through multispectral and Hyperspectral
satellite Imagery. Sensors 9 (2), 1167e1187. doi:10.3390/s90201167.
Church, T., Brandon, J.R., Burgett, G., 2000. GIS Applications in Archaeology, Method
in Search of Theory, Practical Applications of GIS for Archaeologists e A
Predictive Modeling Toolkit. Taylor & Francis, London.
Conolly, J., Lake, M., 2006. Geographical Information Systems in Archaeology
Cambridge.
Demitrack, A., 1986. The Late Quaternary Geologic History of the Larisa Plain,
Thessaly, Greece: Tectonic, Climatic and Human Impact on the Landscape, Ph.D.
thesis, Stanford University.
Espa, G., Benedetti, R., De Meo, A., Ricci, U., Espa, C., 2006. GIS based models and
estimation method for the probability of archaeological site location. Journal of
Cultural Heritage 7, 147e155.
Fry, G.L.A., Skar, B., Jerpansen, G., Bakkestuen, V., Erikstad, L., 2004. Locating
archaeological sites in the landscape: a hierarchical approach based on land-
scape indicators. Landscape and Urban Planning 67, 97e107.
Grundmann, K., 1937. Magoula Hadzimissiotiki. American Antiquity 62, 56e69.
Halstead, J., 1984. Strategies for Survival: An ecological Approach to social and
economic change in the early farming communitites of Thessaly, N. Greece,
Ph.D thesis, King’ s College, Cambridge University.
Iwahashi, J., Kamiya, I., 1995. Landform classification using digital elevation model
by the skills of image procesingdmainly using the Digital National Land
Information. Geoinformatics 6 (2), 97e108 (in Japanese with English abstract).
Jacoli, M., Carrara, A. Gis-based multivariate model for identifying archaeological
sites, Calabria, Southern Italy, I Intern. Congress on Science and technology for
the safeguard of cultural heritage in the Mediterranean basin, Nov. 27eDec. 2,
1995, Catania, Italy.
Jenks, G.F., Gaspall, F.C., 1971. Error on cloropleth maps: definition, measurement,
reduction. Annals of the Association of American Geographers 61 (2), 217e244.
Judge, J.W., Sebastian, L., 1988. Quantifying the Present and Predicting the Past:
Theory, Method and Application of Archaeological Predictive Modelling. U.S.
Department of the Interior, Bureau of Land and Management, Denver.
Kampouroglou, E., 1990. H Geomorphologiki Exelixi toy Kolpou toy Volou apo tin
Neolithiki Epochi mexri Simera, La Thessalie, Quinze anees de recherches
archeologiques, 1975e1990: Bilans Et Perspectives, Actes du Colloque Interna-
tional, Lyon, 12e22 Avril , vol. A, pp. 41e52.
Kohler, T.A., Parker, S.C., 1986. Predictive models for archaeological resource loca-
tion. In: Schiffer, M.B. (Ed.), Advances in Archaeological Method and Theory, vol.
9. Academic Press, New York, pp. 397e452.
Kvamme, K., 1988. Development and testing of Quantitative models. In: Judge, W.,
Sebastian, L. (Eds.), Quantifying the Present and Predicting the Past: Theory,
Method and Application of Archaeological Predictive Modelling, pp. 325e428.
Denver.
Kvamme, K.L., 1990. One-sample tests in regional archaeological analysis: new
possibilities through computer technology. American Antiquity 55 (2),
367e381.
Lock, G.R., Stançic, Z., 1995. Archaeology and Geographical Information Systems:
A European Perspective London.
Mehrer, M., Wescott, K., 2006. GIS and Archaeological Site Location Modelling. CRC
Press, Boca Raton.
Menze, B.H., Sherratt, A.G., 2006. Detection of ancient settlement mounds:
archaeological survey based on the SRTM terrain model. Photogrammetric
Engineering & Remote Sensing 72, 321e327.
Perles, C., 1999. The distribution of magoules in eastern Thessaly. In: Halstead, P.
(Ed.), Neolithic Society in Greece. Sheffield Studies in Aegean Archaeology.
Sheffield Academic Press., Sheffield, pp. 42e56.
Schumann, G., Matgen, P., Cutler, M.E.J., Black, A., Hoffmann, L., Pfister, L., 2008.
Comparison of remotely sensed water stages from Lidar, topographic contours
and SRTM. Journal of Photogrammetry and Remote Sensing 63 (3), 283e296.
Siart, C., Eitel, B., Panagiotopoulos, D., 2008. Investigation of past archaeological
landscapes using remote sensing and GIS: a multi method case study from
mount Ida, Crete. Journal of Archaeological Sciences 35, 2918e2926.
100 D. Alexakis et al. / Journal of Archaeological Science 38 (2011) 89e100

Silverman, B.W., 1986. Density Estimation for Statistics and Data Analysis. Chapman
and Hall, New York.
Van Andel, T., Gallis, K., Toufexis, G., 1992. Early Neolithic Farming in a Thessalian
River Landscape, Greece, Mediterranean Quaternary River Environments
Proceedings. University of Cambridge.
Van Leusen, M., Kamermas, H., 2005. Predictive Modelling for Archaeological Heritage
Management: a Research Agenda. In: Nederlandse Archaeologische Rapporten,
vol. 29. Rijksdienst voor het Oudheidkundig Bodemonderzoek, Amersfoort.
Vouvalidis, K.G., Syrides, G.E., Albanakis, K.S., 2005. Holocene Morphology of the
Thessaloniki Bay: Impact of Sea level rise. Zeitschrift fuer Geomorphologie 137
(N.F. Suppl) Berlin -Stuggart.
Wheatley, D., 1995. Cumulative Viewshed Analysis: A GIS-based Method for
Investigating Intervisibility, and Its Archaeological Application. Archaeology
and GIS: a European Perspective. Routlege, London.
Wheatley, D., Gillings, M., 2002. Spatial Technology and Archaeology. The Archae-
ological Applications of GIS. Taylor and Francis, London and New York.