You are on page 1of 58

RASTER OVERLAY ANALYSIS

(FACTOR COMBINATION)

GmE 203 Principles of GIS


Lecturer: Mylene Villanueva-Jerez, M.Sc.
GIS Analysis Functions

• Query (e.g., identify, select)


• Recoding (reclassify)
• Proximity analysis (buffering, distance)
• Neighborhood analyses (e.g., average, variety)
• Arithmetic operations
• Terrain analysis (e.g., slope, aspect, viewsheds)
• Overlay analysis (logical and arithmetic)
• Network analysis (e.g., routing, allocation)
• Spatial modeling (simulation, projection)
Overlay Analysis

• New spatial data layer from two or more old data layers
– Basic principle : Compare the properties of the
same location in both data layers, and to produce
a new characteristic for that location in the output
layer.
• Can be done within vector and raster
• Should be georeferenced in the same coordinate system and
• Should cover the same area of interest
Overlay Analysis (Raster)

• Cell by cell process which results in the combination


of the two input layers
• Pay attention to the number of possible
combinations that may be possible and understand
the effect on the output layer
Overlay Analysis (Raster)
Overlay Analysis (Raster)
Example using conceptual model to
create a suitability map

Step 1. Stating the problem

Step 2. Breaking the problem down


Process models

Datasets
(data models &
representation
model)
•Step 3. Exploring the datasets

Datasets
(data models &
representation
model)
Step 4. Performing analysis (spatial analysis)
Combination of Factor Maps
• several well defined approaches to factor
combination exist

• these may be applied to land capability


evaluation, habitat assessment, environmental
impact analysis, etc

• the validity of each technique depends upon the


scaling of the factors and the nature of their
interdependence
Key References:
• McHarg I. (1969) Design with Nature, Doubleday, Garden City
NY

• Hopkins L.D. (1977) Methods for generating land suitability


maps: a comparative evaluation, American Institute of
Planning Journal, 4,386-400

• Gordon S.I. (1985) Computer models in environmental


planning, Van Norstrand Reinhold, New York (Ch 5 - Land
capability evaluation)
Factor Combination Techniques
• Boolean analysis (exclusion mapping)
• Ordinal overlay (McHarg technique)
• Linear overlay (rating and weighting)
• Non-linear overlay (physical models)
• Rule of Combination
• Hierarchical approaches
Factor Interdependence
• several procedures make an implicit assumption that
factors are independent of one another

• BUT - Is the influence of soil type the same no matter


what the slope?

• IF NOT - How can such interdependence be handled?


Influence of Factor Scales
Factor Scale Available Combination Interdependence
Approaches

Binary Boolean Logic (AND, OR, XOR, NOT) No


Nominal Rules of Combination Yes
Ordinal Maximum, Minimum (Often No
misused in linear overlay)

Interval Linear, Non-linear and Rules of No/Yes


Combination

Ratio Linear, Non-linear and Rules of No/Yes


Combination
Boolean Analysis
• A Boolean approach to land capability or suitability is
sometimes referred to as "Exclusion Mapping"

• This is because the Boolean approach is a binary IN or


OUT approach with no shades of grey.

• With regard to any particular factor, a land use, activity


etc. is either considered OK or is excluded

• The Boolean approach can be combined with other


techniques when appropriate
Boolean Logic
• Boolean, or logical overlays combine two or more
maps based on a set of logical relationships:
– Logical Intersection: where condition on both maps
must be true. Map A AND Map B
– Logical Union: where condition on one map or the
other (or both must be true) Map A OR Map B
– Exclusive OR: where condition on one and only one
map must be true: Map XOR Map B
– NOT (logical negation): if values on a map are true
they become false, if false they become true: NOT
Map A
In Computers
True is represented as a 1 and
False is represented as a 0

0
1

• These are referred to as BINARY maps


Logical Intersection
• Both regions must be true

Map A Map B

Intersect = MapA AND MapB


Logical Union
• Either one or both maps may be true

Map A Map B

Union = MapA OR MapB


Exclusive OR (XOR)
• Either one map or the other must be true (but
NOT BOTH)

Map A Map B

Exclusive OR = Map A XOR Map B


Map Negation/Inversion (NOT)
• The values become false, false become true

Map A NOT Map A

Negate = NOT Map A


Comparison of Boolean Operators

Map A AND Map B Map A OR Map B


Logical Intersections Logical Unions MUST
MUST MEET ALL criteria MEET ANY criteria

Map A XOR Map B Map A AND NOT Map B


Exclusive OR’s meet one or
the other but NOT BOTH
Implementing Boolean Operations
• For each map to be overlayed, create a new
Boolean field which assigns a 1 to each map
value to be included in the evaluation and a 0
to each map value to be excluded in the
evaluation
Example Reclassification
• Reclass MAP1 to create a new map MAP1A. If
regions coded 2 and 5 only are suitable then set
regions 2 and 5 to 1 and all other map values to 0.
Map1 Map1A
Repeat the reclass operation for all
maps you wish to combine
Map2 Map3

New Map2a from Map2 assigning New Map3a from Map3 assigning
0 to 1 and 4 and assigning 0 to 6 1 to 4 and assigning 0 to 2
The resulting binary maps look like these:

Map1A Map2A

Map3A
Intersect = Map1A AND Map2A AND Map3A

Intersect

The map overlay resulting from the Boolean operation shows


the intersection of the regions from the three input overlays.
Union = Map1A OR Map2A OR Map3A

Union

The binary map overlay resulting from the Boolean OR operation


shows the union of the regions from the three input overlays.
The sequence of the AND and OR
operations is IMPORTANT
• Consider the difference between the following
two operations
– TEST1 = map1a AND map2a
– RESULT1 = test1 OR map3a

– TEST2 = map2a OR map3a


– RESULT2 = test2 AND map1a

• Will the results of these operations be the same


or different?
Result of First Test
Result of Second Test
Comparison of Two Tests

Note the difference between the two results. Errors in ordering


operations is one of the most common errors in Boolean Overlays.
BOOLEAN Syntax
• The previous examples can be combined in a single
operation as follows

test1 = map1a AND map2a


result1 = test1 OR map3a

result1 = (map1a AND map2a) OR map3a

test2 = map2a OR map3a


result2 = test2 AND map1a

result2 = (map2a OR map3a) AND map1a


ORDINAL DATA
• Ordinal Data contains an implication of value
by sequence (good, better, best)
• There is no quantification of the relativities
• Standard mathematical operations are not
valid on qualitative ordinal data
• Factor maps with ordinal scaling should not be
added , multiplied, etc
Popular Usage
• McHarg made manual overlay of Ordinal data
popular
• Continues to be used as if valid
• Order implicitly treated as intervals
Correct Usage
• There is a mode of analysis for ordinal data
which is valid
• This is based on the concept of limiting factor
• It is assumed that for example a land use is
constrained by the lowest rated factor
• It is then appropriate to use the GIS to find the
minimum value by factor in each cell
• In other cases the maximum set of values may
be sought
Minimum Function

= D = min (A, B, C)
EXAMPLES OF ORDINAL FACTOR
COMBINATION
Analog
representa
tions of
ordinal
capability
factors

McHarg, I. (1969) Design


with Nature, Doubleday,
New York
Analog
overlay
(equivalent to
unweighted
linear overlay -
not
mathematicall
y appropriate)

McHarg, I. (1969) Design


with Nature, Doubleday,
New York
Assessment of ordinal capability status
by factor

Xiang, Wei-Ning (1996) A GIS based method for trail alignment


planning, Landscape and Urban Planning, 35,11-23.
The study area
(Sterling Forest)
Environmental
limitation
categories (ordinal
rankings)
Mapping of
Limitations
Combination of
limitations
(using
Maximum)
Change of
development
plans as
consequence
of analysis
Example of
flow-charting
for capability
assessment (no
operations
shown)

Aramburu, M.P. and Escribano, R.


(1993) Golf: a conflicting
recreational activity in the Madrid
Autonomous Area, Landscape
and Urban Planning, 23, 209-220.
More sophisticated example of flow-
charting with operations shown
Linear Overlay
• The process of linear overlay recognizes:
• 1) that ordinal attribute scaling is inadequate for
multiple factor combination (i.e rating is required
rather than ranking), and
• 2) that different factors can have different
significance in their contribution to suitability (i.e.
factors need to be differentially weighted).
• Consequently this is sometime called the rating &
weighting approach.
Linear Overlay
In general
• Suitability = a1 * factor_1 + a2 * factor_2 + ............an * factor_n

• where a1, a2 ....an are the weight applied to the factors


• the equation is evaluated on a cell-by-cell basis (i.e. there is no
capacity to deal with adjacency)
• the factors must run over the same range of values (e.g. 0 to 10)
otherwise there is an implicit weighting in favour of the factor with
the larger range
• the equation is wholly linear (i.e. there is no capacity to deal with
interdependence)
• a linear overlay equation may also be the outcome of linear
regression analysis
Factor Weighting
• Various procedures exist for determining the appropriate
weightings to be applied to the factors.
– from expert opinion or public consultation of some combination of the
two.
– weighting must eventually be a true reflection of the relative
importance of the factors.
– Some authors have suggested an approached based on subdivision of
a finite total weight: given 100 points distribute these among the
factors based on their relative importance. Other approaches included
giving a weight of 1 to the least important factor and then building up
from that point, similar is giving the most important factor a specific
weight (e.g. 10) and working down from there.
– Note that if the ratio of weight between the most important and least
important factor is greater than around 10 then the contribution of
the least important factor to the overall rating is going to be
insignificant.
AHP - the analytical hierarchy
approach - developed by Saaty (1980).
AHP
• The analysis proceeds by conducting a pairwise
comparison of the factors (or policies or objectives).
Each participant is asked independently to establish the
relationships among the factors using the ratio scale of
measurement (see table 1).
• For each decision-maker, therefore, a square objectives
matrix is generated. Next, the matrices are normalised.
This is done by adding the column elements and
dividing each element by the respective column sum.
The rows of the normalised matrix of objectives are
averaged to yield the relative weights. The relative
weights (or factor weightings) are considered
representative of the preferences of the participant.
AHP
• The procedure contains a simple mechanism for
checking the consistency of the judgment made.
• Saaty (1980) illustrates, "if apples are preferred to
oranges and oranges are preferred to bananas, then
apples must be preferred to bananas. Moreover, if
apples are twice as preferable as oranges and oranges
are three times as preferable as bananas, then apples
must be six times as preferable as bananas."
• If the inconsistency ratio - calculated from the
eigenvector of the weighting matrix - is greater than
the chosen threshold then the judgments should be
reconsidered.
Non-linear Overlay
• This category of factor combination includes a wide
range of equation based models: non-linear
regression models, the USLE, cellular automata (e.g.
starlogo and example), neural-network models etc. In
general these techniques are not used for land
capability or suitability mapping but rather for
process and predictive modelling. What will the soil
loss be? In which direction will urban growth occur?
How much traffic will be generated?
Rules of Combination
• A procedure which requires the user to make an
explicit judgment of the effect of combination of two
factors, in all their possible permutations, was called
Rules of Combination by Hopkins (1977). "The rules
assign suitabilities to sets of combinations rather
than to single combinations and are expressed in
terms of verbal logic rather than in terms of numbers
and arithmetic".
Rules of Combination
• The resulting composite assessment can then
become one of the inputs to a new rules matrix
which brings in another factor (or combination of
factors).

You might also like