GIS Data Model

Raster Data model

1
2
(Raster Data Model) A. THE DATA MODEL
geographical variation in the real world is infinitely complex
the closer you look, the more detail you see, almost without limit it would take an
infinitely large database to capture the real world precisely
data must somehow be reduced to a finite and manageable quantity by a process of
generalization or abstraction
geographical variation must be represented in terms of discrete elements or objects
the rules used to convert real geographical variation into discrete objects is the data
model
Tsichritzis and Lochovsky (1977) define a data model as "a set of guidelines for the
representation of the logical organization of the data in a database... (consisting) of
named logical units of data and the relationships between them.
current GISs differ according the way in which they organize reality through the data
model
each model tends to fit certain types of data and applications better than others
the data model chosen for a particular project or application is also influenced by :
- the software available
-the training of the key individuals historical precedent





3
there are two major choices of data model - raster and vector
raster model divides the entire study area into a regular grid of
cells in specific sequence
the conventional sequence is row by row from the top left
corner
each cell contains a single value .
is space-filling since every location in the study area
corresponds to a cell in the raster
one set of cells and associated values is a layer
there may be many layers in a database, e.g. soil type,
elevation, land use, land cover
vector model uses discrete line segments or points to identify
locations
discrete objects (boundaries, streams, cities) are formed
by connecting line segments
vector objects do not necessarily fill space, not all
locations in space need to be referenced in the model
a raster model tells what occurs everywhere - at each
place in the area

a vector model tells where everything occurs - gives a
location to every object
conceptually, the raster models are the simplest of the
available data models
4
B. CREATING A RASTER
consider laying a grid over a geologic map
create a raster by coding each cell with a value that
represents the rock type which appears in the
majority of that cells areas
when finished, every cell will have a coded value
in most cases the values that are to be assigned to
each cell in the raster are written into a file, often coded
in ASCII
this file can be created manually by using a word
processor, database or spreadsheet program or it
can be created automatically
then it is normally imported into the GIS so that the
program can reformat the data for its specific
processing needs
there are several methods for creating raster databases
5
B. CREATING A RASTER

Cell by cell entry
direct entry of each layer cell by cell is simplest
entry may be done within the GIS or into an ASCII file for importing
each program will have specific requirements
the process is normally tedious and time-consuming
layer can contain millions of cells
average Landsat image is around 7.4 x 106 pixels, average TM scene is about
34.9 x 106 pixels
6
run length encoding can be more efficient
values often occur in runs across several cells
this is a form of spatial autocorrelation - tendency for nearby things to be more
similar than distant things
data entered as pairs, first run length, then value
e.g. the array 0 0 0 1 1 0 0 1 1 1 0 0 1 1 1 0 1 1 1 1 would be entered as 3 0
2 1 2 0 3 1 2 0 3 1 1 0 4 1
this is 16 items to enter, instead of 20
in this case the saving is 20%, but much higher savings occur in practice
imagine a database of 10,000,000 cells and a layer which records the county
containing each pixel
suppose there are only two counties in the area covered by the database
each cell can have one of only two values so the runs will be very long
only some GISs have the capability to use run length encoded files

Digital data
much raster data is already in digital form, as images, etc
however, resampling will likely be needed in order that pixels coincide in each
layer
because remote sensing generates images, it is easier to interface with a raster GIS
than any other type
elevation data is commonly available in digital raster form from agencies such as the
US Geological Survey
7
C. CELL VALUES
Types of values
the type of values contained in cells in a raster
depend upon both the reality being coded and
the GIS different systems allow different classes
of values, including :
whole numbers (integers) real (decimal) values
alphabetic values
many systems only allow integers, others which
allow different types restrict each separate raster
layer to a single kind of value
if systems allow several types of values, e.g.
some layers numeric, some non-numeric, they
should warn the user against doing unreasonable
operations
e.g. it is unreasonable to try to multiply the
values in a numeric layer with the values in a
non- numeric layer
integer values often act as code numbers, which
"point" to names in an associated table or legend
e.g. the first example might have the following
legend identifying the name of each soil class : 0
" =no class" 1 = "fine sandy loam" 2 = "coarse
sand" 3 = "gravel "
8
One value per cell
each pixel or cell is assumed to have only one value this is often inaccurate - the
boundary of two soil types may run across the middle of a pixel
in such cases the pixel is given the value of the largest fraction of the cell, or the
value of the middle point in the cell
note, however, a few systems allow a pixel to have multiple values
the NARIS system developed at the University of Illinois in the 1970s allowed
each pixel to have any number of values and associated percentages
e.g. 30% a, 30% b, 40% c
D. MAP LAYERS
the data for an area can be visualized as a set of maps of layers
a map layer is a set of data describing a single characteristic for each location
within a bounded geographic area
only one item of information is available for each location within a single layer -
multiple items of information require multiple layers
on the other hand, a topographic map can show multiple items of information for
each location, within limits
e.g. elevation (contours), counties (boundaries), roads, railroads, urbanized
areas
these would be 5 layers in a raster GIS
typical raster databases contain up to a hundred layers
each layer (matrix, lattice, raster, array) typically contains hundreds or thousands
of cells
important characteristics of a layer are its resolution, orientation and zone

9
















10
Resolution
in general, resolution can be defined as the minimum linear dimension of the smallest
unit of geographic space for which data are recorded
in the raster model the smallest units are generally rectangular (occasionally systems
have used hexagons or triangles )
these smallest units are known as cells, pixels
note: high resolution refers to rasters with small cell dimensions
high resolution means lots of detail, lots of cells, large rasters, small cells
Orientation
the angle between true north and the direction defined by the columns of the raster
Zones
each zone of a map layer is a set of contiguous locations that exhibit the same value
these might be :
ownership parcels
political units such as counties or nations
lakes or islands
individual patches of the same soil or vegetation type
there is considerable confusion over terms here
other terms commonly used for this concept are patch, region, polygon
each of these terms, however, have different meanings to individual users and
different definitions in specific GIS packages
in addition, there is a need for a second term which refers to all individual zones
that
11
have the same characteristics
class is often used for this concept
note that not all map layers will have zones, cell contents may vary continuously
over the region making every cell's value unique
e.g. satellite sensors record a separate value for reflection from each cell
major components of a zone are its value and location(s )
Value
is the item of information stored in a layer for each pixel or cell
cells in the same zone have the same value
Location
generally location is identified by an ordered pair of coordinates (row and column
numbers) that unambiguously identify the location of each unit of geographic space
in the raster (cell, pixel, grid cell )
usually the true geographic location of one or more of the corners of the raster is
also known
E. EXAMPLE ANALYSIS USING A RASTER GIS
Objective
identify areas suitable for logging
an area is suitable if it satisfies the following criteria :
is Jackpine (Black Spruce are not valuable )
is well drained (poorly drained and waterlogged terrain cannot support
equipment, logging causes unacceptable environmental damage )
is not within 500 m of a lake or watercourse (erosion may cause deterioration of
water quality )