Content uploaded by Birger Stichelbaut
Author content
All content in this area was uploaded by Birger Stichelbaut on Sep 30, 2014
Content may be subject to copyright.
Method
The application of First World War
aerial photography to archaeology:
the Belgian images
Birger Stichelbaut1
The First World War left its mark on the ground surface of Europe as perhaps no other human
catastrophe before or since. The author applies modern digital mapping technology to the aerial
photographs taken by the intrepid early pilots, and creates a landscape of military works that would
not have been known in detail to either historians or generals at the time. The GIS inventory
has great potential for historians of the war and is a vital instrument for the management of this
increasingly important heritage.
Keywords: aerial photography, First World War, battlefield archaeology, Belgium
Introduction
The material remains of the First World War are fragile, and under continual threat from
modern land use. This study describes how a specific, non-destructive methodology can
offer new materials for archaeological and historical research into warfare and provide the
means of effective resource management. The method applies new cartographic technology
to photographs that were taken between 1914 and 1918. During the conflict, thousands
of aerial photographs were taken by both sides; they give accurate insights into the density,
distribution and location of military remains, and offer a fuller picture than the trench
maps made on the ground. The research described in this paper focuses on a small sector of
the Belgian Western Front, using aerial photographs mostly taken by the Aviation Militaire
Belge (AvMB), the Belgian air force, and largely covers the Belgian sector of the West
Flanders front line, between Nieuwpoort and Steenstraat (Figure 1).
The research programme had three stages. First, the origin, context and developments of
Belgian military aerial reconnaissance during 1914-1918 were traced. Then over a thousand
photographs were studied, digitised and loaded into a GIS. Finally, an inventory was
made of sites relevant to the Great War, as well as of more traditional archaeological
features. The methodology described below can be extrapolated to the whole Western
and Eastern Front since the primary sources, the photographs, are similar. The only
requirements are access to First World War aerial photographs, the use of GIS and accurate
digital topographical maps or cadastres, and red/green stereoscopic viewers and software.
1Department of Archaeology, Ghent University, Blandijnberg 2, 9000 Ghent, Belgium (Email: birger.stichelbaut@
ugent.be)
Received: 23 November 2004; Accepted: 3 May 2005; Revised: 2 March 2005
antiquity 80 (2006): 161–172
161
First World War aerial photography
Figure 1. Map of West Flanders with place names (dotted line indicates
the approximate position of the front line before the Second Battle of
Ypres). The front line between Bikschote and Nieuwpoort remained almost
unchanged during the war.
Historical background
After the First Battle of the
Marne on 5–11 September 1914,
the First World War became
static. Both sides started to
entrench their armies in the
800km long stretch of land
between the North Sea and the
French–Swiss border. The armies
soon realised the strength and
possibilities of a new weapon,
military aviation carrying out
aerial reconnaissance. Pilots and
observers became the eyes of the
army, a role up until then only
filled by the cavalry. From the
first weeks of the war, aeroplanes
of the AvMB, the Belgian air
force, were sent out to scrutinise
the German military movements.
During 1914, reconnaissance
reports were mainly made ‘at
sight’, using the observers’ eyes
and without cameras. The first
British and Belgian photographs
were taken on 15 and 23 Septem-
ber 1914 (Lampaert 1997:
35-6; Delve 1997: 128). During
the first months of the war, aerial
photography was considered to
be a hobby of a few enthusiastic
airmen. From 1915 onwards the new discipline developed and became widely used by the
military air services.
The infantry command used the photographs of the front-line trenches to get information
for the preparation of raids on enemy trenches and patrols into no-man’s land, to detect
preparations for an offensive (troop movements, ammunition storage, trains, etc.), or to
monitor their own preparation of attacks (looking for weak spots, locating strong defensive
positions, etc.). A second category of photographs was taken during artillery missions.
Photographic interpreters used these for tracing camouflaged artillery positions and locating
other possible targets. Photographs were also useful for verifying the accuracy and results
of the bombardments (Desmet 1921: 40). Other regions were photographed regularly, for
instance, the flooding of the IJzer river and the enemy’s defences and trench fortifications.
The Allies’ own trenches and positions were also photographed, for checking levels of damage
162
Method
B. Stichelbaut
and verifying camouflaged positions. Most images were also used for the production of trench
maps at different scales.
Method
The Belgian First World War photographs have a number of formats depending on the
camera used (13 ×18cm camera with a focal length of 26cm; 18 ×24cm camera with focal
lengths of 52 and 120cm; Anon 1925: 7) but are all panchromatic. The quality of the
pictures can vary enormously and was subject to weather conditions and the stability of
the aircraft, but also on the chemicals used for developing. The pictures are panoramic,
oblique or vertical, depending on what information had been requested. German, French
and British photographs mostly have the same measurements and characteristics of quality
and production. The air force which took these pictures can be identified by the different
markings and numbers on the pictures (i.e. Fl.Ab: Flug Abteilung; MF 52: Escadrille Maurice
Farman 52) but also, for example, by the type of north-facing arrow.
The research focused on 1128 photographs from the KLM/MRA and SGRS-S/A
collections. Each photograph was entered into a database (MS Access), which gave also
archive information about the picture, a description and map reference of any visible
features. The pictures were then digitised, located and loaded into a GIS using Esri ArcView
GIS 3.1 and ILWIS 3.2 Academic. It proved possible to find the precise location of 1039
photographs (92.1 per cent of the 1128 images in the group studied). This corresponds
to an area of 238.6km2. The sites of the largest group of photographs (72 per cent) were
mapped by drawing the four corners of every picture in the same ArcView file as a polygon.
The other 294 photographs were first rectified and georeferenced with exact coordinates
using Image Warp.
Several problems were encountered when trying to identify sites in the photographs.
Belgian and French photographers used many place names that were only used during the
war, although examining wartime trench maps could retrieve many of them. There was
often some difficulty in reconciling the First World War aerial photographs, which show
the landscape in the early twentieth century, with present-day digital maps and cadastral
records which represent the situation no further back than the early 1990s. The division of
land in some areas has changed due to re-allotments and village expansion. Other sites can
be hard to locate because of the lack of usable reference points: some aerial photographs
show only innumerable mud-filled craters. These difficulties were, however, random and
local; the bulk of the group of images studied could be given accurate locations.
Some military structures were hard to detect by their nature. Because of the increasing
number of reconnaissance flights from 1915 onwards, the infantry and artillery learned
to hide their structures from aerial observation with camouflage. They built fake or
dummy constructions, and also misled the enemy by covering buildings and camps with
vegetation, painted canvas and earth. This was mostly carried out on bunkers and artillery
emplacements. Aerial observers soon developed a new technique to find the camouflaged
positions: stereoscopy, used for the first time in a rudimentary form in 1915 (Jaumotte
1919: 162). During reconnaissance flights, two successive photographs were taken of the
same area to make a stereo pair, the images ideally overlapping by 60 per cent. The common
163
First World War aerial photography
Figure 2. Anaglyph of the Esen area (500 ×500m), 2 February 1917. To see this image properly, the green part of stereoscopic
glasses needs to be held on the left. In this case the trenches are constructed with high breastworks, but without stereoscopy this
would be difficult to discern. The ditch of a moated site next to the road and trenches shows up very clearly. In the upper right
corner there is a bunker. Beneath the trenches there are several artillery emplacements.
parts of the pictures were cut out and oriented according to fixed rules (Jaumotte 1919:
164). By looking at this with a stereoscope, an artificial and exaggerated image of the image
area could be created.
Ninety years later, it was still possible to find overlapping photographs and view the terrain
in elevation — and to discover emplacements that otherwise would remain hidden. Digital
stereoscopic views (anaglyphs) were created from these pictures using ILWIS 3.2 Academic.
The common parts were superimposed with the left and right picture respectively in the red
and green colour spectra. The results can only be seen with a pair of (green–red) stereoscopic
glasses. Raised objects such as bunkers and artillery positions can be perceived in 3D relief.
The same can be done for mine craters and other concave features. An example of an anaglyph
can be seen in Figure 2, the German trench system at Esen (2km east of Diksmuide).
All sites were inventoried in ArcView GIS 3.1. The georeferenced pictures were first
inserted in an ArcView projection and the modern-day cadastral map was laid above them
164
Method
B. Stichelbaut
to assess the accuracy and usability of each photograph. Every trace is listed with an accuracy
of at least 5m. The next step was to copy the traces on the photographs to a digital ‘layer’
in GIS. By this, the exact geographical location of each trace could be identified, and also
its length, width and surface area. To get the maximum information the traces had to be
drawn as polygons instead of lines. This was time consuming, but as a method it was able
to contain all the geographical data available, unlike lines.
All the information about the inventoried sites (nationality, inventory number, dating,
surface area, perimeter, coordinates, photograph used, and an interpretation) was stored
in the ArcView ‘Theme Table’, a separate database from the database holding the basic
information of the photographs from phase 1 of the study. While inventorying the
Diksmuide testing area, 3755 different traces were recorded (Stichelbaut 2005). A partial
view from this inventory is reproduced as Figure 6.
Interpretation
In traditional archaeological aerial photographs, most sites are detected as water, snow, crop
and soil marks, and refer to events from past periods. In this case, the majority of the features
being studied belonged to the First World War and were of similar date to the photographs
that had captured them. The photographs had potentially captured information of three
kinds: environmental features, prehistoric and historic sites, and structures and relics related
to the First World War. As a contribution to environmental history, the photographs showed
the extension of the dunes, land use and afforestation at the Front, and the remains of field
systems still visible at the beginning of the twentieth century, and now destroyed. Among
the recorded historical and archaeological traces there are many moated sites and several
post-medieval French and Spanish fortifications like forts and bastioned town walls.
The majority of the recorded features can be assigned to the war, and comprised a
considerable number of defensive and offensive structures and positions. A distinction
could be made between infantry sites, artillery emplacements and structures well behind the
trenches. There were regional differences in the appearance of military features, often related
to geological and topographic conditions. In addition there were several known structures
that did not appear in the battlefield area studied, for example, deep dugouts and mines.
Some of these were picked up using stereoscopy (see above).
Results
In this war, the most common feature observed on photographs of the front line was the
trench system. Trenches can easily be discovered on most photographs, both vertical and
oblique. Made of disturbed earth, their light colour appears clearly against the darker intact
soil holding vegetation. Between the beige parapet and parados, a dark shadow marks the
actual trace of the trench. The longer they are in use, the duller the colour becomes. In
snowy landscapes, trenches show up as dark lines, produced by melting snow. Dummy
trenches were made to mislead wartime air reconnaissance. Their trace is similar to normal
trenches, but they were not constructed as deeply. They can be easily distinguished by their
missing shadows. In woodland, the visibility of trench systems decreases enormously because
165
First World War aerial photography
of shadows and leaf cover. The way around this is to study photographs of the area taken in
theautumnorwinter.
The entrenchments were laid out in three or more broadly parallel lines (fire, support and
reserve trenches; Ellis 1989: 16). The lines were rarely straight, being marked by frequent
fire bays and traverses. The first group of trenches are the fire trenches or tranch´
ees.In
general these trenches are oriented parallel with the front line and have a complex typology
determined by hydrological circumstances, terrain, objectives, visual range, the nationality
of their original engineers and of subsequent occupants who may have modified their layout.
Commonly seen are the redan (salient), traverse (zig-zag), bastion (corner fortification) and
tenaille (advanced outwork). Within each category there are variations according to length
and form of the traverses. An example of two German stratified fighting trenches with round
and square traverses can be seen in detail in Figure 3b. At right angles to the front line are
the communication trenches or boyaus which connect the lines. Often they are much longer
than fighting trenches and their route is generally more sinuous. Traversed, indented and
zigzag (see Figure 3d) routes can be recorded in this category.
Barbed wire entanglements were laid out in front of trenches in successive wide belts
with intervals of between 30 and 100m (Anon 1921: 48, 1998: 92) and in conjunction
with machine gun posts. Entanglements are only visible on high quality images taken on
sunny days (see Figure 3a) or after snowfall. Sometimes the entanglements become visible
because of the geometrical alignment of the stakes (see Figure 3c). In other cases the belt is
noticeable as a dark line, caused by the shadow of the barbed wire (see Figures 2 and 3a:
diagonal line).
Protective concrete bunkers were used for diverse purposes, not always discernible by aerial
photography, since their function was determined by their internal organisation. Most of the
bunkers in the front line were shelters for personnel, artillery and machine guns. Some can
be identified as housing German Minenwerfers because of their rectangular openings in the
roof for firing trench mortars. Most concrete structures in photographs have a light-coloured
shape and are often camouflaged. The presence of artillery can sometimes be deduced by
the proximity of a narrow-gauge railway and blast marks. The use of stereoscopic analysis
can give more information about the nature of their construction and reveal well-hidden
structures.
Narrow gauge or Decauville railways have widths of 0.80, 0.60 and 0.40m. They were used
to supply ammunition and building materials to the trenches and artillery emplacements.
On the air photographs they are recognisable as long, light-coloured lines in the territory
behind trenches and at the front line. They differ from the wooden infantry bridges and
duckboards because of their curved track, and occasionally the iron tracks and wooden
traverses are also visible. An example of a Decauville railway can be seen in Figure 3b.
In advance of both the Allied and German front lines there were listening posts in no-man’s
land armed with machine guns. These forward positions were connected to the trenches
with bridges and paths. In comparison to other elements these small structures show up on
the photographs as small round spots, sometimes horseshoe-shaped. Their position is often
related to passages through barbed wire entanglements.
Identifying artillery emplacements on aerial photographs is complex, as they are most likely
to have been camouflaged. The correct approach is by a comparative study of photographs
166
Method
B. Stichelbaut
Figure 3. Trench system north-west of Bikschote, Belgian AP taken on 8 April 1917. Archive KLM/MRA. Aerial photos box
‘Bikschote B’. (a) General view, (b) detail of two fighting trenches and narrow gauge railway, (c) barbed wire entanglements
and (d) communication trench and two artillery emplacements at the bottom. Moated site in the middle of the AP (black
ditches).
taken at various times. Emplacements can be spotted by the presence of tracks, narrow
gauge railways and telephone cables leading towards it. Also important are blast marks,
black spots on the ground, created by the muzzle fire of the artillery pieces. Another typical
feature is the clustering of artillery pieces in a linear shape. Other difficulties are dummy
emplacements intended to draw the attention away from the real batteries. Field batteries
in the open air can be seen on the photographs by the barrels of the individual pieces, often
167
First World War aerial photography
Figure 4. Camouflaged German artillery position defended by two belts of barbed wire (marked by black oval), 1500m
south of Bikschote (27 December 1916). Archive SGRS-S/A. Aerial photos map 258.
occurring in combination with horses and carts. Artillery pieces in gun breastworks appear
as horseshoe-shaped earth heaps and are very difficult to camouflage. As a result they were
easily spotted and had a short lifetime. Emplacements without cover were most likely to be
unoccupied, unlike covered emplacements. A third type consists of a battery protected by
casemates, protective bunkers for guns made of wooden logs and earth. They provide better
cover and the guns are partially situated underground. The design of these casemates was
trapezoidal, and the opening at the front side could be left open. The casemates are not
always separate structures, there are many gun emplacements with only one elongated earth
heap to provide cover for multiple artillery pieces (Figure 4). When these structures are
located next to rows of trees or fields, their visibility on the photographs drops enormously.
They can only be seen by stereoscopic analysis or by long shadows.
168
Method
B. Stichelbaut
Another category with a low visibility on air photographs is the anti-aircraft artillery.
During the First World War these were conventional field guns on a round earth or concrete
platform to ensure that the barrel could be raised high enough and turned into every
direction. Artillery pieces can also be situated in concrete structures. The study of blast
marks and trench maps can give more information about this. It is also useful that most
batteries consisted of four guns, simplifying the interpretation of four bunkers in a row as an
artillery position. Along the Belgian coastline, between Nieuwpoort and the Dutch border,
the German forces constructed 36 high-range coastal defence batteries with a large calibre.
In general, the batteries consisted of four guns, with narrow gauge railways for their supply
and concrete shelters for ammunition and personnel. Heavy batteries can also be spotted
in the land behind the enemy trenches. In Belgium there are two known examples: the
Leugenboom and Predikboom batteries (380mm) (Carlier 1921: 182).
Behind the battlefield there were many structures and infantry encampments, with the
same characteristics as in the front lines. Trench warfare needed an unceasing supply of
ammunition and building materials. These were centralised in large depots before being
distributed to the fighting units. In ammunition dumps at the rear of the front lines,
boxes were stored in the open air in regular rows. For safety and to avoid detection, the
closer to the front, the smaller the ammunition dumps become, and ammunition nearest
to the front line was stored in concrete structures or dug into fake shell holes. On the air
photographs, ammunition dumps are revealed by the paths leading towards them, rows of
small pits, and their location in relation to artillery positions. Other detectable structures
were barracks and hospitals. Each consisted of an accumulation of rectangular wooden
buildings in rows, and hospitals had a painted red cross on their roof. Also interesting is the
presence of German burial areas close to the trenches. After the war the bodies were moved
to centralised Soldatenfriedhoven as in Langemark. On the photographs these burial areas
are clearly noticeable because of their geometrical layout. Sometimes the German graves
were simply dug in lines; however, more often decorative figures (i.e. crosses) were formed
(Figure 5).
First World War aerodromes have little in common with modern examples. There was
no concrete landing strip, the aircraft taking off and landing from a levelled field. On the
photographs the airfields are visible by the light tracks caused by the landing wheels. There
is always a white ‘landing-T’ that indicates the direction of the wind. Some aerodromes
had wooden sheds, others consisted of canvas tents that could shelter up to three airplanes.
These had a typical hemispherical shape and are light coloured. Both sides used telephone
cables, wireless telegraphy and optical signal posts. Buried cables can be recognised by a
light 2-m-wide strip, caused by freshly thrown up earth. Telephone poles can be spotted on
some photographs. Optical signal posts have a more complex construction, made up of a
number of ditches that conceal the beam from direct observation.
Archaeological and historical traces
Pre-twentieth-century traces are visible on the wartime photographs, mostly showing as
wetmarks and dating from the medieval period. With examination of the images, new
archaeological traces of sites subsequently destroyed by the war or town enlargement can
169
First World War aerial photography
Figure 5. German cemetery, 1km west of Bikschote (31 January 1917). German battery at the right side (marked by black
oval line). Archive SGRS-S/A. Aerial photos map 258.
be recovered. They contain a treasure of information for researching moated sites especially.
The flooding of rivers and the destroyed drainage systems in the battlefields created ideal
recording circumstances for these sites. Moated sites can be seen in Figures 2 and 3 (in
Figure 2 in the centre of the picture next to trenches and crossed by the road; and in Figure 3
as a black square in the centre of the aerial photograph). Medieval moats are also clearly
visible (i.e. at Stuyvekenskerke, Merkem and Diksmuide). Other sites also have a military
nature, for instance medieval and post-medieval city walls and forts. In this case, stereoscopy
can also add new information. The sixteenth-century Fort Nieuwendamme at Nieuwpoort
is a marked example. On modern aerial photographs only parts of the fortress are visible.
On an anaglyph of two First World War pictures, each part of the traces of the earthen fort
can be studied in three dimensions.
Historical potential
During the war, the aerial photographs taken by the intrepid pilots were used to make trench
maps (Chasseaud 1999), on which the trenches were only reproduced schematically, and
marked with conventional symbols. The maps contain unavoidable imprecisions of location
170
Method
B. Stichelbaut
Figure 6. Partial view of the full inventory of traces, 1km north of Diksmuide at the Belgian-German front line.
and identification, and they represent the battlefield at a particular date. For instance on a
map dated 1918, trenches dug and destroyed in 1915 would not be recorded. In contrast,
on photographs taken in 1918, traces of earlier workings can still be seen.
If we compare the data visible on the trench maps to the GIS inventory of the sites that
has been compiled from the aerial photographs, the difference is self-evident. The historian
is privileged to have a detailed overview of 4 years of trench warfare denied to the soldier
at the time. Modern cartographic technology can do far more now than was imaginable
90 years ago. By adding a time dimension in the visualisation of the maps (with the ‘dating’
field in the ArcView ‘Theme Table’) the changing front line can be accurately monitored.
This reveals, for example, the reuse of certain features in different phases of the war (for
example, the incorporation of German trenches in the British trench system after the front
line moved) and the typological evolution of the trenches. The information captured is
of high research potential, in that it should allow historians to better understand this
momentous period in Europe’s history.
171
First World War aerial photography
Conclusion
This study shows that by applying modern cartographic techniques to early-twentieth-
century aerial photographs, it is possible to achieve a detailed overview of the material
remains of the First World War. The scale, density, distribution and accurate location of
these structures, and the way they were modified during the course of the war, provide a
resource not available to either historians or generals at the time.
The digital maps are more comprehensive and reliable than the existing trench maps, and
provide an effective and low-cost instrument for the scientific and cultural management of
this valuable and fragile heritage.
Acknowledgements
This master’s thesis drew on the large number of aerial photographs, primarily taken by the Belgian Army,
that can be found at the Belgian Royal Army Museum (Koninklijk Legermuseum Brussel — Mus´
ee Royale de
l’Arm´
ee) and the Belgian military archives in Brussels (Service G´
en´
eral de Renseignements et S´
ecurit´
e: Section
Archives) and I would like to thank them for letting me use the pictures. M. M´
eganck, W. Gheyle and D.
Herremans are thanked for discussions on this topic and their useful help with GIS and ILWIS. A special word
of thanks goes to the promoter of this research, Professor Dr J. Bourgeois (Department of Archaeology and
Ancient History of Europe, Ghent University) for encouraging and supporting this research and for his valuable
comments on this paper.
References
Anon. 1921. Manual of field works (all Arms). London:
HMSO.
Anon. 1925. Notice sur l’´
etude des photographies
a´
eriennes. Brussels: Defence National.
Anon. 1998. Trench fortifications 1914-1918, A
reference manual. The Battery Press Reference
Series 18. London: Imperial War Museum.
Carlier, A. 1921. La Photographie A´
erienne pendant La
Guerre. Paris: Librairie Delagrave.
Chasseaud, P. 1999. Artillery’s astrologers. A history of
British survey and mapping on the western front
1914-1918. Lewes: Mapbooks.
Delve, K. 1997. World War One in the air, a pictorial
history. Marlborough: Crowood.
Desmet, J. 1921. Rˆ
ole des aviations belge et franc¸aise
sur le front occidental pendant la grande guerre.
Bulletin des sciences militaires 1921-1922: 31-42,
163-76, 343-52, 439-49.
Ellis, J. 1989. Eye-deep in hell, trench warfare in World
War I. Baltimore: The Johns Hopkins University
Press.
Jaumotte, J. 1919. Conf´
erence du Commandant J.
Jaumotte, du Genie. La Conquˆ
etedel’air,Bulletin
Officiel de l’A´
eroclub de Belgique 14: 163-6.
Lampaert, R. 1997. Van pionier tot luchtridder, De
Geschiedenis van het Belgische Militair Vliegwezen
voor en tijdens de Eerste Wereldoorlog.Erpe-Mere:
De Krijger.
Stichelbaut, B. 2005. Great War aerial photography, a
contribution to the Flemish battlefield archaeology,
in J. Bourgeois & M. Meganck (ed.) Aerial
photography & archaeology 2003. A century of
information: 135-48. Archaeological Reports Ghent
University 4. Ghent: Academia.
172
