Fundamentals of GIS - University of Vermont

advertisement
------Using GIS-Fundamentals of GIS
Lecture 5:
Introduction to Raster Analysis
By Austin Troy, Brian Voigt and Weiqi Zhou
University of Vermont
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster data-A Refresher
Raster Elements
–Extent
–# rows
–# columns
–Coordinates
–Origin
–Orientation
–Resolution
–Grid cell
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Methods for storing raster data in a more
computationally and memory efficient way.
• Where a raster layer is random noise, this does not
work.
• Requires repetitive patterns or areas of homogeneity.
• The fewer z values, the easier to compress.
• Simplest method is cell-by-cell encoding where cell
values are stored by row and column number; This is
essentially uncompressed.
• DEM’s and satellite images generally use this structure
because there is typically so much variation.
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Run-length encoding (RLE):
– Compression method that records cell values in groups called
“runs.”
– It records the starting and ending pixel for a “run” with the
same value for a given row, so hundreds of pixels could be
recorded with only two values, if they all have the same
value and are adjacent.
– However, because it measures runs along rows, it is not
efficient for two dimensional areas of homogeneity.
– RLE can reduce file size by 10:1, depending on data.
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Runs:
– Row 2: 3,4
– Row 3: 2, 8
– Row 4: 4,7
– Row 5: 5,7
– Row 6: 2,6
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Chain code:
– This is a more efficient method for dealing with
two-dimensional compression
– This defines a homogeneous two-dimensional area
using cardinal directions and units movements to
define bounding perimeter in relative terms from a
known point
– For instance, go 2 N, 1 W, 1N, 3 W, 1S….etc.
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Here, starting from the
lower left, the computer
would define that
coordinate then code 1N,
3E, 1N, 1W, 1N, 2W, 1N,
1E, 1N, 2E etc…..
• This would define the
perimeter of a
homogeneous area.
• All must have exactly the
same value
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Block code:
– A method that uses square blocks to represent areas
of homogeneous values
– Each block is encoded only with location of one
corner cell and the dimensions; since they are
square, only one dimension needs to be given
– Uses medial axis transformation technique
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Quad tree:
– Divides a grid into hierarchy of quadrants
– Starts with four quadrants; any quadrant that has totally
homogeneous cells will not be subdivided further, but is
stored as a “lead node” which is coded only with that value
and the id of the quadrant.
– Any quadrants with more than one value are subdivided
again into four more quadrants and again the computer
checks for homogeneity.
– It keeps on doing this until it has generated all its leaf node or
until it gets down to the pixel level
– This is known as recursive decomposition
– This is good where one part of a grid is very uniform and the
rest is heterogeneous.
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Quad tree:
Homogeneous
(all one value)
Not homogeneous: more
than one value within
quadrant
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Quad tree: now we break down those quadrants
with non-homogeneous values into four sub
quadrants
Not homogeneous: more
than one value within
quadrant
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Quad tree: and we keep doing this until we’ve come
down to the point where there are only homogeneous
quadrants, even if
those are one cell
in dimension
Not homogeneous: more
than one value within
quadrant
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Data Structure
• Quad tree:
One value (leaf node)
Mixed values (non-leaf)
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Converting Features to Rasters
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Converting Vector to Raster
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Raster Overlay Queries
•The raster data model performs overlay operations
more efficiently than the vector model
•Raster cells have a one-to-one relationship between
layers
•Raster overlay queries: combining two or more
thematic layers to identify relationships between
them such as:
–Areas that are common to all layers
–Areas that meet criteria from each layer
Query example:
(“elevation” > 2500) & (“Slope” > 20)
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Raster Overlay Calculations
•Map algebra can be performed to identify
relationships between layers, or to derive
indices that describe phenomena
•Map algebra calculations create a new output
raster
•Example:
(“Soil_depth_1990”) – (“Soil_depth_2000”) = Soil loss 1990 - 2000
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Source: ESRI
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Map Query: Example 1
• Single layer numeric example: lu_chit = 11
(residential)
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Map Query: Example 1
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
Map Query: Example 2
• Single layer numeric example: lu_chit = 11
(residential)
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Map Query: Example 2
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
Map Query: Example 3
• Con Function
• Results in a binary True/False layer
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Map Query Examples
• Multi-criteria, single layer, categorical map query: looking for all
developed land use types, using attribute codes (11, 12, 13) and
the OR logical operator
• Results in a 1/0 binary layer, showing urbanized areas
Vertical
lines mean
OR
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Map Query Examples
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Map Query Examples
One can then convert this to a feature class or shapefile
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Raster Query: 2 Layer Example
Multi-layer queries use criteria across two or more layers; in this
case we’ll query land use (categorical), elevation (number) and
slope (number)
Let’s say we want to
find identify potential
habitat for a rare plant
that grows at higher
elevation, on steeper
slopes and in
coniferous forest
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Raster Query: Slope
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Raster Query: Multiple Criteria
• Multiple criteria, multiple layers
• Land Cover = Coniferous Forest (42)
• Elevation > 800
• Slope > 20%
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Map Calculation
We can also make calculations between layers (or between a layer
and a constant): here we’ll multiply the k factor (soil erodibility
factor) by slope; let’s just imagine this will yield a more accurate
and spatially explicit index of erodibility that factors in slope at
each pixel
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Map Calculation
• Darker areas feature both steep slopes and erodible soils.
• Advantage over map query approach: result is a continuous index
of values, rather than just a “true” / “false” dichotomy
Lecture Materials by Austin Troy except where noted © 2008
------Using GIS-Fundamentals of GIS
Map Calculation and Query
• We could then run a map query to find areas that have high
erodibility factors and urban land use.
Lecture Materials by Austin Troy except where noted © 2008
------Using GIS-Fundamentals of GIS
Zonal Statistics
• Suppose we had a proposed subdivision map (this one is
made up). We could overlay it on our new index to
determine which proposed subdivisions are problematic
(due to soil erodibility).
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Zonal Statistics
• Summarize the mean,
max or sum for some
value within each of the
bounding units
• Polygon and Raster
• Raster and Raster
• Here we summarize by
subdivision zones the
mean soil erodibility
value (from our
calculation).
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Zonal Statistics
Produces a DBF table with the specified summary statistics
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Zonal Statistics
Now we can plot out the subdivision boundaries (zones) by a
soil erosion statistic. In this case we plot subdivision
boundaries shaded by the mean of the soil erosion statistic.
This represent the mean value of all the soil erosion pixels
underlying a polygon
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Reclassifying Raster Data
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Reclassifying Raster Data
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Reclassifying Raster Data
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Neighborhood Statistics
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Low Pass Filter
• Functionality: averaging filter
– Emphasize overall, general trends at the expense of local
variability and detail.
– Smooth the data and remove statistical “noise” or extreme
values.
• Summarizing a neighborhood by mean or median
– The larger the neighborhood, the more you smooth, but the
more processing power it requires.
– A circular neighborhood: rounding the edges of features.
– Resolution of cells stays the same.
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
High Pass Filter
• Functionality: edge enhancement filter
– Emphasize and highlight areas of tonal roughness, or
locations where values change abruptly from cell to cell
– Emphasize local detail at the expense of regional,
generalized trends
• Perform a high pass filter
– Subtracting a low pass filtered layer from the original
– Summarizing a neighborhood by standard deviation
– Using weighted kernel neighborhood
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Neighborhood Statistics
• Min, max, mean,
standard deviation,
range, sum, variety
• Window size/shape
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Low Pass Filter: Example
• Filtering out anomalies in bathymetric data
Bathymetry mass points:
sunken structures
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
• After turning into raster grid
We see sudden
anomaly in grid
Say we wanted to “average”
that anomaly out
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
• Try a low-pass filter of 5 cells
We can still see those anomalies but
they look more “natural” now
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
• Try a low-pass filter of 25 cells
The anomalies have been “smoothed
out” but at a cost
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
• We can also do a local filter in that one area
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
What about high pass filters?
• Say we wanted to isolate where the wreck was
All areas of sudden change, including
our wrecks, have been isolated
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
Low pass filter for elevation
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
A low pass filter of the DEM done by taking the mean values for a
3x3 cell neighborhood: notice it’s hardly different
DEM
Low pass
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
10 unit square neighborhood
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
20 unit square neighborhood
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
In this high-pass filter the mean is subtracted from the original
It represents
all the local
variance that
is left over
after taking
the means for
a 3 meter
square
neighborhood
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
We do this using Spatial Analyst Tools
>>> Math >>> Minus
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
If we do a high-pass filter by subtracting from the original
the means of a
20x 20 cell
neighborhood,
it looks
different
because more
local variance
was “thrown
away” when
Dark areas represent
taking a mean
things like cliffs and
with a larger
steep canyons
neighborhood
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Using standard deviation is a form of high-pass filter because it is
looking at
local variation,
rather than
regional
trends. Here
we use 3x3
square
neighborhood
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
• Note how similar it looks to a slope map because it is showing
standard deviation, or normalized variance, in spot heights,
which is similar to a rate of change.
• Hence it is emphasizing local variability over regional trends.
• The resolution of the slope is quite high because it is sampling
only every nine cells.
• When we go to a larger neighborhood, by definition, the
resulting map is much less detailed because the standard
deviation of a large neighborhood changes little from cell to cell,
since so many of the same cells are shared in the neighborhood
of cell x,y and cell x,y+1.
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Here is the same function with an 8x8 cell neighborhood.
The coarser
resolution (due
to the larger
neighborhood)
makes it so
that slope rates
seem to vary
more gradually
over space
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Later on we’ll look at filters and remote sensing imagery, but here is
a brief example
of a low-pass
filter on an image
that has been
converted to a
grid. This can
help in
classifying land
use types
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Raster Surface Tools
Arc GIS allows you to
use a digital elevation
model (DEM) to derive:
•Hillshade
•Slope
•Contours
•Aspect
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Raster Surface Tools
DEM
+
Hillshade = Hillshaded DEM
+
=
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Display Options
1. Place the hillshade “under” the DEM in the TOC
2. Make the DEM partially transparent
Lecture Materials by Austin Troy except where noted © 2008
------Using GIS-Fundamentals of GIS
Raster Surface Tools
Slope
Contour
Aspect
Lecture Materials by Austin Troy except where noted © 2008
Fundamentals of GIS
Viewshed Analysis
• This is a multi-layer function that analyzes visibility based on terrain.
• It requires a raster terrain layer and a point layer and produces a
visibility layer (raster) that tells you where the feature can be seen
from, or alternately, what areas someone standing at that feature could
see (remember, line of sight is two way).
• If there are more than one point feature, then each grid cell tells you
how many of the point features can be seen from a given point.
• However in that case, you lose information about the other direction;
You don’t know which features can see a particular grid cell.
• Viewshed analysis can use “offsets” to define the height of the viewer
or of the object being viewed; designated using a new field in the input
layer’s attribute table.
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Viewshed Analysis
Let’s say we’re local planners who are considering sites
for a new waste treatment facility in a valley where the
vacation homes of five rich and powerful executives
are located.
We want it in a place that won’t ruin anyone’s views,
since they comprise 95% of the local tax base.
This generates a grid with three values, representing how
many houses can see a given pixel
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Viewshed Analysis
Red represents areas that can be seen by 1 house, blue by 2 or more
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Viewshed Analysis
In order to compare the viewability of several facilities, separate
viewshed analyses need to be done for each feature.
In the next example we will look at three candidate sites for a
communications tower.
Each will produce a viewability grid.
This grid can then be superimposed on a layer showing residential
areas.
Since each grid will belong to a different tower, we can tell which
tower will be most viewable from the residential areas through
simple overlay analysis.
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Viewshed Analysis
In this case, red is for tower 1, blue for 2 and green for 3
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Proximity
• Can use raster distance functions to create zones based on
proximity to features; here, each zone is defined by the
closest stream segment
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Distance Measurement
• Can create
distance grids
from any feature
theme (point, line,
or polygon)
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Distance Measurement
• Can also weight
distance based on
friction factors,
like slope
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Combining Distance and Zonal Stats
• Can also
summarize
distances by
vector
geography
using zonal
stats
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
------Using GIS-Fundamentals of GIS
Combining Distance and Zonal Stats
• Here we
summarize by
the mean
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Density Functions
•We can also use sample points to map out density raster surfaces. This need to
require a z value in each, it can simply be based on the abundance and distribution
of points.
•Pixel value gives the number of points within the designated neighborhood of
each output raster cell, divided by the area of the neighborhood
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Density Functions
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Fundamentals of GIS
Density Functions
Lecture Materials by Austin Troy, Brian Voigt and Weiqi Zhou except where noted © 2011
Download