A Historical GIS Analysis of the Landscape Compositions: A Case Study of Folding Screens "Rakuchu-Rakugai-zu"

Introduction
Recently, the use of Geographic Information Systems
(GIS) to analyze historical space has attracted interests;
this approach is known as Historical GIS (HGIS)
a subdiscipline in Digital Humanities (William and
Thomas, 2004). In this paper, I propose a new methodology
to analyze compositions of landscape in paintings
with GIS, which I apply for analyzing Japanese screen
paintings known as rakuchu-rakugai-zu (洛中洛外図),
created between the 16th and 18th centuries. Rakuchurakugai-zu
provides contemporary bird’s-eye views of
the city and environs of Kyoto, Japan’s capital at the
time. The paintings capture man-made structures such
as residences and palaces of prominent samurai and
court nobles, temples and shrines; natural features such
as hills and rivers; and festivals and other human activities.
Nearly all of these works consist of a pair of folding
screens of six panels each. At present, over 100 such sets
are known to have survived.
Because the scenes of Kyoto painted on these screens
provide historical information not found in written
texts, they have attracted interests of scholars in various
fields. Particularly noteworthy among these studies are
the following three results of detailed analyses of space
on rakuchu-rakugai-zu screens. Takeda (1966) classifies
these screens into three types, according to their
era of creation and drawing method. (i) The ‘standard’
type, commonly known as the ‘first-generation,’ depicts
scenes from the late 16th century, mainly views of urban
districts of both Shimogyo, i.e., the southern half of
the town, and Kamigyo, i.e., the northern half of the
town, with the surrounding hills as a backdrop. (ii) The
‘variant’ type offers close-up depictions of specific subjects
or districts. (iii) The ‘developed’ type, commonly
known as the ‘second-generation,’ depicts scenes from
the 17th century, particularly of Nijo Castle and downtown
Kyoto, with hills in the background. Takeda argues
that the changes in drawing method reflect a shift from
seasonal nature paintings employing elements of the ‘famous
views’ painting tradition to genre paintings reflecting
trends of the times. The knowledge obtained from the previous research
concerning changes in the subject matter and geographic
range of rakuchu-rakugai-zu is essential to our understanding
of the screen compositions. However, such previous
research has not shed light on how actual views
of Kyoto were transferred to the restricted dimensions
of the screen surface, nor on how real geographic data
were manipulated in the process of making that transfer.
This is because the past analyses have relied solely on
the simple mapping of landmarks. To analyze the compositional
methods employed to paint these scenes, not
enough is simply mapping landmarks; we must also determine
which landmarks and districts were distorted,
and how. In this paper, I propose a new approach, HGIS,
which offers an analytic methodology utilizing GIS to
measure distortions in the landscape depicted on the
screen. This methodology visualizes distortions in the
drawn space by linking the positions of landmarks as
they appear on the screen painting and on survey maps,
and transforming the configuration of the screen accordingly.
Combining the results obtained by this method
with those of previous research should provide us with
more detailed and precise understanding of the drawn
space on these screens.
Methodology
The methodology I introduce here uses GIS spatial analysis
functions to scan the screen surface onto a survey
coordinate grid based on the relative positions of landmarks
in the screen painting. The analytical procedures
go as follows (Fig. 1):
1. Derive coordinate values for landmarks, both on the
screen and on a survey coordinate grid;
2. Generate a link table from these two point-data sets;
3. Use projective transformation and rubber sheeting
techniques to project the screen surface onto the survey
coordinate grid; and
4. Project the areas of the rubber sheet-derived polygons
onto the screen.
This process gives visual representation to differences
between actual space and the space drawn on the screen.
Results show that screens painted in the 17th century and
later distorted actual space more than screens painted in
the 16th century, indicating a decrease in adherence to
perspective-like conventions. This trend toward greater
distortion suggests a shift in landscape drawing methods
away from realism. Increasing deformation may be
attributed to changes in political regime as well as an
expansion of the public’s geographic awareness. I see
significant advantages of using GIS in understanding
Drawn Area and Regularity in Drawing, which I will explain
in detail here.
Drawn Area
Using GIS visualizes a drawn area exactly, which means
we could know which landmarks are included or not
in the painting. Depending upon the area included, we
could know how much of forced inclusion, meaning
spatial distortion, occurs. In the first-generation screens,
for example, the projective transformation-derived polygons
and rubber sheeting-derived polygons are similar in
shape. The second-generation screens, however, reveal a
pronounced tendency to forcibly incorporate landmarks
to the north and south of the city, even though this results
in various expedient changes in the drawn area of the left
screen and a progressive narrowing of the background
scenery in the right screen. Finally, the variant-type
screens may be explainable as the ones with their focus
on specific districts, for which the forced inclusion of
landmarks as employed in the second-generation screens
proved insufficient. This proves distortion progresses
over time.
Regularity in Drawing
Using GIS makes it possible for me to analyze regularity
in drawing, i.e., the locations of spatial abbreviations and
exaggerations on the screens. First-generation screens
exhibit orderly increases in area value, suggesting that
specific conventions similar to the rules of perspective
were followed to achieve the precise geographic positioning
of landmarks. The result is a realistic rendering
of geographic space. In contrast, second-generation and
variant-type screens exhibit numerous instances of distorted
space, and area value increases variously from
locale to locale. This suggests that valued less was the
actual positional relationships of landmarks in the drawing
of these scenes. This means that GIS clearly shows
that in terms of their rendering of real-space positional
relationships the first-generation screens are superior to
these later ones.
Conclusions
Until now, there has been little research applying GIS
to analyze landscape drawing methods. This case study
shows, however, such an analysis is feasible when applied
to paintings covering a wide area of the city scape,
such as rakuchu-rakugai-zu screens. The analytical
methodology presented in this study involves not merely
mapping landmarks that appear on the screens, but also
projecting a “virtual screen” onto a survey map. I was
able to implement this process for the first time by using
the ‘adjust,’ ‘area calculation,’ and ‘table join’ functions of GIS analysis. The results obtained with this methodology
offer insights from a geographic point of view into
the approaches taken to space drawing and landmark positioning
in these works.
Analyzing landscape drawing methods, conventional
art history mentions that from the 17th century on, the
depiction of politically significant landmarks became a
norm and took priority over the accurate rendering of
geographic locations. That is, landmarks took on the aspect
of symbols, which the creators of the screens increasingly
incorporated, paying less attention to their
positional relationships.
In short, it turned out that my case study of rakuchurakugai-zu
screens of the 16th and 17th centuries supports
this point of conventional art history not from a
subjective but an objective point of view as using GIS
can visualize drawn area and regularity in drawing exactly,
measured in quantity (Fig. 2). While this overview
has provided a sense of general trends over time, I hope
to follow up this study with more detailed analyses of
individual screens in the context of the sociopolitical environments
in which they were created.
Acknowledgements
This research was conducted with assistance from the
Japan Society for the Promotion of Science and the Ministry
of Education, Culture, Sports, Science and Technology,
Global COE Program, “Digital Humanities Center
for Japanese Arts and Cultures”, Ritsumeikan University.
References
Takeda, T. 武田恒夫 (1966). Rakuchu-rakugaizu
to sono tenkai 洛中洛外図とその展開. In Kyoto
kokuritsu hakubutsukan ed: Rakuchu-rakugai-zu 洛中
洛外図, kadokawa shoten 角川書店, 1966, pp.17-32.
William, G. and Thomas, III (2004). Computing and the
Historical Imagination. In Schreibman, A., Siemens, R.,
Siemens, R. G. and Unsworth, J. (eds), A Companion
to Digital Humanities, Oxford: Blackwell Publishing,
pp.56-68.
Figure 1. Procedures Figure 2. From Realism to Deformation