Applications of GIS II: Cartographic generalisation and pattern

GEO 243.1 | FS 2015!
Introduction to spatial analysis with GIS | Lecture 10
Applications of GIS II:
Cartographic generalisation and pattern
recognition"
Martin Tomko1"
Geographisches Institut!
[email protected]!
!
!
1
Inputs from R Weibel!
V10 | App GIS II: Generalisation
2 | Intro!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Last week(s)"
•  We have seen an example of GIS-based modeling —
modeling snowdrift — and we have deconstructed this
example from perspectives of !
–  Representation of reality in a GIS analytical process (GIS thinking
process)!
–  Choice of appropriate data models and data structures!
–  Choice of appropriate operations to operationalise these models!
•  You have participated ina GITTA tutorial on Suitability
Analysis and Multicriteria analysis.!
Today"
•  Second application example with focus on appllication of the
already familiar operations in a complex GIS analysis"
1
V10 | App GIS II: Generalisation
3 | Outlook!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Remaining weeks…"
•  04.05.2015: GITTA Tutorat „Accessibility“!
•  11.05.2015: Lecture „GIS in Switzerland“ à A quick look at
the GIS field in CH – industry, activities, governmental
organisations, data. Guest lecture from Dr. Ralph
Straumann!
•  18.05.2014: Repetitorium – email me questions you want to
cover. If no questions, no repetitorium!!!
•  22.05.2015 – handout of the last assignment (note different
date – error in the overview from start of semester.) !
•  25.05.2015: No lecture!
V10 | App GIS II: Generalisation
4 | Intro!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Outline: Deconstruction of an Application of GIS"
•  In this lecture we are again looking at some examples of
GIScience methods in geographical research
•  The examples are supposed to:
–  Show you some of the areas of research in GIScience;
–  Explain the context/positioning of GIScience with respect to physical
and human geography;!
–  Make you think about reality, conceptual models, data
structures and operations in the context of this research;
–  The examples are independent of software – they are about
dealing with information with some geographic component…;
–  The examples should reinforce the material I have already given
you (review the material!).
2
V10 | App GIS II: Generalisation
5 | Intro!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Learning objectives"
Using the example of cartographic generalisation and pattern
recognition, you will:!
ü  Learn that complex processes can be solved by
decomposition into simple partial operations (components);!
ü  You will be able to name and briefly describe the main
generalisation procedures;!
✓  You will be able to explain how pattern recognition can help
to uderstand and identify fundamental processes/objects;!
✓  You will be able to identify and apply the known operations
on fields to problems in pattern analysis and generalisation;!
✓  You will be able to explain the principles of simple line
generalisation algorithms ( Douglas-Peucker, smoothing
with moving window average).!
!
V10 | App GIS II: Generalisation
6 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Contents"
1.  Cartographic generalisation (Recap from GEO 113)"
2.  Breakdown of the overall process: Generalisation
operations!
3.  GIS Operations for cartographic generalisation!
– 
– 
Operations on Fields / Rasters!
Operation on Entities / Vectors!
4.  Cartographic pattern recognition!
– 
– 
What is a pattern!
GIS operations for pattern recognition!
3
V10 | App GIS II: Generalisation
7 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Cartographic generalisation = Abstraction"
•  When transiting from reality to a conceptual model, we
abstract the reality — we choose elements, that are
important (relevant) for us – for our use.!
•  Furthermore, when reducing the scale of a map, the
available space (paper, screen) is also reduced; map
symbols cover more space than the objects in reality."
IGN 1:25,000
IGN 1:100,000
V10 | App GIS II: Generalisation
8 | 1. Cartographic generalisation!
IGN 1:250,000
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Cartographic symbolisation = Generalisation"
•  As spatial databases are a-priori unsymbolised, we need to
plan how we represent spatial data (symbolisation, choice of
cartographic key). These symbols need space à we need
to generalise!
Spatial DB
Map
Symbolisation"
Generalisation"
4
V10 | App GIS II: Generalisation
9 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Cartographic generalisation: Definition"
•  Intl. Cartogr. Association (1973): “the selection and simplified representation of detail appropriate to the scale and/or the purpose of a map”!
•  Schweiz. Ges. für Kartographie (2002: 41): “Die Generalisierung ist die
massstabsgerechte inhaltliche und grafische Vereinfachung der
komplexen Wirklichkeit auf der Grundlage digitaler Landschaftsmodelle
oder Karten grösseren Massstabes. Sie besteht hauptsächlich aus der
zweckentsprechenden Auswahl und Zusammenfassung der Objekte
sowie der möglichst lagegenauen, charakteristischen, richtigen und
eindeutigen grafischen Darstellung.”!
•  Conscise (Weibel): Reduction of the content of a geodatabase retaining
the objects, forms and structures essential for the target scale and
purpose of the map.!
•  Shortest: “Remove unnecessary, keep relevant.“!
Ü ICA (International Cartographic Association) (1973): Multilingual dictionary of technical terms in cartography. Wiesbaden,
Franz Steiner Verlag!
Ü SGK (Schweiz. Gesellschaft für Kartographie) (2002): Topografische Karten – Kartengrafik und Generalisierung.
Kartografische Publikationsreihe, Nr. 16 (erhältlich bei http://www.kartographie.ch/publikationen/index.html)!
V10 | App GIS II: Generalisation
10 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Ü SGK (Schweiz. Gesellschaft für Kartographie) (2002): Topografische Karten –
Kartengrafik und Generalisierung. Kartografische Publikationsreihe, Nr. 16
(erhältlich bei http://www.kartographie.ch/publikationen/index.html)!
Cartographic generalisation: partial components"
5
V10 | App GIS II: Generalisation
11 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Ü SGK (Schweiz. Gesellschaft für Kartographie) (2002): Topografische Karten –
Kartengrafik und Generalisierung. Kartografische Publikationsreihe, Nr. 16
(erhältlich bei http://www.kartographie.ch/publikationen/index.html)!
Cartographic generalisation: partial components"
V10 | App GIS II: Generalisation
12 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Contents"
1.  Cartographic generalisation (Recap from GEO 113)!
2.  Breakdown of the overall process: Generalisation
operations"
3.  GIS Operations for cartographic generalisation!
– 
– 
Operations on Fields / Rasters!
Operation on Entities / Vectors!
4.  Cartographic pattern recognition!
– 
– 
What is a pattern!
GIS operations for pattern recognition!
6
V10 | App GIS II: Generalisation
13 | 2. Generalisation operations!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Reduction of complexity: components"
•  Manual generalisation by a cartographer – a
holistic process.!
•  This is too complex to be formalised and
automated in a single step!
•  Scientific approach to reduce complexity:
division into smaller, tractable components. !
•  For generalisation: generalization operators!
à Somewhat identifiable in the previous slide!
à Gen.operations allow for the deconstruction of the
holistic process of generalisation into tractable
parts (sub-processes).!
Figure: http://www.memini.it/dividi-e-impera/
Ü Harrie, L. & Weibel, R. (2007): Modelling the Overall Process of Generalisation. In: Mackaness, W.A., Ruas, A. & Sarjakoski,
L.T. (eds.): Generalisation of Geographic Information: Cartographic Modelling and Applications. Elsevier Science, 67-87.!
V10 | App GIS II: Generalisation
14 | 2. Generalisation operations!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Possible generalisation operations"
Ü Hake, G., Grünreich, D. & Meng, L. (2002): Kartographie. Berlin:
de Gruyter!
Ü Shea, K.S. & McMaster, R.B. (1989): Cartographic generalization
in a digital environment: When and how to generalize. Proceedings
Auto-Carto 9, pp. 56-67.!
7
V10 | App GIS II: Generalisation
15 | 2. Generalisation operations!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Challenges"
•  The division into individual operations simplifies the
complexity of the generalisation problem – but we need to
find concrete solutions for each generalisation operation.!
!
!
!
!HOW to …?!
•  A Generalisation operator (e.g., line simplification) can be
operationalised by implementing different generalisation
algorithms.!
•  These algorithms can be distinguished with respect to the
applicable object class, target scale, etc…!
“A line is not always a line”
(What do these examples represent?)
V10 | App GIS II: Generalisation
16 | 2. Generalisation operations!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example: Generalisation of streets"
Toolbox of algorithms for the simplification of streets (Lecordix ‘98)
Min Break! Max Break! Accordeon!
Plaster!
Gaussian!
Sec. 3
Douglas!
Ü Lecordix, F., Plazanet, C. & Lagrange, J.-P. (1997): A Platform for Research in Generalization: Application to Caricature.
GeoInformatica, 1(2): 161-182.!
8
V10 | App GIS II: Generalisation
17 | 2. Generalisation operations!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example: Generalisation of streets"
Example of a context-sensitive operation: Feature
Displacement using Elastic Beams.!
Before!
After!
Ü Bader, M. (2001): Energy Minimizing Methods for Feature Displacement in Map Generalization. Dissertation, Geogr. Inst. UZH.!
Ü Bader, M., Barrault, M. & Weibel, R. (2005): Building Displacement by Means of a Ductile Truss. Int. Journal of Geographical
Information Science, 19(8/9): 915-936.!
V10 | App GIS II: Generalisation
18 | 2. Generalisation operations!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
What does this mean for GEO 243?"
•  These algorihtms:!
–  Apply mathematical methods and geometric algorithms that will be
covered in Master studies;!
–  These are not available in generic GIS (GeoMedia und ArcGIS), but
they are implemented in specialist software (1Spatial products)!
•  “Normal” GIS also expose several cartographic
generalisation operations!
•  By being creative and combining and parametrising these
operations, one can achieve a lot.!
•  Example of automatic generalisation using ArcGIS for Dutch
maps: Stoter et al. (2014)!
Ü Stoter, J., Post, M., van Altena, V., Nijhuis, R. & Bruns, B. (2014): Fully automated generalization of a 1:50k map from 1:10k
data. Cartography and Geographic Information Science, 41(1).!
9
V10 | App GIS II: Generalisation
19 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Contents"
1.  Cartographic generalisation (Recap from GEO 113)!
2.  Breakdown of the overall process: Generalisation
operations!
3.  GIS Operations for cartographic generalisation"
– 
– 
Operations on Fields / Rasters"
Operation on Entities / Vectors!
4.  Cartographic pattern recognition!
– 
– 
What is a pattern!
GIS operations for pattern recognition!
V10 | App GIS II: Generalisation
20 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Generalisation Operators on Fields/Rasters"
•  Cartographic data are mostly vectors. They allow:!
–  rich attributes!
–  flexible symbolisation!
–  Target resolution is well tuneable – not bound to a fixed raster cell
size!
•  Generalisation is however possible also on fields/rasters:!
–  Needed for raster GIS data (results of models, e.g., suitability
analysis).!
–  Raster operations are computationally simpler.!
–  Rasters are better at representing uncertainty (e.g., gradual
transitions in the case of uncertain boundaries).!
10
V10 | App GIS II: Generalisation
21 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Examples of Field/Raster Operations"
•  We cover three operations on fields/rasters:!
–  Smoothing filter!
–  Majority filter!
–  Mathematical Morphology!
•  All are available in Geomedia"
V10 | App GIS II: Generalisation
22 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Smoothing filter"
Original
3x3 Window
7x7
15x15
11
V10 | App GIS II: Generalisation
23 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Majority filter "
SlopeHangneigung
≤ 30 (black)
V10 | App GIS II: Generalisation
24 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Majority filter "
Original
3x3 Window
7x7 Window
12
V10 | App GIS II: Generalisation
25 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Majority filter "
Original, zoomed in
3x3 Window
7x7 Window
V10 | App GIS II: Generalisation
26 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Examples of Field/Raster Operations"
•  What algorithms can one use to realise
these operations?!
–  You already have the basics.!
•  What commands would you use in
GeoMedia Grid?!
13
V10 | App GIS II: Generalisation
27 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Smoothing and majority filter"
•  Realisation using focal operations:!
–  Smoothing filter: Average or Median within a moving window à
known from GZGI (GEO 123.2)!
–  Majority filter: Majority – value of the most frequent class within the
moving window substituted for current value!
•  In GeoMedia Grid: Grid > Statistical > Local Scan!
–  (“Local” is confusing, as this is not a local operation)!
•  Larger windows, more bandwidth ==> more smoothing!
•  Possible with/without weighting!
Focal Majority of
1
1
1
0
1
1
1
0
0
1
1
0
1
1
1
0
1
1
0
0
1
1
1
0
0
1
1
0
1
1
1
0
=
V10 | App GIS II: Generalisation
28 | 3. GIS Operations for generalisations (fields/rasters)!
1
1
Changed value
“Undecideable”
(border effect)
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Mathematical Morphology"
•  Mathematical Morphology: operations of image processing based
on set theory and topology.!
•  A “Structuring Element” B is applied on the image.!
•  Most important operations (for us):!
– 
– 
– 
– 
Erosion: reduction by k cells (where k = (WidthB) / 2 +1)!
Dilation: expansion by k cells!
Opening: Erosion, followed by Dilation!
Closing: Dilation, followed by Erosion!
Structuring
Element B
Erosion
Dilatation
Opening
Closing
Ü http://en.wikipedia.org/wiki/Mathematical_morphology !
14
V10 | App GIS II: Generalisation
29 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Mathematical Morphology"
•  GeoMedia - no operations for Mathematical Morphology.!
•  Fantasy!: you can simulate Erosion, Dilation, Opening und
Closing:!
–  Use Grid > Zone > Buffer"
–  “Structuring Element” - width of the buffer (= k cells)!
•  Following examples show “Closing” (Dilation, followed by
Erosion) with k = 5, where:!
–  black: “foreground” (Slope ≤ 30 Grad)!
–  yellow: “background” (Slope > 30 Grad)!
V10 | App GIS II: Generalisation
30 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Mathematical Morphology"
Original, enlarged
Dilation with k = 5;
Grey indicates distance
Binary
Rendering of dilation
15
V10 | App GIS II: Generalisation
31 | 3. GIS Operations for generalisations (fields/rasters)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Mathematical Morphology"
Original
Erosion with k = 5;
Grey indicates distance
Binary dilation
Binary erosion
= Final result of Closing
V10 | App GIS II: Generalisation
32 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Contents"
1.  Cartographic generalisation (Recap from GEO 113)!
2.  Breakdown of the overall process: Generalisation
operations!
3.  GIS Operations for cartographic generalisation!
– 
– 
Operations on Fields / Rasters!
Operation on Entities / Vectors"
4.  Cartographic pattern recognition!
– 
– 
What is a pattern!
GIS operations for pattern recognition!
16
V10 | App GIS II: Generalisation
33 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Generalisation operations on Entities / Vectors"
•  The majority of cartographic data are vectords (e.g.
VECTOR25) which have advantages when it comes to
symbolisation etc!
•  BUT: Generalisation algorithms for vector data are more
complex than the equivalent algorithms for rasters!
•  Despite that, all commerical GIS contain at least
rudimentary functions for vector generalisation!
•  In GeoMedia:"
–  Analysis > Aggregate: not only for Spatial Join (see Üb. 3), but also
for aggregation of neighbouring polygons or lines sharing an attribute
value!
–  Tools > Simplify: Douglas-Peucker Algorithm!
–  Tools > Smooth: Line smoothing using moving average!
V10 | App GIS II: Generalisation
34 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Simplification vs. Smoothing (McMaster & Shea 1992)"
•  Simplification of shape
through elimination of
vertices!
•  Output line contains a
subset of the original
vertices.!
•  Smoothing through
Approximation (e.g.,
average of coordinates)!
•  Output line typically contains
the same amount or more
vertices than the input!
Ü McMaster, R.B. & Shea, K.S. (1992): Generalization in Digital Cartography. Association of American
Geographers.!
17
V10 | App GIS II: Generalisation
35 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Line Simplification"
•  Task: “Simplify the shape of a cartographic line by removal
of ‘unnecessary’ vertices”. (nach McMaster and Shea, 1992)!
•  Suited for lines with sharp edges (e.g., boundaries)[mostly
man made boundaries].!
•  Problem:"
–  How to define “unnecessary” or “irrelevant”?!
–  What geometric or semantic criteria do you apply to select the
important vertices?!
V10 | App GIS II: Generalisation
36 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Douglas & Peucker algorithms (1973)"
•  “Classical”, broadly used —discussed in GEO 113.!
•  Uses a Tolerance band: Vertices, that exceed this band are considered
important and are retained.!
http://commons.wikimedia.org/wiki/File:Douglas-Peucker_animated.gif
Ü Douglas, D.H. & Peucker, T.K. (1973): Algorithms for the Reduction of the Number of Points Required to
Represent a Digitized Line or its Caricature. Canadian Cartographer, 10(2): 112-122.!
18
V10 | App GIS II: Generalisation
37 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Douglas & Peucker algorithms (1973)"
•  “Classical”, broadly used —discussed in GEO 113.!
•  Uses a Tolerance band: Vertices, that exceed this band are considered
important and are retained.!
•  Animation: http://gitta.info/Generalisati/en/html/GenMethods_learningObject3.html
Ü Douglas, D.H. & Peucker, T.K. (1973): Algorithms for the Reduction of the Number of Points Required to
Represent a Digitized Line or its Caricature. Canadian Cartographer, 10(2): 112-122.!
V10 | App GIS II: Generalisation
38 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example Douglas/Peucker: Wanderweg bei S-chanf"
Original line
312 Vertices
100 m
200 m
19
V10 | App GIS II: Generalisation
39 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example Douglas/Peucker: Wanderweg bei S-chanf"
Tolerance ε = 5 m (accuracy of Vector25 = 3 - 8 m)
102 Vertices
100 m
200 m
V10 | App GIS II: Generalisation
40 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example Douglas/Peucker: Wanderweg bei S-chanf"
Tolerance ε = 10 m
61 Vertices
100 m
200 m
20
V10 | App GIS II: Generalisation
41 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example Douglas/Peucker: Wanderweg bei S-chanf"
Tolerance ε = 50 m
20 Vertices
100 m
200 m
V10 | App GIS II: Generalisation
42 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Beispiel Douglas/Peucker: Wanderweg bei S-chanf"
Toleranz ε = 100 m
17 Vertices
100 m
200 m
21
V10 | App GIS II: Generalisation
43 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example Douglas/Peucker: Wanderweg bei S-chanf"
Tolerance ε = 200m
11 Vertices
100 m
200 m
V10 | App GIS II: Generalisation
44 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example Douglas/Peucker: Wanderweg bei S-chanf"
Tolerance ε = 500m
3 Vertices
100 m
200 m
22
V10 | App GIS II: Generalisation
45 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Line Smoothing"
•  Task: shape simplification through smoothing and approximation of the
original line.!
•  Suitable for lines with smoothly transitioning, “rounded” forms (e.g.
rivers, railways)!
•  Corresponds to low pass filtering or moving average!
•  Parameters: window size (obligatory), weight (optional; here: distance)
from central vertex p[i])!
Does this look familiar?
Many variants/option for
weighting and the tuning
of the smoothing are
possible (McMaster/Shea)
Ü McMaster, R.B. & Shea, K.S. (1992): Generalization in Digital Cartography. Assoc. of American Geographers.!
V10 | App GIS II: Generalisation
46 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example of line smoothing using moving average"
Original line
312 Vertices
100 m
200 m
23
V10 | App GIS II: Generalisation
47 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example of line smoothing using moving average"
Window size ± 7 points (total 15)
Equally weighted points
312 Vertices
100 m
200 m
V10 | App GIS II: Generalisation
48 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example of line smoothing using moving average"
Window size ± 7 points (total 15)
Equally weighted points
312 Vertices
100 m
200 m
24
V10 | App GIS II: Generalisation
49 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Example of line smoothing using moving average"
Window size ± 3, 7, 15, 21 vertices
Equally weighted points
312 Vertices
100 m
200 m
V10 | App GIS II: Generalisation
50 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Line simplification of polygon boundaries"
Original:
•  Vegetation polygon from the
SNP
25
V10 | App GIS II: Generalisation
51 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Line simplification of polygon boundaries"
Line simplification:
•  Tolerance = 10 m
V10 | App GIS II: Generalisation
52 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Line simplification of polygon boundaries"
Line simplification:
•  Tolerance = 20 m
26
V10 | App GIS II: Generalisation
53 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Line simplification of polygon boundaries"
Line simplification:
•  Tolerance = 50 m
V10 | App GIS II: Generalisation
54 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Combination line simplification and smoothing"
Simplification:
•  Tolerance = 50 m
Smoothing:
•  Window size ± 3 Pte
•  Weight 0.5 for vertices
other than central vertex
27
V10 | App GIS II: Generalisation
55 | 3. GIS Operations for generalisations (entities/vectors)!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Polygon simplification – beware of topology"
www.aurin.org.au - Morandini et al, 2013,2014
V10 | App GIS II: Generalisation
56 | 1. Cartographic generalisation!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Contents"
1.  Cartographic generalisation (Recap from GEO 113)!
2.  Breakdown of the overall process: Generalisation
operations!
3.  GIS Operations for cartographic generalisation!
– 
– 
Operations on Fields / Rasters!
Operation on Entities / Vectors!
4.  Cartographic pattern recognition"
– 
– 
What is a pattern!
GIS operations for pattern recognition!
28
V10 | App GIS II: Generalisation
57 | 4. Cartographic pattern analysis!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Pattern recognition in generalisation"
•  Good generalisation requires deep understanding of the
map:!
–  Meaning of the individual objects!
–  Distinguishing between relevant and irrelevant objects!
•  Example: generalisation of settlements – buildings in city
centers are generalised differently to those in suburban
areas.!
•  BUT: “city centers” are not well defined and typically are not
explicitly stored in spatial databases as objects!
•  “city centers”, “Inner cities” or “CBDs” are important also in
other contexts:!
–  Spatial planning, traffic planning!
–  Search engines, to find “Hotels in the center of Zurich”!
V10 | App GIS II: Generalisation
58 | 4. Cartographic pattern analysis!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Identification of city centers"
•  If city centers are not identified explicitly, they may be
implicitly stored in spatial DBs - can they be identified
thorugh pattern discovery techniques?!
•  Idea"
–  In a topographic DB there are many objects that belong to a city
center;!
–  If one can identify high numbers of these, or if they are clustered in
some locations in high concentration, then the “cityness” is high;!
•  Lüscher & Weibel (2013) tried this for British cities.!
Ü Lüscher, P. & Weibel, R. (2013). Exploiting Empirical Knowledge for Automatic Delineation of City Centres from Large-scale
Topographic Databases. Computers, Environment and Urban Systems, 37(1): 18-34.!
29
V10 | App GIS II: Generalisation
59 | 4. Cartographic pattern analysis!
Overview"
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Special procedure – not covered here
(see. Lüscher et al. 2009)
KDE
Buffer
Web-based questionnaire
to identify parameters
of city centers
Ü Lüscher, P., Weibel, R. & Burghardt, D. (2009). Integrating ontological modelling and Bayesian inference for the recognition
of urban concepts in cartographic vector data. Computers, Environment and Urban Systems, 33(5): 363-374.!
V10 | App GIS II: Generalisation
60 | 4. Cartographic pattern analysis!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Definition of the pattern “city center”"
•  This procedure is similar to a Multi–criteria evaluation
(MCE).!
•  As for a MCE, we must “operationalise” the criteria that are
used to characterise a city center.!
•  BUT: a city center is a vaguely defined concept – which
relates to human cognition and conceptualisation.!
•  Thus, a user experiment has been conducted to identify the
most important characteristics/criteria of a city center :!
–  Internet –based survey of UK residents!
30
V10 | App GIS II: Generalisation
61 | 4. Cartographic pattern analysis!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Operationalisation of a “city center”"
3 types of object classes have been
defined to operationalise a city center:
•  F = Frequency-based: frequent
objects à KDE
•  L = Landmark-like: rare objects
(Town hall, cathedralHBF) à
euklidean buffer
•  A = Area-like: Objects, that can
not be represented as a set of
objects (residential areas,
industrial zones) à special
procedure, not covered here
(Lüscher et al. 2009).
V10 | App GIS II: Generalisation
62 | 4. Cartographic pattern analysis!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Computation of a City Centre Typicality"
•  For each of the 17 object classes a separate surface is computed.!
•  The city center typicality for a point in space is a weighted sum of the 17
surfaces à local operation over 17 layers!
•  In the formula below:!
–  cm and ca are normalisation constants, so that 0 ≤ typicalitycitycentre ≤ 1.!
–  ca stands for all the negative criteria: retail parks (wret), Industrial areas (wind),
residential areas (wres), natural areas (wnatural).!
31
V10 | App GIS II: Generalisation
63 | 4. Cartographic pattern analysis!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Results"
(b) Isolines of
typicality for
Bristolà a
Typicality of 0.5
was applied as
limit criterion
(a) City Centre Typicality vs. the area of the city centers. Peaks – multicenter cities with
auxiliary centers (e.g., Oerlikon in Zürich)
V10 | App GIS II: Generalisation
64 | 4. Cartographic pattern analysis!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Evaluation of the results through comparison"
First, the resulting city centers
have been compared with
reference polygons through spatial
intersection and the degree of
match evaluated à precision,
recall and F1-Score:
Reference polygons: collected
from guide books, planning
documents etc...
32
V10 | App GIS II: Generalisation
65 | 4. Cartographic pattern analysis!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Evaluation of the results through comparison (2)"
In a second round, a KDE surface of point coordinates of georeferenced Flick
photos (flickr.com) has been computed and a 80%-Isoline polygon of this surface
has been intersected with the typicaity polygons, and again analysed through
Precision, Recall and F1-Score.
V10 | App GIS II: Generalisation
66 | Wrap-up!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
Summary"
•  Example: Cartographic generalisation as a representative of a
complex holistic process.!
•  Division into partial processes (steps) enables automation.!
•  These partial processes can be executed as common GIS operations.!
•  Many of these GIS operations are already familiar to you:!
–  Focal operations for the generalisation of raster data!
–  Moving average for smoothing of Lines (a 1D focal operation)"
–  Multi – criteria analysis for the identification of city centers, and the
operationalisation of ist criteria!
–  KDE and euklidean distance for the realisation of partial
parameters of typicality of city centers!
–  Weighted summation as a Local operation over 17 layers!
–  Spatial Intersection to compare polygons (Evaluation of the
Results)!
33
V10 | App GIS II: Generalisation
67 | References!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
References"
•  Bader, M. (2001): Energy Minimizing Methods for Feature Displacement
in Map Generalization. Dissertation, Geogr. Inst. UZH.!
•  Bader, M., Barrault, M. & Weibel, R. (2005): Building Displacement by
Means of a Ductile Truss. Int. Journal of Geographical Information
Science, 19(8/9): 915-936.!
•  Douglas, D.H. & Peucker, T.K. (1973): Algorithms for the Reduction of
the Number of Points Required to Represent a Digitized Line or its
Caricature. Canadian Cartographer, 10(2): 112-122.!
•  Hake, G., Grünreich, D. & Meng, L. (2002): Kartographie. Berlin: de
Gruyter!
•  Harrie, L. & Weibel, R. (2007): Modelling the Overall Process of
Generalisation. In: Mackaness, W.A., Ruas, A. & Sarjakoski, L.T. (eds.):
Generalisation of Geographic Information: Cartographic Modelling and
Applications. Elsevier Science, 67-87.!
V10 | App GIS II: Generalisation
68 | References!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
References"
•  Lecordix, F., Plazanet, C. & Lagrange, J.-P. (1997): A Platform for
Research in Generalization: Application to Caricature. GeoInformatica,
1(2): 161-182.!
•  Lüscher, P., Weibel, R. & Burghardt, D. (2009). Integrating ontological
modelling and Bayesian inference for the recognition of urban concepts
in cartographic vector data. Computers, Environment and Urban
Systems, 33(5): 363-374.!
•  Lüscher, P. & Weibel, R. (2013). Exploiting Empirical Knowledge for
Automatic Delineation of City Centres from Large-scale Topographic
Databases. Computers, Environment and Urban Systems, 37(1): 18-34.!
•  McMaster, R.B. & Shea, K.S. (1992): Generalization in Digital
Cartography. Association of American Geographers.!
34
V10 | App GIS II: Generalisation
69 | References!
GEO 243.1 | Intro to spatial analysis with GIS | FS 2015
M. Tomko, GIUZ, Uni Zürich
References"
•  SGK (Schweiz. Gesellschaft für Kartographie) (2002): Topografische
Karten – Kartengrafik und Generalisierung. Kartografische
Publikationsreihe, Nr. 16 (erhältlich bei
http://www.kartographie.ch/publikationen/index.html)!
•  Shea, K.S. & McMaster, R.B. (1989): Cartographic generalization in a
digital environment: When and how to generalize. Proceedings AutoCarto 9, pp. 56-67.!
•  Stoter, J., Post, M., van Altena, V., Nijhuis, R. & Bruns, B. (2014): Fully
automated generalization of a 1:50k map from 1:10k data. Cartography
and Geographic Information Science, 41(1).!
35