4 . 1 I N T R O D U C T I O N

Starting a chapter by this title implies a lot of explanation to the reader. Terms like ‘geomatics’

and ‘physiognomic landscape research’ promise a wide interest in a diversity of scientifi c do-

mains, especially when the geomatics component is the main focus.

In the pioneer stage of conducting physiognomic landscape studies by use of automated proce-

dures most scientists discerned the limitation of computer capacities and the availability and

accuracy of data. However, they affi rm positively the role of computations; geo-information

was already mentioned. This chapter surveys the expected role by the key-word geomatics. The

geomatics defi nition evolved for the last decennium into “Geomatics is a fi eld of activity which,

using a systematic approach, integrates all the means used to acquire and manage spatial data

required as part of scientifi c, administrative, legal and technical operations involved in the pro-

cess of production and management of spatial information. These activities include, but are not

limited to, cartography, control surveying, digital mapping, geodesy, geographic information

systems, hydrography, land information management, land surveying, mining surveying, pho-

togrammetry and remote sensing [url 1]” (Roswell and Tom, 2009).

However, the scope of this chapter is narrowed. For that reason the defi nition of physiognomic

research as given in the introduction chapter is a starting point and will be used as a reference.

R O N V A N L A M M E R E N

GEOMATICS IN
PHYSIOGNOMIC
LANDSCAPE RESEARCH
A DUTCH VIEW

74 Geomatics in physiognomic landscape research. A Dutch view

The geomatics defi nition from the Dutch perspective is given. Followed by a stepwise descrip-

tion of the geomatic items: geodata, geodata processing and geodata visualisation. Afterwards,

geomatics and physiognomic landscape research will be linked again twice, fi rstly, from the

geomatics perspective, and secondly, from the physiognomic landscape perspective. The fi nal

section concludes the relation between both domains and sets an outlook. However, the chap-

ter is greatly based on the experiences in the Netherlands, but references will be made to an

international setting.

4 . 2 G E O M A T I C S – A N E X P A N D I N G D O M A I N

The current ISO defi nition of geomatics shows that the original focus on spatial information/

geodata “a technology and service sector focusing on the acquisition, storage, analysis and

management of geographically referenced information for improved decision-making” (Ca-

nadian Council of Land Surveyors, 2000) has been changed in favour of management and

decision-making by integration and systematic approach items. Besides, ISO uses geomatics

and geographic information science as synonyms.

A Dutch textbook on Geographic Information Systems and Spatial research (Hendriks and

Ottens, 1997) presents Geomatics as the domain that integrates modelling on the levels of con-

ceptual and logical representation of the spatial reality by developing geo-information method-

ologies and theory. The analysis of conceptual and logical models, as well as their relations of

interest of different application domains, fuels these developments (Molenaar, 1997).

As we may understand from the more recent defi nitions, and notice from the reported inven-

tions and developments (Fisher, 2006), the domain has dramatically changed and is still

changing since the early days of Canadian GIS, CGIS [url 2], as did, and does, this sector in the

Netherlands (Bregt and Van Lammeren, 2000). All these changes have an impact on the Dutch

physiognomic landscape studies.

4.2.1 Data sources

In the relation with physiognomic landscape research the acquisition of geodata is still of pri-

mary importance, because the nature of the input datasets the scope of the analysis outcome.

For that reason, fi rst an overview of present data of interest is shown. Second to that, the ways

to visualise the input and derived data needs a thorough look, after all we are dealing with the

visible landscape. Finally, the (automated) functions to analyze these geodata needs will be

exploited.

A geomatics-time line (fi gure 1), which shows awareness of reported evolutions and is partly

infl uenced by the GIS timeline [url 3], gives an overview of the change in available geodata.

Looking at the acquisition of geodata it is obvious that from the study of De Veer and Bur-

rough (1978) and in line with the CGIS setup, the fi rst geodatasets were manually made by

a procedure that was very similar to the raster approach as used for physiognomic landscape

research. This rasterising way of getting geodata was based on the defi nition of a variable (for

example ‘space-mass difference’), the spatial extent and a spatial resolution (the raster-cell

size). Via a physical overlay of a pellucid paper, with a drawn raster on it, over a hard copy

map (“don’t forget the fi ducial marks!”), per raster cell a value related to the variable was writ-

ten down.

4.2.2 Data availability

The introduction of vector-based geodata started as early as the raster approach originated by

the work of Sutherland (2003). However, in contrary to CGIS it took a while before it strongly

pushed, in a more practical sense, data developments in Computer Aided Design, and as origi-

nally developed. However, in the Netherlands, the role of vector geodata was serious from 1987

onwards, when the fi rst versions of GeoPakket in relation to SiCAD (in 1987) and ArcInfo (in

1983) were introduced in the offi ces of the Dutch Administrations. Since that moment the map

series of the Dutch Topographical Service of the Land Registry (Topografi sche Dienst Kadaster)

began to become available as geodata (Top10Vec). Also, the early satellite images gave a boost

to geodata in the early 80’s, by Landsat-TM images [url 4].

1975

geo data

MAPMAPMAP

Intergraph/ESRIIntergraph/ESRIIntergraph/ESRI

Top10VecTop10Vec

FGCDFGCD
3D topology3D topology

AHNAHN Computer visionComputer visionComputer visionComputer vision
Web 2.0

LBSTerrestrial LidarTerrestrial Lidar

Crowd sourcingCrowd sourcing

raster
vector

pictures
video

1985

1995

2000
2005
2007

2015

CGIS

Top10VecTop10Vec

Computer visionComputer visionComputer visionComputer vision

Figure 1

Geomatics events of importance for geodata availability

76 Geomatics in physiognomic landscape research. A Dutch view

Since the well-known Geodata Act of the Clinton government (FGDC, 1994 [url 5]) the interest

in authorised geodata became globally a serious item. In the Dutch Ministries in cooperation

with their departments, like Dutch Topographical Service of the Land Registry and Census of-

fi ces, started rapidly to develop standardised geodatasets of national concern, that gradually

replaced the hard copy maps. The last hard copy topographic map series of the Netherlands,

scale 1:50,000, dated from 1984 [url 6]. Besides the description of position based data was no

longer a primarily a real map case, but instead geodata many organisational aspects change

gradually. For example, the latest, 2008, Dutch Act on landscape planning demands the use

of geodatasets in the procedure of creating and deciding upon municipality zoning plans (by

January 2010, [url 7]).

Driven by the EU Inspire initiative (2009, Annexes) more of these enactments will follow. The

guiding role of the Inspire Annexes is interesting in the case of physiognomic studies. What

type of geodata has been described, and, may support such studies?

As soon as the airborne Lidar technology was available, the Dutch Ministry of Public Works

(Ministerie van Verkeer en Waterstaat) started a campaign with this technology, to map,

during the period 1997-2003, the elevation of the Dutch land area. This ‘actual state of

the Dutch land elevation’ (Actueel Hoogtebestand Nederlandthe Dutch land elevation’ (Actueel Hoogtebestand Nederlandthe Dutch land elevation’ ( (AHN-1)), was the fi rst high-

resolution digital elevation dataset [url 8]. The national authorities of Europe are encour-

aged by the European Inspire directive [url 9], which defi ned 34 spatial themes to develop

geodatasets for. A closer look at the appendixes that describe these themes shows that the

majority of the themes are about orthogonal two-dimensional geo-referenced (2D) geodata.

However, many new developments with respect to CAD and 3D visualisation have been initi-

ated to capture and deliver three-dimensional geo-referenced (3D) geodata. Experiments in

these directions are on-going and are most promising (Xu et al, 2010; Döner et al 2010). Yet,

Figure 2

Point cloud that represents a forest stand (by Van Leeuwen, 2010)

the debate on feasible data structures and fl awless topology rules is still on. Especially in re-

lation to the (combination of) airborne and terrestrial Lidar (Tang et al, 2008; van Leeuwen,

2010), 3D referenced data will offer more geometric details of real world phenomena, which

may suit physiognomic landscape studies. Currently, the translation from point clouds, the

measured points of refl ection, into 3D objects for landscape visualisation remains challeng-

ing (fi gure 2).

4.2.3 Occasional geographic data

The Internet, on the other hand, has gradually become the main source of data. Wherrett

(2000) presented the Internet as a medium to send out questionnaires in relation to landscape

perception studies. Particularly the concept of participation and interaction by the Internet, as

promoted by the concept Web 2.0 [10], has brought forward many Internet communities who

store and share data. Mobile phones and digital cameras make it possible to geotag all data

types ranging from mobile messages to photographs and videos, and once done, these data ob-

jects can be easily tracked by location based services (lbs)(Raper, 2007).

Successively applications like Flickr [url 11], Panoramio [url 12], Locr [url 13], Google Earth

[url 14] and Bing [url 15] offer the many dedicated volunteers to geotag their photographs and

put these on the Internet. The huge amounts of photographs do offer a great ‘crowd source’

of data that may be of use for physiognomic landscape studies (Jones et al., 2008). Of interest

with these photographs is the variety of perspective projections. Sometimes orthogonally pro-

jected photographs are available.

One data type that is not mentioned in the time line is the set of 3D rasters, based on volume

image elements, and mostly called voxels. Though it started as a promising development in a

GIS setting (Marschallinger, 1996), the interest seems to have faded out. However, in some ge-

ological and geomorphologic studies the data type as such is still in use (Clevis et al., 2006), as

in some remote sensing studies. Computer gaming and medical studies, like CT-scan analysis,

still favour these data types [url16].

4.2.4 Trend watch

With respect to the previous sketch of the geomatics development it is obvious that the variety

of available and accessible digital geodata has increased dramatically. This variety does offer

many options to be used in physiognomic landscape studies (fi gure 3). Yet, drawbacks still exist.

Geodata can be 2D or 3D referenced and the reference systems can vary. The VGI datasets may

be biased. An Internet search for photographs of the Eiffel tower (keywords in English, German Eiffel tower (keywords in English, German Eiffel tower

and Dutch) on July 26th 2010 showed 1,550,000, 256,000 and 24,400 photographs respective-

78 Geomatics in physiognomic landscape research. A Dutch view

ly. For the Hoge Veluwe, a Dutch National Park, the total of available photographs was 65,300.

In any case not all datasets are accessible due to commercial or privacy reasons.

The role of all previously mentioned data types is to describe current, recent past and present

states of the visible landscape and as such it could be useful.

It seems that rarely geodata is produced for the function of physiognomic landscape research.

Consequently available data is not always optimal, and needs to be transformed or interpreted

before it can be used in this kind of research.

4 . 3 T W O O R T H R E E D I M E N S I O N A L G E O D A T A

The subject of physiognomic studies has been defi ned before as ‘the visible landscape’. Visibility in

this case meant from the human’s eye perspective. As Mark (1999) explained, geodata is merely a

representation of the things that exist and represent, in our case, the ‘visible landscape’. Compara-

ble with his approach in this text (Mark, 1999), the words ‘phenomena’ and ‘entities’ point at the

things that exist. The words ‘objects’, ‘features’ and associated words like ‘attributes’ and ‘values’

refer to the representations of phenomena and entities in the formal system of the digital world.

Y-axis: 3D data defi nition (top) to 2D data defi nition (bottom)

X-axis: 3D geo-reference (left) to 2D geo-reference (right)

(x, y, z)da

(x, y, z)

(x, y, z)dbz

Figure 3

Available data types for physiognomic landscape studies

4.3.1 Digital landscape model attributes

The representation of the visible landscape phenomena and entities includes minimally a

topographic surface. The surface represents the continuous phenomena elevation. This surface

however may include more delineated features like pits, tops, ridges, edges, faults and stream

patterns. This representation is known as a digital elevation model (DEM). However such a DEM

is not a representation of the visible landscape. Man-made entities like buildings, roads, canals,

plantings, as well as natural vegetation, have to be clearly represented as well. Gathering and

adding these defi nable delineated objects to the DEM will generate fi nally a landscape object

model (LOM) or digital landscape model (DLM). Such models may be understood as the rep-

resentation of the visible landscape (fi gure 4). Wassink (1999) labelled the visible landscape

by the nowadays old fashioned annotation ‘landscape as a whole’. This label originated from a

more static and fi xed physiognomic landscape approach.

The different types of geodata, as presented in the second paragraph, could support the repre-

sentation of the visible landscape in many ways. The 2D or 3D referenced geodata (raster and

vector) could offer, for example after an interpolation process, a DEM coverage of a certain

geographical extent [url17]. For example in a 2D geo-referenced setting the terrain layer (of

fi gure 4) can be represented via surface, vector or raster defi nition. Besides the terrain layer,

the volume layer (of fi gure 4) can also be generated in this way. In a 3D geo-referenced setting

the terrain layer as well as the volume layer can be constructed by a 3D geometry and topology

of morphologic objects. By adding 2D and 3D referenced objects, like representations of build-

Figure 4

A visible landscape representation

(source: Wassink, 1999)

On bottom the DEM (terrain layer).

Successive layers like network (line and area

objects) and volume layer (3D objects) put

forward a DLM (on top)

Landscape as a whole

The volume layer

The network layer

The terrain layer

80 Geomatics in physiognomic landscape research. A Dutch view

ings, trees and other artifacts, a DLM will be created out of the DEM. For example a 2D geo-

referenced elevation dataset, like the Dutch AHN, can be supported by a selection of vertically

extruded 2D objects to generate a DLM out of a DEM as presented in fi gure 6.

If a ‘complete’ DEM or DLM coverage doesn’t exist then many other software functions are

available to create these models by a number of processing steps.

In many visible landscape studies in which a DLM is used, the spatial resolution or precision of

the model that suits the study is problematic. Many studies refer to this item as a scale issue.

However, this item deals, in fact, with the geometric precision and accuracy of the point, line,

area and volume features defi ned in a vector structure, and the granularity of the raster cells

and voxels. As well as the geodata representations of the terrain layer, the network layer and

volume layer are also problematic, as the relations between these layers is still under exposed.

Currently, it looks like the only solution to tackle this issue seems to be additional data sam-

pling and adding ancillary geodata.

Photographs have a number of attributes that are implicitly related to the image. These graphic

attributes like colour hue, colour saturation (grey value) and colour brightness are related to

the smallest feature of a photograph, the image element (pixel). The combination of such pix-

els offers patterns and structures. These patterns and structures are cognitively understood by

humans. In the domain of computer vision researchers try to mimic algorithmically these facili-

ties of the human brain (Szeliski, 2010). Videos also have the same basic implicit variables per

frame (Zhou, 2010), as a video consists out of series of related stills (scene) and series of scenes

(video narrative).

Geodata however consists mostly of geometrically well-defi ned features like points, lines, poly-

gons, volume objects (vectors) and raster cells or voxels. One or many thematic variables can be

linked to these features to explain the meaning of the features in terms of characters and num-

bers. The values, the range of numbers and characters, of the attribute domains are constrained

by a measurement scales (Gibson et al., 2000; Open GeoSpatial Consortium inc. [url 18]). In

landscape physiognomic studies these graphic and thematic variables fulfi l an important role.

4.3.2 Digital landscape model visualisation

The previously introduced DLM has many digital expressions. The visualisation of such DLM

expressions is the most important interface to discuss landscape physiognomy in most studies.

Geovisualisation has to be understood as defi ned by Dykes et al. (2005): “Geovisualisation

can be described as a loosely bounded domain that addresses the visual exploration, analysis,

synthesis and presentation of geospatial data by integrating approaches from cartography with

those from other information representation and analysis disciplines, including scientifi c visu-

alisation, image analysis, information visualisation, exploratory data analysis and GI Science.”

Figure 5 shows how real world phenomena, the visible landscape as perceived in reality, is

represented and infl uenced by four different transformations before we can perceive the visu-

alisation of it (Lammeren, Houtkamp, et al, 2010). The fi gure expresses the importance of

transformations. The fi rst transformation has been discussed in the previous paragraph. The

second transformation (TII) shows the preparation of the data for visualisation. Different com-

binations of 2D and 3D defi ned objects and layers may need to be generated to fi nally result

in a visualisation. All types of fi gure 5 can be combined, for example, Google Earth examples

may illustrate such visualisations by the historical landscape paintings of Florence [url 19] and

the 3D buildings layer of Amsterdam [url 20]. The latter (fi gure 6) shows different 3D house

geometries (3D) plus mapped textures (images used to show the ‘realistic’ facades of the build-

ings). The former shows geotagged images of historic paintings of Florence and located into the

original view direction. Photo’s and paintings like in the example of Florence, put forward the

subject of atmospheric conditions that’s not yet covered by geodata and only gradually by geo-

visualisation (Daniel and Meitner, 2001).

We may conclude that the issue of missing and imprecise geodata, in the case of physiognomic

landscape studies, can be partly dealt with by including other data types, especially geotagged

photographs and videos in the visualisation of such data. The automated processing of such

combined datasets is much harder however.

Real world

TI TII TIII TIV

2D 2D 2D 2D

3D 3D 3D

Geo data Geo-visualisation Computer display Perception

Figure 5

Transformations in the Geovisualisation process (source: Van Lammeren, Houtkamp, et al., 2010)

TI geodata acquisition; TII geovisualisation defi nition; TIII display rendering; TIV perception triggers in 2D and 3D (parallax and/or depth

cues). Light grey arrows refer to 3D referenced geodata types

82 Geomatics in physiognomic landscape research. A Dutch view

4 . 4 G E O D A T A P R O C E S S I N G

All geodata types and the options to visualise these have previously been discussed with the

meaning to show the variety of options to represent the visible landscape. The sheer variety

relates to the fact that that a dataset, which will be used to analyse physiognomic landscape

items, must be prepared by pre-processing before the more analytical processing can be per-

formed.

4.4.1 Pre-processing

Pre-processing includes the transformation of geo-references (map projection, 3D into 2D, 2D

into 3D), of geodata (like from vector into raster, raster into vector, geotagged images into

vector) and of feature classes (for example points into lines, points into areas, points into vol-

umes). It could also include (re-) classifi cation of attributes and attribute domains to better fi t

to the physiognomic landscape analysis (like ratio measurement class variable classifi ed as an

interval measurement class variable).

A very special class of transformation is related to the data of volunteers, also known as volun-

tary geographic information. These geotagged photographs and videos can be used, thanks to

the results in pattern recognition by computer vision research to stitch or construct photogram-

metrically multi-facetted scenes like those offered by Microsoft Photosynth [url 21] and even

3D-models and -scenes [url 22] (Snavely, 2006; Pollefeys, 2002). These very promising devel-

opments must lead to an integration of horizontal and vertical defi nitions of reference.

Figure 6

Amsterdam Royal Palace without (left) and with (right) 3D models

4.4.2 Re-classification and interpretation

After fi nishing the geodata pre-processing, the processing in line with the physiognomic land-

scape analysis may be started. Depending on the pre-processed data types available these pro-

cessing steps may be as different as the physiognomic landscape interests.

The simplest analysis seen from a computational point of view may be the re-classifi cation of

2D- or 3D-geodata layer, geotagged image or video. Re-classifi cation could support for example

user interpretation and appreciation by ranking or ordering, user understanding of classes of

interest for policy making and user labelling of features (e.g. Wascher, 2005). Such classifi ca-

tion may also follow after other types of processing. In more detail the re-classifi cation can

comprehend feature classes or thematic class values (Chrisman, 2003).

4.4.3 Simple geometric analysis

As mentioned before, most analysis in the domain of physiognomic landscape research is basi-

cally starting with a digital landscape model. Besides classifi cations, a number of new attrib-

utes may be described and calculated describing visual properties of the landscape model (e.g.

Ode et al., 2010). Geometric attributes of interest that may be used are: location, direction,

distance, altitude, size (length, area, volume), shape and topological relation. All these at-

tributes can be calculated for a single object, multiple objects and interrelated objects. Besides

attributes like spatial density, distribution and variability can be derived as a next step. Most

algorithms available by geodata processing software will support this type of geometric and

topologic analysis functions.

4.4.4 Visibility oriented analysis

An interesting dispute is always what type of object initiates the data processing. It always

means that a second geodataset is involved by which point, line, area and volume objects have

been described that will be used as the starting objects for the physiognomic landscape analy-

sis. For example the algorithmic principles of visibility studies, is typifi ed as visibility querying

on a digital landscape model (Batty, 2001; De Floriani and Magillo, 2003). Before the real

querying the continuous visibility mapping by TIN (vector) and discrete visibility mapping by

raster is started from point objects that represent vantage points.

In a raster format the range of local, focal, zonal and global functions, as originally defi ned by

Tomlin (1990), work in this way. For the voxel format there are comparable classes to be found

(Marschallinger, 1996; Clevis et al., 2006).

84 Geomatics in physiognomic landscape research. A Dutch view

4.4.5 De Veer and Burrough revisited

Current geodata types are, in other words, able to be processed in many ways. If we take into con-

sideration the examples of de Veer and Burrough (1978), then we may conclude that fi nding ‘con-

cave objects’ is a matter of pre-processing (especially classifi cation) 2D geodata, querying the area

objects that represent spaces and ordering afterwards on size and shape of the selected objects.

The ‘breadth of view’ approach is able to do this via the isovist fi eld’s concepts of Benedikt

(1979), which have been implemented. After pre-processing and defi ning the points that repre-

sent vantage points, a number of attributes may be calculated like lines of sight, fi elds of sight

and derivatives like the shape of the fi eld of sight. Also the viewshed techniques could be cat-

egorised as such an approach.

The third approach, ‘raster’ as they called it, is in fact a combination of classifi cations of attrib-

utes belonging to the objects and a transformation from vector based objects into a raster ge-

ometry. The resolution of the raster cell size will be a most critical factor in a correct ascribing

the intended values to each individual raster cell. Many studies still use this approach.

4.4.6 What validity?

In this paragraph the main classes of geodata processing have been introduced in short. Each

class delivers data output by which characteristics of the landscape physiognomy are descrip-

tively explored, quantitatively explained or qualitatively predicted. In the academic research

tradition the main question concerns the validity of these results.

From the position of this section in this chapter it could falsely be understood that the validity

of the results is just a matter of selecting appropriate processing tools, like cross-validation of

the used and resulted data. Or, on the other hand, the processing tools themselves could be

also subject to validation, as has been illustrated by the study of Riggs and Dean (2007) about

viewshed processing.

However, as put forward in the previous paragraphs, the variety of geodata that helps to repre-

sent the study of landscape physiognomy, are suspects themselves. Some geodata is generated

by given defi nitions and procedures within the context or praxis of a certain application do-

main. A formal defi nition of visual landscape entities and phenomena seems an irrefutable is-

sue. Fisher (1999) already explained that the nature of uncertainty is based on the well or poor

defi ned classes of objects and their spatial delineations. In well-defi ned situations uncertainty

is caused by errors and is probabilistic. In poorly defi ned situations uncertainty could be a mat-

ter of vagueness (weak defi nition) or ambiguity (confusing defi nition).

Indeed occasionally geodata has been generated without any given defi nition or procedure.

Most geotagged photographs and tweets bear this origin. But does it mean that it is vague or

ambiguous data? The enormous amounts of such geodata, accessible and available by social

networks, do unveil vagueness and ambiguity. Perhaps expert validation of physiognomic land-

scape fi ndings will be followed up by validation via social networks and E-communities. The

works of Sheppard (e.g. Sheppard and Cizek, 2009), promoting the ethics of visualisation, also

give clues for such more contextbased validation. The conclusion of Ode et al. (2010) that “The

results show that the different data sources were more or less adequate to use in different contexts

and for different purposes”, fuels this perspective.

4 . 5 G E O M A T I C S M E E T S P H Y S I O G N O M I C L A N D S C A P E
R E S E A R C H

From a geomatics perspective the link with physiognomic landscape studies may be based on

the extent of the geodata types and related variables to be used as input, to be processed into

a certain output variable, and to be visualised. The data types, as introduced in the second

section, may be a starting point. The way this data could be combined, processed and fi nally

visualised offer a combinatory set of options that may be of relevance for physiognomic studies

(fi gure 7). It will show clearly that the methodological soundness is a tough case.

4.5.1 Geodata ensembles

Figure 7 consists of three blocks. The fi rst block, entitled data, shows all data types as in-

troduced in section three of this chapter. The second block, pre-processing, summarises all

Data Pre-processing Processing

I II III IV V VI
raster cells a

voxels b

volumes 3D 3D 3D 3D 3D 3D c 1 1 1 1 1

surfaces 3 3 3 3 3 d 1 1 1 1 1

polygons

lines

point

pictures e

video to f
vid pic poi lin pol sur vol vox ras

from
1 = both, or one

Figure 7

Geodata ensembles

dark blue = original data; light blue = created data; box = based on photogrammetric functions

86 Geomatics in physiognomic landscape research. A Dutch view

transformations available to realise a specifi c type of digital landscape model as introduced in

section four. The third block, processing, ranges groups of data ensembles that may support

physiognomic landscape studies. The ensembles are based on original data types (block one) or

originated from the pre-processing results (block two) and take into consideration the process-

ing options as introduced in the fi fth section.

4.5.2 Ensemble-related pre-processing

The focus of the second block is on transformation options. The x-axis shows the from data and

the y-axis the to data. The fi rst column of the block presents the transformation of a video into

an image via a frame or still. Images can be used in many forms of transformation. As such the

pixels of an image may be converted into a grid. However in case of an image made via a per-

spective projection the geo-referencing may be a diffi cult topic to handle. Yet, the many pattern

recognition functions, resulting from computer vision research, could support the transforma-

tion from specifi c objects of an image into specifi c points, lines and polygons. Such derived data

may be used for a photogrammetric construction of a 3D-model, which in the fi gure is labelled

as volume data (indicated by 3D).

Transformations of points into lines, polygons and surfaces that fi nally represent elevation, by

the terrain and/or volume layer (fi gure 4), as well as lines into points, polygons and surfaces

and polygons into points, lines and surfaces, are very common functions in a 2D-reference

system. The transformations of these vector-based data types into raster datasets are also com-

mon. The transformation of points into volumes is possible in case of 3D referenced point sets.

Examples of such data are given by terrestrial Lidar data, that show transformation is also pos-

sible in cases of datasets with line and polygons objects defi ned by a 3D reference. Surfaces,

sometimes described as 2.5D-referenced because once visualised they give a three-dimensional

impression even though the geo-reference is still 2D, offer transformation options into both

directions. Creating points, lines and polygons in a 2D or 3D reference are possibilities. Besides,

surfaces could be transformed into a 3D (volumes), voxel and raster model. In fact, the volume

data type (3D) offers the same classes of transformations. However these transformation func-

tions have to include the conversion of 3D topology. Depending on the algorithm there is not a

single result. However that’s often the case in many of these transformations.

The seventh column presents the transformation of voxels. The creation of a volume model

needs an intermediate step in which 3D referenced points and lines of signifi cance have to be

found in favour of the construction of polygons and volumes. The last column presents the

transformation of raster data into polygons and surfaces.

4.5.3 Processing of single data ensembles

The third block of fi gure 7 gives an idea of the data ensembles that may be the object of pro-

cessing options related to physiognomic landscape studies. The fi rst column (fi gure 7: I) of the

third block points at the processing of a single data type. Examples of physiognomic landscape

applications that can be found by studies with raster cells (fi gure 7a) are very familiar with the

raster approach of de Veer and Burrough (1976). A raster cell is, in most of these studies, the

location identifi er for a number of variables of interest for a specifi c analysis. These variables

may originate from thematic themes and geometric items. For example McGarigal et al. (2002)

introduced landscape metrics; a number of variables originally thought useful for landscape

ecology studies. However, the concept of the matrix-patch-corridor model can be understood

metaphorically and used in physiognomic landscape studies (Kamps and Van Lammeren, 2001;

Palmer and Hoffman, 2004).

Voxel analysis (fi gure 7b) is not often found in physiognomic landscape studies. However,

if geomorphology Clevis et al., 2006) or layered Isovist fi elds are included, the so-called

Minkowski model (Benedikt, 1979), then voxel analysis has a lot to offer. The role of volume

models (fi gure 7c) is at the moment mainly related to studies of perception and assessment.

The lack of well-defi ned data structure and 3D topology blocked the availability of suitable

analysis methods, like 3D Boolean operators, that offer immediately quantitative numbers of

the calculated results. Surface analysis (fi gure 7d) is available in many ways, especially for the

discovery of height derivatives like contours, slope types, slope aspect, edges and drainage pat-

terns. At the moment the tools to process images (fi gure 7e) and videos (fi gure 7f) are mostly

related to perception and appraisal by assessing single, pairs and series of images. Such studies

are still in line with the studies like Schroeder (1988). Besides usability, analysis, like naviga-

tion and orientation, in relation to the visualisation of the above data types are of interest for

physiognomy studies.

4.5.4 Processing steps of multiple data ensembles

The other columns of the third block (fi gure 7: II up to VI) show the ensembles that make use

of integrated datasets. In all cases the DLM may be based on a surface (2.5D) or volume (3D)

data types or a combination of both. The most common combination by now is the one where

volume objects are placed on a surface. Depending on the physiognomic landscape analysis, in

this example visibility, the data model could be extended by:

• Points, in the case of view point based visibility studies;
• Lines in the case of route based visibility studies;
• Polygons in the case of neighbourhood or specifi c landscape unit based visibility studies;
• Rasters in the case of all previously mentioned types of studies.

88 Geomatics in physiognomic landscape research. A Dutch view

In fact all ‘visualscapes’ analysis tools as introduced by Llobera (2003), including isovist and

viewshed, are look-a-likes. All of them derive values related to visibility variables from the

above mentioned data ensembles. A serious 3D approach is suggested by the ViewSphere ap-

proach as discussed by Yang et al. (2007).

Recently, new combinations of data show the surplus of options of how geomatics meets physi-

ognomic landscape studies. What seems promising is the integration of surface (2.5D) or vol-

ume (3D) data with:

• Images in case of (semi-) photo realistic renderings via texture mapping or having loca-
tion based billboards or panoramic views in the digital landscape model for example like

Streetview [url 23] to capture specifi c visibility items;

• Videos in case of dynamic renderings of objects or location based video streams also to cap-
ture visibility items.

4 . 6 P H Y S I O G N O M I C L A N D S C A P E R E S E A R C H M E E T S
G E O M A T I C S

Another approach to show how both domains meet may be given by the four landscape per-

spectives of Antrop (2007). A landscape perspective is defi ned as the way that human are con-

fronted with the landscape. The four perspectives are the vertical, the horizontal, the mental

and that of the meta-reality. The fi rst two perspectives are mainly related with the primary

cognition and defi nition as a human may sense. The latter two are more related to derivation

and inference from the fi rst two perspectives. For that reason the fi rst two are primarily linked

to data as input for processing and the latter two as output data of processing (fi gure 8). All

derived variables are group-listed in the column furthest right in fi gure 8.

meta-reality perspective

mental perspective

horizontal perspective

vertical perspective

mindmap

shape indicators

landscape metrics

philosophical-psychological

parametrical-reductionistic

genius loci

landscape characteristics

Data – Input

Data – Output

Figure 8

Physiognomic landscape perspectives (based on Antrop, 2007)

The meta-reality and mental perspectives are recognisable in the studies of Zube et al. (1982)

and Daniel and Vining (1983). 30 years later such an extensive literature review, as they did,

could be of serious interest to fi nd out how human-landscape interaction studies have been de-

veloped since and how these paradigms have evolved. Dutch studies of the past decennia have

checked, and it may be considered, that the paradigms (table 1) are detectable and, in each of

these applied geomatics, are traceable.

4.6.1 Expert approach

The expert-ecological approach may be recognised in the dissertation of Wassink (1999). The

dissertation presents a methodological attempt on a qualitative landscape classifi cation to de-

fi ne landscape form transformations of the Dutch landscape. A landscape morphological model

has been developed, based on a layered concept (fi gure 5) of terrain forms (digital elevation),

network pattern and vertical landscape features like buildings, trees and shrubs. The methodo-

logical framework was applied for the brook valley landscapes of the Pleistocene areas of the

Netherlands. The study shows that this multilayered model may visually support the insight in

specifi c relations between geomorphology, networks and vertical features in relation to land-

scape forms. The geodata in case of networks and vertical features were derived from the Dutch

topographical data. Kamps (2001) repeated the analysis of Wassink’s study by using raster data

and landscape metrics (McGarigal et al., 2002). Steenbergen et al. (2009) made the same type

of study for the Dutch polder landscapes.

The works of Wassink (1999) and Steenbergen et al. (2009) bridges the expert-ecological ap-

proach with the formal-esthetical approach. The latter is much more detectable in Steenbergen

et al. (2003, 2008), with the use of architectural variables like vista, rhythm, symmetry and

order in a variety of landscapes (like Italian Renaissance villa landscape and Dutch Polder land-

scape). Kerkstra et al. (2007) analysed vista’s to fi nd a leading architectural design principal for

designs of the undulating Zuid Limburg area in the south of the Netherlands.

Table 1

Tentative summary of differences in human-landscape interaction paradigms.

context liability sensitivity validity usability

expert - ecological research preference medium medium medium-low low-medium

expert - formal esthetic artist view low medium-high low high

psychophysical mutually well
defi ned

high low medium-good medium

cognitive - psycological differ : cognitive,
perceptive, affective

high-medium high-medium medium-low medium-high

experiential-
phenomological

inside/outsideness low-medium high pm - personal local

90 Geomatics in physiognomic landscape research. A Dutch view

4.6.2 Psychological and psychophysical approaches

In the psycho-physical approach the Dutch research groups show still many interests. These

studies still start with defi ning space and physiognomic character by the type and amount of

landscape features (Werkgroep Helmond, 1974; Blaas, 2004; Roos Klein-Lankhorst et al., 2002,

2004, 2005; Van Lammeren et al, 2010; Weitkamp, 2004, 2007, 2010). These types of studies

are still grounded in an expert tradition but do validate the results by expert and respondent

tests. These types of studies also bridge the expert-ecological approach and the psycho-physical

perception approach. Most of these studies are based on stated references and not on revealed

ones (Sevenant, 2010).

The psychological approach does have some Dutch examples. These are still in line with the

previous studies of Coeterier (1996). Especially photo and video montages have been used as

input data. The link with geodata is not always the case (Tress and Tress, 2003).

4.6.3 Phenomenological approach

The more recent studies, in relation to Web 2.0 developments and as promoted by Coeterier

(2002), which are in line with the humanistic or phenomenological, have been performed.

These are especially studies (Lammeren et al., 2009) in which an attempt at classifi cation of

landscape photographs, which were taken by tourists, has been made, in relation to tourist

landscape interest and their in-situ behaviour. In that study volunteer data (geodata as well as

geotagged photos) has been used.

The Web 2.0, including the trends of geotagged photographs and augmented reality, heavily

supports the challenging future of this research approach.

4 . 7 T E N D E N C I E S A N D P E R S P E C T I V E S

In the previous sections the connection between the expanding domain of geomatics and the

variety of physiognomic landscape studies have been outlined. Tendencies and perspectives

are derived from the Dutch studies. One item is very sure and in line with previous writings of

Ervin and Steinitz (2003). The nature of ongoing physiognomic landscape studies is not only

dominated by strict ecological, formal-esthetical, psycho-physical, psychological and phenom-

enological approaches. The studies that show a cross-reference approach are increasing and

seem very promising, thanks to geomatics.

4.7.1 Increase of geodata

The increase of geodata as result of institutional, societal and technology developments, show

a variety of physiognomic landscape representations and, by creating data ensembles, there

are many options to have geodatasets that fulfi l the defi nition of landscape entities and the

intended processing results. The most striking trends discovered are the increased precision of

data, three-dimensional geometry of objects, data type integration (supported by computer vi-

sion), 3D references (horizontal plus vertical references), 3D geo-scenes (due to 3D reference)

instead of 2D geodatasets, time series (initiated by digital photography and ground based Li-

dar) and even real-time geodata based on GPS tagged photos and video enabled smart phones.

These trends will bring forward the need for geodata standards for physiognomic landscape

studies including a related ontology.

4.7.2 Outcome of data processing

Geo-computational innovations improved options to calculate many thematic, geometric and

topology-based variables, and, even time-series based variables have dramatically increased.

Besides listing the variables they could be linked to the type of physiognomic landscape studies

(fi gure 9).

landscape classification

landscape manifestation

landscape configuration

scenery

existential meaning

spatial configuration

based on Wassink 1999

based on Antrop 2007

viewing
conditions

personal-
demographic

characteristics

landscaping
processes

natural substratum embedded factors

Figure 9

Physiognomic landscape research and landscape studies

92 Geomatics in physiognomic landscape research. A Dutch view

In relation to the availability of data ensembles the following types of variables are used:

• Basic vector and raster analysis tools based, including density and distribution variables
(description substratum and embedded factors; confi guration, classifi cation);

• Landscape metrics (confi guration, classifi cation);
• DEM derived, like slope, aspect, curvature (description substratum and embedded factors;

confi guration);

• 3D Boolean operators that support geometric and topologic variables (confi guration; mani-
festation; (re)classifi cation);

• Viewsphere, Viewshed and Isovist derived variables (manifestation; confi guration, classifi -
cation);

• Interface perception variables based on eye tracking, time responses, interface tracking
(manifestation; classifi cation).

4.7.3 Impact of visualisation

Important research stimuli came from the many ways to visualise geodata and the variety and

simplicity of interfaces. Based on the type of physiognomic landscape studies (fi gure 9) that

each have their peculiar sets of variables, it is obvious that a high variety of visualisations are in

use. These visualisations are primary based on the traditions of cartography.

In the perception-oriented studies’ landscape manifestation, geomatics have been used to cre-

ate (dynamic) landscape visualisations. Based on these visualisations derived variables related

to usability and affective appraisal have been the subject of studies.

The nature of the interface for both, derived variables visualisation and landscape visualisa-

tion, have become the subject of studies too.

4.7.4 Improving methodologies

The increase in data and new variables, the latter as result of processing options, has infl u-

enced methodology. Most striking are the options to compare variables in relation to measured,

perceived and simulated data. Even (cor)relations between variables can be generated like

viewing graphs and visibility paths. In all of them, landscape confi guration, manifestation and

classifi cation studies, an increase of variables can be discovered. There is an increased interest

in the nature of a data due to the role of volunteers to collect, to review and to respond on land-

scape data. Specifi cally, the demographics of users underpin the fi ndings of studies via demo-

graphic group based variables.

As noticed by Ervin and Steinitz (2003), even the so-called “viewer predisposition, or purpose”

and the impact of other senses-related variables to measure landscape characteristics and per-

ception, like noise, smell and crowdedness, become part of studies.

However one of the most important gains of geomatics is the automated reporting function.

This function captures datasets and processing steps by fl ow diagrams and meta-datasets,

which supports the discussion of results and makes an easy adaptation of the methodology pos-

sible.

4.7.5 Meeting previous demands

Let’s fi nish with the study of De Veer and Burrough (1978), who asked policy makers and consult-

ants to score applications of physiognomic landscape studies. Those days fi ve categories scored

high: vulnerability designation (e.g. visibility of a new building, road or power line); suitability

designation (e.g. for different recreation activities); public landscape preferences (e.g. as deter-

mined by a questionnaire using colour photographs of selected landscapes); landscape evaluation

(using parameters such as diversity, rareness, or replacement possibilities) for conservation plan-

ning and landscape design (the creation of new, or modifi cation of old landscapes).

The main conclusion from the questionnaire was that users’ demands for physiognomic land-

scape mapping vary enormously, both in terms of mapping scale and map content (De Veer

and Burrough, 1978). With the contribution of geomatics we may notice that the variation of

physiognomic landscape studies does increase. Applied geomatics in physiognomic landscape

research will dramatically increase by the availability of mobile Internet services that will sup-

port citizens to become more aware of their surrounding environment and to participate in

spatial planning procedures.

As mentioned by peers from many application domains (Tucci, 2010), the contribution of geo-

matics does not only consist of the application of the latest information technology based data

and data processes, but it helps to create new methodological pathways, especially related to

data acquisition and processing. Alongside that, geomatics proffer new and innovative ways of

describing reality, which offers a wider spatial range related to more precision and accuracy of

physiognomic features. Examples are: the enormous quantities of data relating to a single geo-

graphic location and generated at different times (past, present and future), the possible addi-

tion of extra variables to each of the representations during the research, the powerful extent

to which topology rules spatial analysis, the variety of visualisation options, the fact that data

can be created via volunteer sessions and Internet services and the acquired data, information

and knowledge can be widely disseminated on-line via external databases in the ‘cloud’ and

Internet sites.

94 Geomatics in physiognomic landscape research. A Dutch view

All together geomatics doesn’t only provide important contributions to physiognomic land-

scape studies, but it also create more awareness, different appreciation and more sustainable

utilisation of landscapes. At least that’s what may be proved in the near future.

The Dutch-Flemish physiognomic landscape research community tends to follow, and some-

times initiate, most of the here fore mentioned global geomatic developments. However this

community is not afraid of using data ensembles and variables in line with cross-reference ap-

proaches to evolve and challenge physiognomic landscape studies.

R E F E R E N C E S

Antrop, M. (2007) Perspectieven op het landschap: achter-

gronden om landschappen te lezen en te begrijpen. Gent,

Academia Press.

Batty, M. (2001) Exploring isovist fi elds: space and shape in

architectural and urban morphology. Environment and

Planning B: Planning and Design 28; 123–150

Benedikt, M.L. (1979) To take hold of space: isovists and

isovists fi elds. Environment and Planning B: Planning and

Design 6; 47-65

Blaas, M. (2004) Landscape Research Helmond: Geo-processed.

Wageningen University (MSc thesis Geo-Information

Science GRS 2004-1)

Bregt, A., and Lammeren, R. van (2000) 30 jaar GIS in Wage-

ningen. Kartografi sch Tijdschrift 26 (3); 15-21

Burrough, P.A., and Veer, A.A. de (1982) Een informatiesys-

teem landschapsbeeld. Automatische produktie van

landschapskaarten ten behoeve van de ruimtelijke pIan-

ning. Planning 17; 2-15

Burrough, P.A., and Veer, A.A. de (1984) Automated produc-

tion of landscape maps for physical planning in the

Netherlands. Landscape Planning ll; 205-226

Clevis, Q., Tucker, G.E., Lancaster, S. T., et al. (2006) A simple

algorithm for the mapping of TIN data onto a static grid:

Applied to the stratigraphic simulation of river meander

deposits. Computers & Geosciences 32; 749–766

Coeterier, J.F. (1996) Dominant attributes in the perception

and evaluation of the Dutch landscape. Landscape and

urban planning 34; 27-44

Coeterier, J.F. (2002) Lay people evaluation of historic sites.

Landscape and Urban Planning 59; 111–123

Canadian Council of Land Surveyors, Canadian Institute of

Geomatics, Geomatics Industry Association of Canada

(2000) Geomatics. (Human Resources Study)

Chrisman, N. (2003) Exploring geographic information

systems. 2nd edition. New York, etc., John Wiley &

Sons.

Daniel, T.C., and Vining, J. (1983) Methodological issues in

the assessment of landscape quality. In: Altman, I., and

Wohlwill, J. (eds.) Behavior and the Natural Environ-

ment: Advances in Theory and Research. New York, Ple-

num Press, pp 5-45

Daniel, T.C., and Meitner, M. (2001) Representational valid-

ity of landscape visualisations: the effects of graphical

realism on perceived scenic beauty of forest vistas. Jour-

nal of Environmental Psychology (2001) 21; 61-72

Floriani, L.D., and Magillo, P. (2003) Algorithms for visibility

computation on terrains: a survey. Environment and

Planning B: Planning and Design 30; 709–728

De Veer, A.A., and Burrough, P.A. (1978) Physiognomic land-

scape mapping in the Netherlands. Landscape Planning

5; 45-62

Döner, F., Thompson, R., Stoter, J., et al. (2010) 4D cadastres:

First analysis of legal, organizational, and technical

impact. With a case study on utility networks. Land Use

Policy 27; 1068–1081Policy 27; 1068–1081Policy

Dykes, J., MacEachren, A., and Kraak, M.J. (2005) Exploring

Geovisualization. In: Dykes, J., MacEachren, A., Kraak,

M.J. (eds.) Exploring Geovisualization. Amsterdam, etc.,

Elsevier, pp 3-19

Elsner, G.H. (1971) Computing Visible Areas from Proposed

Recreational Developments. A Case Study. Pacifi c South-

west Forest and Range Experimental Station, California.

Ervin, S., and Steinitz, C. (2003) Landscape visibility com-

putation: necessary, but not suffi cient. Environment and

Planning B: Planning and Design 30; 757 -766

Fisher, P.F. (1999) Models of uncertainty in spatial data. In:

Longley, P., Goodchild, M. et al. (eds.), Geographical

Information Systems, Second Edition. New York etc., John

Wiley & Sons. pp 191-205

Fisher, P.F., ed. (2006) Classics from IJGIS: Twenty years of the

International Journal of Geographical Information Science

and Systems. S.l.: CRC press

Gibson, J.J. (1966) The senses considered as perceptual systems.

Boston, Houghton-Miffl in.

Gibson, C.C, Ostrom, E. and Ahn, T.K. (2000) The concept

of scale and the human dimensions of global change: a

survey. Ecological Economics 32; 217–239

Hebblethwaite, R.L. (1973) Landscape assessment and clas-

sifi cation techniques. In: Lovejoy, D. (ed.) Land Use and

Landscape Planning. Aylesbury, Leonard Hill, pp 19-50

Hendriks, P., and Ottens H. (1997) Geografi sche Informatie

Systemen in ruimtelijk onderzoek. Assen, Van Gorcum.

Jones, C.B., Purves, R. S., et al. (2008) Modelling vague

places with knowledge from the Web. International

Journal of Geographical Information Science 22 (10);

1045–1065

Kamps, S. and Lammeren, R. van (2001) Quantitative land-

scape description. In: Conference Proceedings of ECLAS

2001, pp 64-75

Kent, M. (1985) Visibility analysis of mining and waste tip-

ping sites. A review. Landscape and Urban Planning 13;

101-110

Llobera, M. (2003) Extending GIS-based visual analysis: the

concept of visualscapes. International Journal for Geo-

graphical information science 17 (1); 25-48

Kerkstra, K., Vrijlandt, P., et al. (2007) Landschapsvisie Zuid

Limburg. Provincie Limburg/Wageningen University

Mark, D. (1999) Spatial representation: a cognitive view.

In: Longley, P., Goodchild, M. et al. (eds.) Geographical

Information Systems, Second Edition. New York etc., John

Wiley & Sons, pp 81-89

Marschallinger, R. (1996) A voxel visualation and analysis

system based on Autocad. Computers & Geosciences 22

(4), 379-386

McGarigal, K., Cushman, S.A., et al. (2002). FRAGSTATS: Spa-

tial Pattern Analysis Program for Categorical Maps. Am-

herst, University of Massachusetts (http://www.umass.

edu/landeco/research/fragstats/fragstats.html)

Molenaar, M. (1997) Conceptuele ruimtelijke gegevensmo-

dellen. In: Hendriks, P., and Ottens, H. (eds.) Geografi s-

che Informatie Systemen in ruimtelijk onderzoek. Assen,

Van Gorcum, pp 21-44

Ode, Å., Tveit, M.S., and Fry, G. (2010) Advantages of using

different data sources in assessment of landscape change

and its effect on visual scale. Ecological Indicators 10;

24–31

Palmer, J.F., and Hoffman, R.E. (2001) Rating reliability and

representation validity in scenic landscape assessments.

Landscape Urban Planning 54; 149–161

Pollefeys, M., Gool, L. van, Vergauwen, M., et al. (2002)

Video-to-3D. International archives of photogrammetry

remote sensing and spatial information sciences 34 (5);

579-584

Raper, J., Gartner, G., et al. (2007) Applications of location-

based services: A selected review. Journal of Location

Based Services 1; 89–111

Riggs, P.D., and Dean D.J. (2007) An Investigation into the

Causes of Errors and Inconsistencies in Predicted View-

sheds. Transactions in GIS 11(2); 175–196

Roos-Klein Lankhorst, J., Buijs, A.E., Berg, A.E. van den, et al.

(2002) Belevings GIS versie februari 2002. Wageningen,

Alterra (werkdocument 2002/08; planbureau-werk in

uitvoering)

Roos-Klein Lankhorst, J, Vries, S. de, Lith-Kranendonk, J. van,

et al. (2004) Modellen voor de graadmeters landschap,

beleving en recreatie; Kennismodel Effecten Land schap

Kwaliteit KELK, Monitoring Schaal, BelevingsGIS.

Wageningen, Natuurplanbureau vestiging Wageningen

(planbureaurapport 20)

Roos-Klein Lankhorst, J., Vries, S. de, Buijs, A., Berg, A.E. van

den, et al. (2005) Belevings GIS versie 2: waardering van

het Nederlandse landschap door de bevolking op kaart.

Wageningen, Alterra (alterra-rapport 1138)

Roswell, C. and Tom, H. (2009) Standards guide, ISO/TC 211

geographic information/geomatics (2009-06-01)

Rijksplanologische Dienst (1966) Tweede nota over de

ruimtelijke ordening in Nederland. The Hague.

Schroeder, H.W. (1988) Visual impact of hillside develop-

ment: comparison of measurements derived from aerial

and ground-level photographs. Landscape and Urban

Planning 15; 119–126

Sheppard, S. R. J., and Cizek, J. (2009) The ethics of Google

Earth: Crossing thresholds from spatial data to land-

scape visualisation. Journal of Environmental Manage-

ment 90(6); 2102–2117ment 90(6); 2102–2117ment

Sevenant, M. (2010) Variation in landscape perception and

preference. Experiences from cases studies in rural and

urban landscapes observed by different groups of respond-

ents. Ghent University, Department of Geography (PhD-

dissertation)

Snavely, N., Seitz, S.M., and Szeliski, R. (2006) Photo Tour-

ism. Exploring Photo Collections in 3D. ACM Transac-

tions on Graphics 25(3); 835-846

Szeliski, R. (2010) Computer Vision: Algorithms and Applica-

tions. University of Washington (draft may 2010).

Steenbergen, C.M., and Reh, W. (2003) Architecture and

Landscape: The Design Experiment of the Great European

Gardens and Landscapes. Basel etc., Birkhäuser.

Steenbergen, C.M., Meeks, S., and Nijhuis, S. (2008) Compos-

ing landscapes: analysis, typology and experiments for

design. Basel etc., Birkhäuser.

Steenbergen, C.M., Reh, W., Nijhuis, S., and Pouderoijen,

M. (2009) The polder atlas of the Netherlands. Bussum,

Thoth.

Sutherland, I.E. (2003) Sketchpad: A man-machine graphical

communication system. Based on the original dissertation

(1963). University of Cambridge, Computer Laboratory

(Technical report nr. 574).

Tandy, C.R. (1967) The Isovist Method of Landscape Survey,

in: C.R. Murray (ed.) Methods of Landscape Analysis.

London: Landscape Research Group, pp 9-10

Tang, T., Wenji, Z., et al. (2008) Terrestrial laser scan survey

and 3D TIN model construction of urban buildings in

a geospatial database. Geocarto International 23 (4);

259 - 272

Tomlin, D. (1990) Geographic information systems and carto-

graphic modeling. Prentice-Hall

96 Geomatics in physiognomic landscape research. A Dutch view

Tomlinson, R.F. Calkins, H.W. and Marble, D.F. (1976) Com-

puter handling of geographical data. UNESCO Natural

Resources Research XIII, Paris

Tress, B., and Tress, G. (2003) Scenario visualization for par-

ticipatory landscape planning. A study from Denmark.

Landscape and Urban Planning 64; 161–178

Tucci, G., Algostino, F., Bucalossi, L., et al. (2010) Cultural

heritage and sustainable valorization in the governorate

of Tratous with reference to the Euromed IV project: the

contribution of geomatics. In: Ioannides, M., Fellner, D.

et al. (eds.) Digital Heritage. Proceedings 3rd internation-

al conference, EuroMed 2010. New York, etc., Springer,

pp 399-408

Van den Berg, A. (2009) Handboek Natuurbeleving en gezond-

heid. Wageningen University (Course syllabus, October

2009)

Van Eetvelde, V. and Antrop, M. (2009) A stepwise multi-

scaled landscape typology and characterisation for

trans-regional integration, applied on the federal state of

Belgium. Landscape and Urban Planning 91: 160–170

Van der Ham, R.J.M., and Iding, J.A. (1971) De landschap-

stypologie naar visuele kenmerken. Methodiek en gebruik.

Wageningen, Afdeling Landschapsarchitectuur, Land-

bouwuniversiteit Wageningen

Van der Ham, R.J.M., Schut, G.F.E., and Iding, J.A. (1970)

Een voorstel voor een nieuwe landschapstypologie naar

visuele kenmerken. Stedebouw en Volkshuisvesting 11;

421-438

Van Lammeren, R., Goossen, M., and Roncken, P. (2009)

Sensing Landscape History with an Interactive Location

Based Service. Sensors (9); 7217-7233

Van Lammeren, R., J. Houtkamp, J. et al. (2010) Affective

appraisal of 3D land use visualization. In: Computers,

Environment and Urban Systems 34; 465-475

Van Leeuwen, M., and Nieuwenhuis, M. (2010) Retrieval of

forest structural parameters using LiDAR remote sens-

ing. European Journal of Forest Resource

Wascher, D. (2005) European Landscape Character Areas.

Typologies, Cartography and Indicators for the Assessment

of Sustainable Landscapes (Alterra Report No. 1254)

Wassink, W.Th. (1999) Beekdallandschappen. Een morfolo-

gisch onderzoek in de zandgebieden van Nederland. Wage-

ningen University (PhD-dissertation)

Weir, M.J.C. (1976) Landscape evaluation in Great Britain.

A short review. Mededelingen Werkgemeenschap Land-

schapsecologisch Onderzoek 3; 15–20

Weitkamp, G. (2004) Landschappelijke beleving geparametri-

seerd. Een methode voor het bepalen van vijf belevingsaspec-

ten aan de hand van de top10vector. Wageningen University

(MSc thesis Geo-Information Science GRS 2004-28)

Weitkamp, G., Bregt, A., Lammeren, R. van, and Berg, A. van

den (2007) Three sampling methods for visibility meas-

ures of landscape perception. In: Winter, S., Duckham,

M. et al. (eds.) COSIT 2007. New York, etc., Springer-

Verlag, pp 268-284

Weitkamp, G. (2010) Capturing the View: A GIS based proce-

dure to assess perceived landscape openness. Wageningen

University (PhD-dissertation)

Werkgroep Helmond (1974) Landschapsonderzoek Helmond.

Wageningen, Afdeling Landschapsarchitectuur Hoge-

school Wageningen.

Wherret, J.R. (2000) Creating Landscape Preference Models

Using Internet Survey Techniques. Landscape Research

25 (1); 79-96

Xu, W., Zhu, Q., et al. (2010) Design and Implementation of

3D Model Database for General-Purpose 3D GIS. Geo-

spatial Information Science 13 (3); 210-215

Yang, P.P., Putra, S.Y., and, Li, W. (2007) Viewsphere: a GIS-

based 3D visibility analysis for urban design evaluation.

Environment and Planning B: Planning and Design

Zhou, G. (2010) Geo-Referencing of Video Flow From Small

Low-Cost Civilian UAV IEEE. Transactions on automation

science and engineering 7 (1); 156-166

Zube, E.H., Sell, J.L., and Taylor J.G. (1982) Landscape

perception: research, application and theory. Landscape

planning 9; 1-33

[1] http://neo.aseanfl ag.com/?option=com_content&view

=article&id=65&Itemid=55

[2] http://www.igs.net/~schut/cli.html

[3] http://www.casa.ucl.ac.uk/gistimeline/ (time line only

available until 1/1/2000 )

[4] http://landsat.gsfc.nasa.gov/about/history.html

[5] http://www.fgdc.gov/policyandplanning/executive_

order

[6] http://www.tudelft.nl/live/pagina.jsp?id=d11f4b18-

63dc-46cc-a71e-14741f67928f&lang=nl

[7] http://www.vrom.nl/pagina.html?id=46094

[8] http://www.ahn.nl/

[9] http://inspire.jrc.ec.europa.eu/

[10] http://oreilly.com/web2/archive/what-is-web-20.html

[11] http://www.fl ickr.com/map/

[12] http://www.panoramio.com/

[13] http://www.locr.com/

[14] http://earth.google.com/

[15] http://www.bing.com/

[16] http://en.wikipedia.org/wiki/Voxel

[17] http://www2.jpl.nasa.gov/srtm/

[18] http://www.opengeospatial.org/

[19] http://bbs.keyhole.com/ubb/ubbthreads.php?ubb=sho

wfl at&Number=1323176

[20] see Google Earth – active layer 3D buildings

[21] http://www.photosynth.net/

[22] http://phototour.cs.washington.edu/

[23] http://www.google.com/intl/en_us/help/maps/

streetview/

URL (checked July 2010)