Journal of Archaeological Science: Reports 22 (2018) 32–44

Contents lists available at ScienceDirect

Journal of Archaeological Science: Reports

journal homepage: www.elsevier.com/locate/jasrep

Landscape applications of photogrammetry using unmanned aerial vehicles T

James O'Driscoll
Archaeology Department, University of Aberdeen, Scotland, United Kingdom of Great Britain and Northern Ireland

A R T I C LE I N FO A B S T R A C T

Keywords: Photogrammetry is quickly becoming an important, cost effective technique for recording cultural heritage.
Photogrammetry Beyond the micro-scale of site evaluation, however, there are also effective landscape applications, with drone-
Drone based image collection allowing for large-scale survey. This combination of highly portable technology, which is
GIS not fully automated, can be used to create accurate and dense three-dimensional models at a fraction of the cost
Topographical survey
of LiDAR, and often at a much high spatial resolution. Yet, despite this, few studies have assessed the viability of
Hillforts
Landscape archaeology
this technique in regard to landscape studies. Those that have, such as Muñoz-Nieto et al. (2014), highlight the
effectiveness of this technique and its ease of use. This paper assesses the viability of this technology for mapping
large archaeological sites such as hillforts, providing a case study for its application to landscape archaeology.

1. Introduction manipulated in a number of ways. The digital environment can be

Photogrammetry is quickly becoming an important, cost effective
technique for recording cultural heritage, with particularly impressive
and innovative applications in excavation and artefact recording (De
Reu et al., 2014; Dellepiane et al., 2013; Roosevelt et al., 2015; Lerma
and Muir, 2014). Beyond the micro-scale of site evaluation, however,
there are also effective landscape applications, with drone-based image
collection allowing for large-scale survey. This combination of tech-
nology can be used to capture overlapping geo-referenced vertical and
oblique photographs in order to create accurate and dense three-di-
mensional models at a fraction of the cost of LiDAR, and often at a much
high spatial resolution. Furthermore, the hardware is highly portable,
allowing the user to access a wider range of sites. The surveys have
become fully automated and modern structure from motion software
allows the user to create ortho-rectified aerial photographs and DSM's
(Digital Surface Model) at the click of a button. Yet, despite this, few
studies have assessed the viability of this technique in regard to land-
scape studies. Those that have, such as Muñoz-Nieto et al. (2014),
highlight the effectiveness of this technique and its ease of use. This
paper assesses the viability of this technology for mapping large ar-
chaeological sites such as hillforts, providing a case study for its ap-
plication to landscape archaeology.
2. Drone survey and photogrammetry
Topographical surveys can contribute greatly to the identification of
previously unseen surface features and can be used to produce three
dimensional computer models of the surveyed area, which can be
viewed from any perspective, free of vegetation or features that may
obscure view. The height and direction of the sun can be manipulated
to create shading and shadow effects, making interpretation easier.
Micro-topographic surface features of buried or destroyed archae-
ological deposits occasionally exist, and can be identified with the ex-
aggeration of the vertical axis by a set multiple (Newman, 1997, 11).
Until recently, the acquisition of topographical data was undertaken
with a total station or D-GPS (Differential-Global Positioning Systems).
Barratt et al. (2000, 141) have speculated that a single total station
team would capture about 1000 points in a day, whereas one person
using a D-GPS systems can record up to 2000 points every hour. More
recently, LiDAR (Light Detection and Ranging) surveys have been used
for archaeological purposes to collect large amounts of elevation data.
This technique allows millions of points to be acquired in a matter of
hours at an accuracy of + or −7 cm vertically, and + or −15 cm
horizontally, usually with a resolution of 1–16 data points per metre
squared. The common weakness in these techniques is that data col-
lection and processing can be time consuming and costly.
Terrestrial photogrammetry has been used by archaeologists since
the early eighties (Fussell, 1982) but the prohibitive expense of hard-
ware and processing equipment meant that it was not a viable tech-
nique for most projects. With the advent of low-cost digital cameras and
increased computer processing power in the late 2000s, however, this
technique became a more viable option for artefact analysis and small-
scale site evaluation. However, even until recently, photogrammetric
surveys of monuments were undertaken using cumbersome methods
such attaching cameras to poles or kites and relying on an extensive
system of ground control points for software programmes to knit the

E-mail address: James.odriscoll@abdn.ac.uk.

https://doi.org/10.1016/j.jasrep.2018.09.010
Received 26 June 2018; Received in revised form 12 September 2018; Accepted 14 September 2018
2352-409X/ © 2018 Elsevier Ltd. All rights reserved.
J. O'Driscoll
photographs together (see McCarthy, 2014).
In the past few years, the rapid development of drones and structure
from motion software, as well as the substantial decrease in their cost,
has made them a viable option for archaeologists. While they have
quickly become the go-to platform for taking site and excavation pho-
tographs, the ability of this technology to further enhance archae-
ological investigations has not yet been fully realised. Even low-cost
drones aimed at casual, non-professional users can now be integrated
with free mapping and flight planning software for automated flight
and data capture to create highly detailed three dimensional models.
There are a number of problems with this approach, most notably
the inability of the technique to penetrate vegetation and tree canopy,
and the absolute accuracy of the collected data. The former is also a
problem for LiDAR (though usually not promoted as such), with data
collected in environments with dense conifer forestry often being
unusable (for example, see O'Brien and O'Driscoll, 2017, 214–215).
Absolute accuracy depends on the type and accuracy of the GPS in-
tegrated with the drone. With non-professional consumer drones this
can vary considerably and can sometimes be in the range of + or
−50 m or more. If accessing the erosion rate of a coastal monument, for
example, where multiple surveys of the same environment are needed,
this is inadequate and a much higher absolute accuracy is needed.
There are a number of different options to correct this, such as setting
up control points with a DGPS or using smart ground control points like
AeroPoints, buying a RTK-GPS enabled drone, or georeferencing the
collected data in GIS. However, if accessing an individual site in a one
off survey, the relative accuracy of the data is more important. This
depends on the height of the drone during survey and quality of the
camera used, but is usually between 1 and 2 cm, which is more than
accurate to produce a highly detailed and accurate model.
3. Practical applications of drone survey
For this study, a DJI Mavic Pro was used. This drone weights
0.734 kg including the in-built camera, battery and gimbal and can be
folded into a 0.083 m (height) by 0.083 m (width) by 0.19 m (length)
package, making this a highly portable unit. Using the freely down-
loadable Drone Deploy app on a mobile phone or tablet, the software
connects to the Mavic Pro via a USB cable linked to the controller. Once
the app automatically connects to the drone, the user then creates a
project by zooming into the area of interest on a satellite map and
creating a polygon that outlines the extent of the survey area. The user
then assigns the height of the drone, which determines the resolution of
the model and number of data points per metre squared. This allows the
software to automatically generate a flight path, with the number of
traverses increasing as the user decreases the height of the drone. The
direction of these traverses can also be adjusted if necessary. The per-
centage of overlap is automatically set to 65% sidelap and 75% fron-
tlap, though for more accurate readings the user should increase both to
at least 80%. Once these settings have been input, the user simply
uploads the flight plan to the drone via the upload button. Once this is
completed, the user selects ‘begin collection’ and the drone will auto-
matically take off, rise to the desired height, and move to the first
traverse where it will automatically begin taking images (Fig. 1). The
user can view where the drone is on their phone/tablet and also view
what is visible from the camera. A survey can be planned in a matter of
minutes in the field, allowing for archaeologists to collect survey data
on the fly (pun intended!).
Like many forms of modern digital archaeology, photogrammetry is
easily implemented. It is therefore easy to view this technology in terms
of its inputs and outputs, without any knowledge of its internal work-
ings. The algorithms and processes used by photogrammetry pro-
grammes are complex and are continually developing. It is not the
concern of this paper to summarise the evolution of this technique in an
archaeological context (see Remondino, 2014 for a comprehensive
outline). Instead, the author proposes to quickly outline the basic
33
Journal of Archaeological Science: Reports 22 (2018) 32–44
methodologies used by photogrammetry programmes as well as how to
use this software in a practical setting.
The collected images are imprinted with XYZ coordinates, allowing
the photogrammetry software to locate where the photographs were
taken in three-dimensional space, although this is not essential for the
software to produce good quality outputs (Many non-commercial
drones cannot geo-tag photographs and can still produce excellent
quality DSM's). The software is able to create a three-dimensional
model due to the multiple overlapping images at slightly differing po-
sitions, giving each photograph a unique angle of the select area/object
while also providing it with keypoints (reference points) visible to
multiple photographs. As such, it is important that multiple photo-
graphs overlap (hence the increase in the percentage of overlap applied
above). In instances where there is sufficient overlap, the number of
keypoints per pixel can be in excess of 10,000. This is more than suf-
ficient to derive an accurate model.
For software programmes such as Pix4D, the creation of a DSM and
orthomosaic is as simple as dragging the photographs into the folder
(Fig. 1). Once these have been uploaded to the cloud, the programme
automatically begins to process the data, giving outputs such as a DSM,
orthomosaic, LAS file, Mesh OBJ file, etc. Depending on the variables
outlined above, a survey can have a resolution in excess of hundreds,
and sometimes thousands of points per square metre. This programme
will also automatically create a three dimensional model that you can
view and manipulate immediately after completion of the processing.
Similar to the way setting-up and undertaking image capture with a
drone has become a straightforward and easy process, the practical use
of the photogrammetry software is also uncomplicated. While these
outputs can immediately be manipulate in GIS without any processing
from the user, programmes such as Pix4D does allow users to further
enhance or clean-up data, but in many instances this is just for aesthetic
reasons. The remainder of this paper will outline a number of different
case studies to highlight the applicability of this new survey technique
to landscape studies.
4. Case study: Cahercommaun, Co. Clare, Ireland
Cahercommaun fort is position at the southwestern edge of
Tullycommon townland in Co. Clare, Ireland (Fig. 2). The site is c.
145 m above Ordnance Datum and overlooks a deep north–east/
south–west ravine, approximately 30 m in height. The fort itself is ap-
proximately 0.68 ha in total area and comprises three widely spaced
enclosing elements that abut the edge of the ravine to the north. These
defend the eastern, southern and western approaches to the fort, while
the north is protected by steep natural cliff face. The monument was
partially excavated by Hugh O'Neill Hencken (1938) in 1934, who ar-
gues that the site was an important centre of a regional Early Medieval
chiefdom known as Tulach Commain.
Cahercommaun is more broadly situated within the extensive karst
landscape of the Burren. This environment was formed in the last gla-
cial maximum, when ice sheets eroded the overlying soil and exposed
the limestone surface. The lack of soil cover is less suitable for intensive
farming which has led to the survival of much of the ancient landscape,
including significant Early medieval remains (Hull and Comber, 2008).
Survey of the site was funded by a Royal Irish Academy grant which
aimed to record in detail the Western Stone Forts on the tentative list of
UNESCO World Heritage sites in Ireland. In total, 166 geo-tagged aerial
images were captured, resulting in the generation of a DSM and or-
thomosaic 11.69 ha in size (Figs. 3–4). This took approximately 17 min
to collect in-field. The DSM consists of 13,519,874 three-dimensional
data points with an average density of 186.84 points per square metre.
This data was then processed in GIS to produce hillshade (Fig. 5) and
slope (Fig. 6) models which were analysed to create a plan (Fig. 7) of
the visible archaeological features.
Cahercommaun is a particularly good case study, as there have been
a number of surveys undertaken to record the layout of the fort. Despite
J. O'Driscoll Journal of Archaeological Science: Reports 22 (2018) 32–44

Fig. 1. Workflow of survey planning and data collection.

Fig. 2. Aerial photograph of Cahercommaun fort, Co. Clare, Ireland.

34
J. O'Driscoll Journal of Archaeological Science: Reports 22 (2018) 32–44

Fig. 3. DSM of Cahercommaun fort, Co. Clare, Ireland, derived from remotely captured aerial photographs.

this, the photogrammetry survey identified a number of unrecorded relevance was the use of GIS visibility analysis (Fig. 8). This highlighted
features, including levelled field systems and possible hut structures. that views from the interior of the inner enclosure are restricted by the
These became particularly visible when using GIS analytical techniques substantial defences even at the north, where the wall is less sub-
such as slope or hillshade analysis to view the data. Of particular stantial. Access to the top of these walls is provided by steps built into

Fig. 4. Orthomosaic of Cahercommaun fort, Co. Clare, Ireland, derived from remotely captured aerial photographs.

35
J. O'Driscoll Journal of Archaeological Science: Reports 22 (2018) 32–44

Fig. 5. Hillshade model of Cahercommaun fort, Co. Clare, Ireland, derived from remotely captured aerial photographs.

the inner face. Visibility analysis highlights the extensive views from explain why there is no evidence for terracing in the middle and outer
this area, with particularly excellent views of the eastern and western enclosing elements, as such features may have been practical elements
approaches to the fort, as well as visibility of the valley to the north. incorporated into the defences of the inner wall to augment the visi-
The heightened visibility from the top of the inner enclosure may bility of the immediate landscape to those inside the interior. As such, it

Fig. 6. Slope model of Cahercommaun fort, Co. Clare, Ireland, derived from remotely captured aerial photographs.

36
J. O'Driscoll
Fig. 7. Plan of features recorded by the photogrammetry survey at Cahercommaun fort, Co. Clare, Ireland.
was not necessary to incorporate such features, allowing the walls to be
thinner which, in turn, resulted in reduced costs in labour and re-
sources. In fact, the practicalities of the walls themselves come into
question when we consider the viewshed between the inner and outer
enclosure. Views from this position are restricted by the outer wall,
which in antiquity would have been taller than what is visible today.
This could indicate that the outer two enclosing elements may have
been constructed primarily to convey the status and power of the oc-
cupants to those approaching the fort, though they could have also had
a number of subsidiary functions, such as cattle management and de-
fence.
At Cahercommaun, the photogrammetery survey has provided us
with a better understanding of the site. While the enclosing elements
could be regarded as defensive in nature, the photogrammetry survey
and viewshed analysis highlights that the height of the outer wall would
have restricted the visibility of people located between the inner and
outer defences. It may be more plausible to regard the middle and outer
defences principally as status symbols that could have had a secondary
defensive function.
5. Case study: Glanbane, Co. Kerry, Ireland
Glanbane hillfort in Co. Kerry, Ireland is positioned on the eastern
edge of the Slieve Mish Mountains overlooking a broad valley that
connects the bays of Dingle and Tralee and sits at approximately 86 m
Ordnance Datum (Fig. 9). The site comprises two widely spaced en-
closing elements that occupy an area of 6.58 ha, though these have been
completely levelled and are only visible as crop marks in some aerial
photographs. Geophysical survey mapped the extent of these defences,
as well as the interior of the fort. This revealed a number of possible
circular wooden structures within the interior, including a large, 45 m
post structure positioned near the centre of the interior partially trun-
cating a 55 m earthen enclosure (O'Brien and O'Driscoll, 2017). Trial
excavations over the two enclosing elements revealed that each con-
sisted of a rock-cut ditch up to 3.1 m wide and 1.35 m deep, with an
37
Journal of Archaeological Science: Reports 22 (2018) 32–44
internal stone faced bank. Radiocarbon dates from the basal layers of
the internal ditch indicate that Glanbane hillfort was constructed
around 1150–900 BCE and is therefore broadly contemporary with the
majority of other large hilltop enclosures in Ireland known as hillforts.
Considering that we know the extent and layout of the enclosing
elements at Glanbane and there is little in the way of any topographical
low-relief features apparent on the surface, this provides an excellent
case study for the application of drone and photogrammetry technology
to map and record levelled archaeological sites. In total, 235 geo-tagged
aerial images were captured, resulting in the generation of a DSM and
orthomosaic 21.43 ha in extent (Fig. 10). This took approximately
21 min to collect in-field. The DSM consists of 20,797,450 three-di-
mensional data points with an average density of 200.5 points per
square metre.
The results of the survey revealed the extent of the inner enclosing
element, with the outer example being less extensively mapped and
topographical features visible only on the north, northwestern and
southeastern sides. This suggests that the internal bank ranged from 7
to 10 m wide and the external example 6–7.5 m wide. This is important
additional information, as neither excavation nor geophysical survey
conclusively confirmed the extent of these features. We can now suggest
that banks of the hillfort would not have been more than 1–1.5 m high
(considering a combination of the width of the banks and the depth of
the external ditches, which presumably would have been the main, if
not sole, source of material for its construction), though there is the
possibility that this would have been topped with a wooden palisade
and evidence for this has been destroyed. The photogrammetry survey
was also useful in clarifying the presence of an entrance at the north-
east. This is apparent as a tapering of the two enclosing elements, where
the average separation between the features narrows from 16 m to 8 m.
There is also a possible corresponding 6 m break in both enclosing
elements, though this is difficult to confirm due to obstruction from a
modern farm track. This probable entrance appears faintly in the geo-
physical survey results, though not as clearly and well defined as in the
photogrammetry survey. The survey was also useful in helping us to
J. O'Driscoll Journal of Archaeological Science: Reports 22 (2018) 32–44

Fig. 8. GIS viewshed analysis of topographical data from drone derived data at Cahercommaun, Co. Clare, Ireland.

identify that the large structures near the centre of the fort are located Haughey's Fort, Co. Armagh is a good comparison for Glanbane. Here, a
at the highest point of the interior, something which is not noticeable large 25–30 m wooden structure delimited a series of substantial pits
considering the flat nature of most of the site (Fig. 11). containing carbonized grain, coarse ware pottery, fragments of quern
The survey at Glanbane has provided important insights regarding stones, metal artefacts and fragments of bronze and gold (Mallory,
the layout and possible size of the enclosing elements and positioning of 1995, 78; Mallory and Warner, 1988), leading Mallory (1995, 84) to
the central structures within the interior. The results revealed a possible argue that ‘defended as it may be, [Haughey's Fort] was primarily the
entrance at the north-east comprising a narrowing of the defences and focus of ritual activities’ (Mallory, 1995, 84). A similar interpretation
corresponding gaps. This is an important discovery, as nearly all of the could be ascribed to Glanbane.
recorded Bronze Age hillforts in Ireland have simple entrances con-
sisting of small corresponding breaks in the defensive works, with rare
6. Case study: Turin Hill, Angus, Scotland
example of more complex entrance at such sites like Carn Tighernagh,
Co. Cork possible being a later refurbishment of the fort in the Iron Age
Turin Hill in Angus, Scotland is a complex of fortifications com-
or Early Medieval period. The ‘pinching’ of the enclosing elements near
prising two hillforts and three smaller ringforts located on the highest
the entrance at Glanbane is a unique feature only found at this site. The
point of a roughly east-west running ridge of hills overlooking an ex-
survey also revealed the width of the hillfort banks, which may infer
pansive tract of land to the north, east and west (Fig. 12). The earliest
that the height of the earthworks would not have been substantial.
fort at the site occupies an area of 4.68 ha and is defined by a low relief
However, considering the combination of the depth of the ditches (up to
bi-vallate earthwork comprising a bank-ditch-bank-ditch arrangement
1.35 m) and possible height of the banks (1–1.5 m), these features may
on the western side, with a series of natural craigs defending the eastern
have still formed a practical defensive barrier 2–3 m in height. The
approach. Inside this fort, on the southern side, is an oblong enclosure
presence of the large structures at near the centre of the fort at the
with a single robbed out stone wall occupying an area of 0.67 ha. Three
height point of the interior may be a hint of its ritual nature and
smaller ringforts are located at the crest of ridge each occupying areas
function. Structural evidence from the contemporary Irish hillfort of
of 0.085 ha, 0.09 ha and 0.1 ha respectively.

38
J. O'Driscoll Journal of Archaeological Science: Reports 22 (2018) 32–44

Fig. 9. Aerial photograph of the levelled hillfort at Glanbane, Co. Kerry, Ireland.
(Bing Maps 2018)

Fig. 10. Slope model of Glanbane, Co. Kerry, Ireland, derived from remotely captured aerial photographs.

39
J. O'Driscoll Journal of Archaeological Science: Reports 22 (2018) 32–44

Fig. 11. Contours derived from photogrammetry derived model with results of geophysical survey overlain.

Fig. 12. Aerial photograph of the complex of fortifications on Turin Hill, Angus, Scotland.

40
J. O'Driscoll
Fig. 13. Slope model of Turin Hill, Angus, Scotland, derived from remotely captured aerial photographs (top), compared with slope model of equivalent area derived
from LiDAR data.
The wealth of archaeological features on the hill has attracted nu-
merous researchers, with a number of ground surveys having been
undertaken. The most detailed plan was produced by Alexander and
Ralston (1998) who recorded the enclosing elements of the various
forts, as well as identifying unrecorded hut platforms and additional
earthworks. More recently, LiDAR data has also become available for
the site. The available plans and topographical data for Turin Hill,
therefore, makes this an excellent case study for the application of
drone and photogrammetry technology, where we can assess if these
41
Journal of Archaeological Science: Reports 22 (2018) 32–44
new techniques are accurate enough to identify the archaeological
features already recorded or map any new features, while also allowing
comparisons to be made with other data collection methods like LiDAR.
Survey of the site was funded by the Leverhulme funded
Comparative Kingship Project. A total of 342 geo-tagged aerial images
were captured, resulting in the generation of a DSM and orthomosaic
12.76 ha in extent (Fig. 13). This took approximately 32 min to collect
in-field. The DSM consists of 26,578,680 data points with an average
density of 500.8 points per square metre. The survey results allowed for
J. O'Driscoll
Fig. 14. Hillshades of Turin Hill derived from LiDAR (right) and photogrammetry (left) data.
a more accurate assessment of the visible surface remains and clarified
the relative chronology of the fortifications on the hill. Previous plans
and the more coarse LiDAR data were unable to clearly demarcate if
and where the individual forts overlap (Figs. 13 and 14). In contrast,
the photogrammetry results clearly reveal that the large bi-vallate
hillfort (T1 and T2) is truncated on the western side by the oblong fort
(T5), which itself it truncated by the central ringfort (T7) (Fig. 15).
Similarly, although a series of small upstanding mounds (T10) are also
apparent in the LiDAR survey, up to five additional examples were
identified in the photogrammetry results, while three more hut struc-
tures and an additional part of the enclosing elements (T9) were only
visible with photogrammetry.
The survey also revealed interesting aspects relating to the defences
of each fort. For example, the builders of the larger bi-vallate hillfort
seemed to have positioned the enclosing elements on a natural break of
slope to enhance their appearance and apparent size when viewing or
approaching the site from the exterior. The survey also recorded linear
depressions at the crest of the inner bank, possibly indicating the pre-
sence of a palisade slot (which is supported by the geophysical survey
results). In some areas, parallel linear features approximately 3.35 m
apart define the limits of external bank and this could represent the
inner and outer wall face. An unrecorded low-relief bank (T3) extends
from the western side of the outer enclosing element, before dividing in
two, with the inner bank extending towards the east–west running craig
to the south while the outer bank extending to incorporate the wes-
ternmost ringfort. This is also visible in the LiDAR data. The earthwork
Fig. 15. Plan of features recorded by the photogrammetry survey at Turin Hill, Angus, Scotland.
42
Journal of Archaeological Science: Reports 22 (2018) 32–44
clearly incorporates the westernmost ringfort (T6), which is probably
later than the large bi-vallate hillfort. We might therefore suggest that
this newly discovered bank is a later addition, built to visually in-
corporate the ringfort into the design of the larger fort in order to
substantiate the legitimacy and status of the ringfort occupants through
direct visual links with the past.
On the lower craig overlooked by the forts, T9 represents the
earthwork recorded by Alexander and Ralston (1998), however, a
second low-relief feature is apparent in both the LiDAR and photo-
grammetry data. This is in-line with the T6 extension, and effectively
cuts off access to the lower craig, creating a variation of a nuclear
hillfort. These types of forts are apparent in Scotland from the Early
Medieval period, a date that would correspond well with the probable
construction of the three small ringforts on the summit.
The photogrammetry survey recorded the oblong fort and three
ringfort, all of which are at risk of damage by local agricultural activity,
making detailed survey of these sites timely. However, little in the way
of additional information was gained from the photogrammetry results.
Scattered throughout the interior of the bi-vallate hillfort are a
number of possible hut platforms. Five of these have previously been
recorded by Alexander and Ralston and three are visible in the LiDAR
data. In contrast, eight possible hut sites were identified in the photo-
grammetry results, comprising circular or oval terraces approximately
5–14 m in diameter cut into the natural slopes of the hill. Furthermore,
up to 14 raised flat-topped platforms approximately 4–7.5 m in dia-
meter are apparent in the photogrammetry results that were not
J. O'Driscoll
recorded by Alexander and Ralston. These features are also visible in
the LiDAR data though they are less defined. It is only in the photo-
grammetry data that we can accurately measure the size and shape of
these features. These platforms may represent some form of structures,
where the build-up of material over time raised the floor levels, re-
sulting in these distinctive features visible on the surface. This inter-
pretation is partly supported by recent geophysical survey at the site.
Despite the previous earthwork and topographical surveys at Turin
Hill, detailed photogrammetry survey revealed a number of previously
unrecorded archaeological features that have transformed our under-
standing of the site. We can now confirm the relative chronology of the
fortifications on the hill: the large bi-vallate hillfort being truncated by
the oblong fort which in turn is truncated by the central ringfort
(though this does not necessarily mean the other two small ringfort are
contemporary with this). We might also recognise the site as a type of
nuclear hillfort, considering the newly discovered earthworks on the
lower craig to the east protects this area and creates a terraced division
(possibly reflecting divisions in the status of those living on the hill)
that defines this type of fortification. We can also argue that the hill was
occupied by domestic structures, although these cannot be linked with
any specific phase of use of the hill, as without excavation, it is im-
possible to speculate regarding their chronology.
7. Summary of results
As demonstrated in the case studies above, this technology can be
used to significantly enhance our understanding of a site, even where
these has already been detailed earthwork surveys or where LiDAR data
is available. At Cahercommaun in Co. Clare, Ireland, for example, the
photogrammetry results were used in GIS to demonstrate the im-
practical characteristics of the outer enclosing elements as defensive
works, allowing us to interpret the impressive walls as a mark of the
status and power of the occupants.
The survey at Glanbane hillfort in Co. Kerry, Ireland, demonstrated
the ability of drone and photogrammetry technology to accurately de-
tect and map the extent of low-relief topographical features that are
almost imperceptible at ground level. This allowed us to confirm the
presence of an entrance at the north-east of the hillfort, with evidence
of some form of elaboration not usually found at contemporary sites in
Ireland. Other minute surface variations within the interior revealed
that the large structures near the centre of the fort were built at the
highest point on a slight rise, further emphasising the importance of
these features.
The LiDAR available at Turin Hill in Angus, Scotland provided an
excellent comparison for newly collected photogrammetry data. In
many cases, the LiDAR survey identified similar low-relief features,
though the photogrammetry results were at a much higher resolution,
allowing for more accurate recording and measurements to be ex-
tracted. In some instances, the photogrammetry survey identified fea-
tures not visible in the LiDAR data, including a number of probable
circular hut structures and minute variations in the form of the defences
of the large bi-vallate hillfort.
8. Conclusions
There are obvious disadvantages to photogrammetry, the most no-
table being its inability to penetrate vegetation and tree canopy.
However, without the proper conditions and sensors, LiDAR can also
struggle in these environments. There are also the stated issues with
regard to the absolute accuracy of photogrammetry data, thought there
are ways to alleviate these issues (see above).
Despite these problems, there are clear advantages to this type of
methodology, specifically those relating to cost and time, as well as ease
of use, portability and the ability to do on-the-fly surveys. The in-
tegration of drones has allowed for exponentially larger areas to be
survey much quicker than ground based methods of data capture, with
43
Journal of Archaeological Science: Reports 22 (2018) 32–44
far greater accuracy and resolution. We now have the ability to examine
large-scale monuments and landscapes using these techniques and as
such, this technology should be seen as a serious alternative to other
more conventional forms of topographical data collection.
As well as doing surveys of monuments and environments as those
identified above, there are numerous archaeological applications to this
type of survey, such as management and mapping of coastal archae-
ology at risk of erosion or built heritage, for example. Furthermore, the
collected data can be made available to the public through various web-
hosts such as Sketchfab, facilitating more accessibility and increasing
engagement.
The rapid development of drones and affiliated hardware is already
positively impacting archaeology in more conventional ways, with as
drone aerial photography being used for site detection, survey and
excavation, and is now positioned to make high resolution topo-
graphical data a viable option for many more archaeologists. Compact
LiDAR systems have recently been developed for drones, and though it
is currently prohibitively expensive, we will quickly see this technology
becoming more cost effective and widespread. There are also more
novel methods of site detection, such as hyper-spectral, multi-spectral
or thermal imaging, that can now become more regularly implemented
in the field via drones, allowing for more rapid means of sub-surface
remote sensing.
Conflict of interest statement
Compliance with ethical standards
This research complies with all ethical standards.
Funding
Royal Irish Academy Directed Research on UNESCO World Heritage
Sites; Leverhulme Trust.
Conflict of interest
The authors declare that they have no conflict of interest.
Acknowledgments
The author would like to thank the various colleagues who read
early drafts of this paper, as well as the two anonymous reviewers
whose comments and suggests helped to improve and focus the final
paper. I would also like to thank the Royal Irish Academy and the
Leverhulme funded Comparative Kingship Project for support.
References
Alexander, D., Ralston, I., 1998. Survey work on Turin Hill, Angus. Tayside Fife Archaeol.
J. 5, 36–49.
Barratt, G., Gaffney, V., Goodchild, H., Wilkes, S., 2000. Survey at Wroxeter using carrier
phase, differential GPS surveying techniques. Archaeol. Prospect. 7, 133–143.
De Reu, J., De Smedt, P., Herremans, M., Van Meirvenne, P., Laloo, P., De Clercq, W.,
2014. On introducing an image-based 3D reconstruction method in archaeological
excavation practice. J. Archaeol. Sci. 41, 251–262.
Dellepiane, M., Dell'Unto, N., Callieri, M., Lindgren, S., Scopingno, R., 2013.
Archaeological excavation monitoring using dense stereo matching techniques. J.
Cult. Herit. 14, 201–210.
Fussell, A., 1982. Terrestrial photogrammetry in archaeology. World Archaeol. 14 (2),
157–172.
Hencken, H., 1938. Cahercommaun: a stone fort in County Clare. R. Soc. Antiqu. Irel. 68,
1–82.
Hull, G., Comber, M., 2008. Caherconnell, Co. Clare and cashel chronology. Archaeol.
Irel. 22 (4), 30–33.
Lerma, J., Muir, C., 2014. Evaluating the 3D documentation of an early Christian upright
stone with carvings from Scotland with multiple images. J. Archaeol. Sci. 46,
311–318.
Mallory, J., 1995. Haughey's Fort and the Navan Complex in the Late Bronze Age. In:
Waddell, J., Shee Twohig, E. (Eds.), Ireland in the Bronze Age. The Stationary Office,
Dublin, pp. 73–86.
J. O'Driscoll
Mallory, J., Warner, R., 1988. The date of Haughey's Fort. Emania 4, 36–40.
McCarthy, J., 2014. Multi-image photogrammetry as a practical tool for cultural heritage
survey and community engagement. J. Archaeol. Sci. 43 (3), 175–185.
Muñoz-Nieto, A., Rodríguez-Gonzálvez, P., González-Aguilera, D., Fernández-Hernández,
J., Gómez-Lahoz, J., Picón-Cabrera, J., Herrero-Pascual, J., Hernández-López, D.,
2014. UAV Archaeological Reconstruction: The Case Study of Chamartin Hillfort
(Avila, Spain).
Newman, C., 1997. Tara: An Archaeological Survey. Discovery Programme Monograph 2
(Dublin).
44
Journal of Archaeological Science: Reports 22 (2018) 32–44
O'Brien, W., O'Driscoll, J., 2017. Hillforts, Warfare and Society in Bronze Age Ireland.
Archaeopress, Oxford.
Remondino, F., 2014. UAV: platforms, regulations, data acquisition and processing. In:
Remondino, F., Campana, S. (Eds.), 3D Recording and Modelling in Archaeology and
Cultural Heritage: Theory and Best Practice. British Archaeological Report
International Series 2598pp. 73–88 Oxford.
Roosevelt, C., Cobb, P., Moss, E., Olson, B., Ünlüsoy, S., 2015. Excavation is digitization:
advances in archaeological practice. J. Field Archaeol. 40 (3), 325–346.