GNR402: Introduction to Geographic Information

Systems: RASTER ANALYSIS

Surya S. Durbha, PhD

Professor
CSRE, IITB

 Outline
❖ Background
❖ Local Operations
❖ Neighborhood Operations
❖ Zonal Operations
❖ Physical Distance Measure Operations
❖ Comparison of Vector and Raster Data
Analysis

RASTER OPERATIONS
 RASTER OPERATIONS
❑    A raster GIS must have capabilities for Input of data, various
housekeeping functions, operations on layers, output of data and
display  layers

❑ Basic Display

❑    Simplest type of values to display are integers

❑ On a colour display , each integer value is assigned a unique
colour (Fig.1)
❑     If the values have a natural order, a sequence of  colors is used
(Fig.2) 

      There must be a legend to explain the meaning of each color.

 RASTER OPERATIONS

 Other Types of Display: 

❑ Display the data as a surface

❑ Contours can be shown through the pixels along the
lines of constant value (Fig.3)
❑ The surface can be shown in oblique-perspective
view (Fig.4)
 Fig. 1
Fig. 2

Fig. 3 Fig. 4
 LOCAL OPERATIONS
❑   produce a new layer from one or more input layers
❑    the value of each new pixel is defined by the
values of the same pixel on the input layers(s)
❑ neighbouring or distant pixels have no effect
❑ Arithmetic operations make no sense unless the
values have appropriate scales of measurement
 Regrouping  
❑     Is carried out using only one input layer
1. assign a new value to each unique value on the
input layer
➢ useful when the number of unique input values is
small 
 LOCAL OPERATIONS
2. assign new values by assigning pixels to classes or ranges
based on their old values
➢ useful when the old layer has different values in each
cell,
e.g., elevation or satellite images
3. sort the unique values found on the input layer and replace by
the rank of the value

➢ e.g. 0, 1, 4, 6 on input layer become 1, 2, 3, 4 respectively

➢ applications : assigning ranks to computed scores of
capability, suitability etc.
➢ some systems allow a full range of mathematical operations
 Overlay Operations
➢ an overlay occurs when the output value
depends on two or more input layers
➢ many systems restrict overlay to two input
layers only (Fig. 5)
Examples :
1. output value equals arithmetic average of
input values
2. output value equals the greatest (or least) of
the input values
3. layers can be combined using arithmetic
operations
4. combination using logical conditions
e.g. if y > 0, then z = 1, otherwise z = 0
Fig. 5
 Overlay Operations
5. assign a new value to every unique combination of input
values by using cross tables

1 2 3 4
1 1 1 2 2
2 1 1 1 2
3 2 1 1 2
 Applications of Local Operations
Change Detection Study
❑ Use unique combinations produced by the ‘combine’
operator to trace the change of the cell value (e.g. change
of vegetation cover).
❑ Most useful for GIS models that require mathematical
computation on a cell-by-cell basis.
http://libraries.maine.edu/Spatial/gisweb/spatdb/gis-lis/pix/gi94105i01.jpeg
http://www.earthzine.org/wp-content/uploads/2011/07/Figure-11.jpg
http://www.innovativegis.com/basis/mapanalysis/topic22/Topic22_files/image005.png
 Applications of Local Operations

Soils:
❑ Revised Universal Soil Loss Equation (RUSLE) uses six
environmental factors in the equation:
A=R K L S C P
A=average soil loss
R=rainfall-runoff erosivity factor
K=soil erodibility factor http://www.scotland.gov.uk/Resource/Img/47121/0020519.jpg

L=slope length factor

S=slope steepness factor
C=crop management factor
P=support practice factor
❑ Each factor can be prepared as as an input raster.
o Multiply the rasters in a local operation to produce the output raster of
average soil loss.
Applications of Local Operations
 NEIGHBOURHOOD OPERATIONS
   The value of a pixel on the new layer is determined
by the local neighborhood of the pixel on the old
layer
Filtering
       A filter operates by moving a "window" across the
entire raster
e.g. many windows are 3x3 cells
➢    the new value for the cell at the middle of the
window is a weighted average of the values in the
window
➢ by changing the weights we can produce
different effects:
Examples filters: NEIGHBORHOOD OPERATIONS
1.
.11 .11 .11
.11 .11 .11
.11 .11 .11
➢ Replaces each value by the simple unweighted average of it and its eight
neighboring values
➢ smooth's the spatial variation on the layer
2.
0.5 0.5 0.5
0.5 0.6 0.5
0.5 0.5 0.5
➢ slightly smooth's the layer
3.
-.1 -.1 -.1
-.1 1.8 -.1
-.1 -.1 -.1
➢   slightly enhances local details by giving neighbors negative weights
 NEIGHBORHOOD OPERATIONS
❑ The surrounding cells are chosen for their distance and /or
directional relationship to the focal cell.
❑ Common neighborhoods include rectangles, circles, wedges,
etc.
❑ Output from a neighborhood operation can show summary
statistics including maximum, minimum, range, sum, mean,
median, and standard deviation
 APPLICATIONS OF NEIGHBORHOOD OPERATIONS
❑ Important application of neighborhood operations is data
simplification
❑ Moving average method reduces the level of cell value
fluctuation in the input raster.
❑ The output raster of moving averages represents a
generalization of the original cell values.
Variety of vegetation types:
❑ Use variety as a measure
o Tabulate the number of different cell values in the neighborhood, and assigns
the number to the focal cell.
o Wildlife species another example
❑ Neighborhood operations are common in Image processing
o Filtering, convolution, or moving window.
 APPLICATIONS OF NEIGHBORHOOD OPERATIONS
Terrain Analysis:
❑ Slope aspect, and surface curvature measures of a
cell are all derived from neighborhood operations
using elevations from its adjacent neighbors.
Gravity Sprinkler System:
❑ Installing a gravity sprinkler irrigation system
requires information about elevation drop within a
circular neighborhood of a cell.
Site selection studies:
❑ Due to the ability to summarize statistics within a
defined area, a neighborhood operation can be used
to select sites that meet a study’s specific criteria.
 Slopes and aspects

➢  if the values in a layer are elevations, we can compute the

steepness of slopes by looking at the difference between a
pixel's value and those of its adjacent neighbors
➢ the direction of steepest slope, or the direction in which the
surface is locally "facing", is called its "aspect“ (Fig. 6)
➢ slope and aspect are useful in analyzing
vegetation patterns, computing energy balances
and modeling erosion or runoff
➢   aspect determines a direction of runoff
➢ it can be used to sketch drainage paths for
runoff
DEM Slope Aspect

Fig. 6
 Raster Analysis
They are particularly beneficial in the
construction of environmental models,
such as the
▪ analysis of soil erosion,
▪ forestry,
▪ hydrological applications.

 OPERATIONS ON EXTENDED NEIGHBOURHOODS

Distance
➢ calculate the distance of each cell from a cell or the nearest of
several cells
➢ each pixel's value in the new layer is its distance from the
given cell(s)
Buffer zones
➢ buffers around objects and features are very useful GIS
capabilities
➢   e.g. build a buffer of 500 m wide around the road network
➢ buffer operations can be visualized as spreading the
object spatially by a given distance (Fig. 7)
Fig. 7
 OPERATIONS ON EXTENDED NEIGHBORHOODS

➢ the result could be a layer with values :

➢ 1 If in original selected object
➢ 2 If in buffer
➢ 3 If outside object and buffer
➢ Application include noise buffers around roads, safety
buffers around hazardous facilities
➢ In many programs the buffer operation requires the user
to first do a distance operation, then a reclassification of
the distance layer
➢ the rate of spreading may be modified by another layer
representing "friction"
➢ the friction layer could represent varying cost of travel.
This will affect the width of the buffer - narrow in areas of
high friction, etc.

 
 OPERATIONS ON EXTENDED NEIGHBORHOODS

Visible area or "viewshed“

➢ given a layer of elevation, and one or more viewpoints, compute the

area visible from at least one viewpoint (Fig. 8)
      e.g. value = 1 if visible, 0 if not
➢ useful for planning location of facilities such as smokestacks, or
surveillance facilities such as fire towers, or transmission facilities

 
Fig. 8
 OPERATIONS ON ZONES (GROUPS OF
PIXELS)
Identifying zones
➢ by comparing adjacent pixels, identify all patches or zones having the
same value
➢ give each such patch or zone a unique number
➢ set each pixel's value to the number of its patch or zone

Areas of zones
➢ measure the area of each zone and assign this value to each pixel
instead of the zone’s number
➢ alternatively output may be in the form of a summary table

Perimeter of Zones
➢ measure the perimeter of each zone and assign this value to each pixel
instead of the zone’s number
➢ alternatively output may be in the form of a summary table
➢ length of perimeter is determined by summing the number of exterior
cell edges in each zone
 OPERATIONS ON ZONES (GROUPS OF PIXELS)

Zonal operations compute a new value for each location of the

existing values from a specified layer that are associated not
(just) with the location itself but with all locations that occur
within its zone on another specified layer (Fig. 9)

OPERATIONS ON ZONES (GROUPS OF PIXELS)

http://www.innovativegis.com/basis/mapanalysis/topic22/Topic22_files/
image029.gif
OPERATIONS ON ZONES (GROUPS OF PIXELS)

http://webhelp.esri.com/arcgiSDEsktop/9.3/published_images/
SAc_Zonal_Stats_input_zone_and_value.gif
OPERATIONS ON ZONES (GROUPS OF PIXELS)

http://www.innovativegis.com/basis/mapanalysis/topic22/Topic22_files/
image011.png
 OPERATIONS ON ZONES (GROUPS OF PIXELS)

Comparing the topographical characteristics of different soil textures:

• Use soil raster that contains the categories of sand, loam, and clay
as the zonal raster and slope, aspect, and elevation as the input
rasters.
• By running a series of zonal operations, we can summarize the
slope, aspect and elevation characteristics associated with the three
soil textures.

Fig. 9
 OPERATIONS ON ZONES (GROUPS OF PIXELS)

Distance from zone boundary

➢ measure the distance from each pixel to the nearest part of its
zone boundary, and assign this value to the pixel
➢ boundary is defined as the pixels which are adjacent to pixels of
different values
Shape of zone
➢ measure the shape of the zone and assign this to each pixel in the
zone
➢ one of the most common ways to measure shape is by comparing
the perimeter length of a zone to the square root of its area
 COMMANDS TO DESCRIBE CONTENTS OF
LAYERS
One layer
➢ generate statistics on a layer
➢ e.g. mean, median, most common value, other statistics

More than one layer

➢ compare two maps statistically
➢ e.g. is pattern on one map related to pattern on the other?
➢ e.g. chi-square test, regression, analysis of variance

Zones on one layer

➢ generate statistics for the zones on a layer
➢ e.g. largest, smallest, number mean area
EXAMPLE ANALYSIS USING A RASTER GIS
EXAMPLE ANALYSIS USING A RASTER GIS
 EXAMPLE ANALYSIS USING A RASTER GIS

Objective
Identify erosion prone areas :
An area that satisfies the following criteria:
➢ has high intensity rainfall
➢ has with less soil depth
➢ has less vegetation
➢ has steeper slope
Raster description :
Resolution 100 m, area 0.5 km by 0.5 km
 Layer 1 : Slope Map 1 steeper slope 2 lower slope
1 1 1 2 2
1 1 1 2 2
1 1 2 2 2
1 2 2 2 2
1 2 2 2 2
 EXAMPLE ANALYSIS USING A RASTER GIS
Layer 2 : Soil Depth Map 1 low 2 high
1 1 1 1 2
1 1 1 1 2
1 1 1 2 2
2 2 2 2 2
2 2 2 2 2
Layer 3 : Vegetation Map 1 less 2 more
1 1 1 2 2
1 1 1 2 2
1 1 2 2 2
1 2 2 2 2
1 2 2 2 2
Layer 4 : Rain Fall Map 1 high 2 low
1 1 1 2 2
1 1 2 2 2
1 1 1 2 2
1 1 1 2 2
1 2 2 2 2
 ANALYSIS STEPS

Layer 1: Layer 2: Layer 3: Layer 4:

Slope Soil Depth Vegetation Rain Fall

Overlay Overlay

Layer 5: Layer 6:
Steeper Slope & Low Less Vegetation
Soil Depth & High Rainfall

Overlay

Layer 7:Erosion Prone Areas

 ANALYSIS STEPS
 Layer 5 : Steep Slope and low soil depth
1 1 1 0 0
1 1 1 0 0
1 1 0 0 0
0 0 0 0 0
0 0 0 0 0
Layer 6 : Low Vegetation and High Rain Fall
1 1 1 0 0
1 1 0 0 0
1 1 0 0 0
1 0 0 0 0
1 0 0 0 0
Layer 7 : Erosion Prone Areas
1 1 1 0 0
1 1 0 0 0
1 1 0 0 0
0 0 0 0 0
0 0 0 0 0
 WEIGHTED OVERLAY ANALYSIS

http://webhelp.esri.com/arcgiSDEsktop/9.3/published_images/
SAc_Overlay_Example_inputs.gif
location for a new urban park
 Best
location to
construct a
new school

Forest and agriculture were more favorable than residential housing in this model. It
was desired to locate the school on flat slopes, near recreation sites, and far from
existing schools.http://webhelp.esri.com/arcgiSDEsktop/9.3/index.cfm?TopicName=Overlay_analysis_sample_applications
‘Much of the life of the mind consists in applying concepts to things’
(Fodor 1998:24)