Lecture 1

Vector and Raster Data Models

Tomislav Sapic
GIS Technologist
Faculty of Natural Resources Management
Lakehead University
 Spatial data can Vector and Raster Data Model
represent real world
features with discrete
boundaries (such as roads,
buildings, lakes, rivers,
administrative
boundaries) as well as real
world phenomena with
non-discrete boundaries
(such as precipitation and
nutrient levels, terrain).

In general, the vector

data model is used to
represent the former and
the raster data model to
represent the latter.

However, the lines od

separation are not firm. As
well, often making of a GIS
file in one model requires
use of files from the other
model. Source: Lo et al.
(2007)
Raster vs. Vector
Data Model
Representations
of the Real
World
Although discrete objects
such as buildings, ski lifts,
roads, are usually not
represented by a raster data
model, they are handy for
illustrating a direct
comparison of representation
between a raster and a vector
data model.
Elevation, as a continuous land
(field) characteristic, is usually
stored in a vector file for easier
data management, but is more
suitable to be represented by a
raster data model.
Source: Heywood et al. (2006)
 Vector Data Model
Spatial features are usually stored in separate theme GIS files.

Roads are typically

stored in a separate line
vector GIS file, called,
e.g., roads
A small pond might be
stored as part of a
Rivers would typically wetlands GIS file.
be stored in a separate
line vector GIS files,
called, e.g., rivers or
streams. A forest area (stand) is
usually delineated and
stored in a separate
Lakes would typically be poly GIS file, often
stored in a polygon called fri.
(poly) vector GIS file,
called, e.g., lakes.
 Vector Data Model
The three main geometric shapes used in the vector data model, to
represent real world features are:
Point
Line
Points, lines and polygons are Polygon Line
stored in separate, point, line, Polygon Point
and polygon, GIS files.

Each GIS file along with spatial

features also contains a table in
which each row (record)
represents one of the spatial
features.

(An exception to the above are

so called multipart features,
where one row represents
multiple features.)

Columns (fields) in the table

are used to store data
(attributes) describing each
feature.
 Point GIS File

The shapefile represented by this

layer contains point features
representing municipalities.

A feature is represented by a row

The Shape column (field) contains (record) in the files table.
the spatial information, i.e., x and y
coordinates for each point.
 Line GIS File In ArcGIS, vertices are not
usually visible but can be
Line features are made of two or more vertices (sing. vertex).
made visible during the
A vertex is a point georeferenced by an x,y pair.
editing of the line.
Each record (row) in the table represents a whole line (all vertices
composing the line), which is often just a segment of a longer line.

End vertices are called

nodes.
 Polygon GIS File
Each polygon is
represented by one
Polygon features are made of one or record in the files
more lines that enclose(s) an area. table.

Since lines are

made up of
vertices that means
that polygons are,
in the end, made
up of vertices
(points), i.e. they
are represented by
a group of x.y
coordinates.
 Circles, as well, are
actually a series of
vertices, i.e. a series of
straight lines (not
consisting of an infinite
number of points!).

This is changing as well, with GIS software

now giving an option to connect vertices with
parameterized segments, such as circular arcs,
elliptical arcs, and Bzier curve.
 x,y Pairs as Spatial Representation

Ultimately, GIS vector files, point, line

or polygon files, have their spatial
representation stored in the
computer memory through x,y
coordinate pairs.
There are no lines stored the GIS
software recognizes the type of the
file (a point, line, or a polygon GIS
file) and displays the features as
points only or connects the points
(vertices) with lines and displays
them as lines or fills the line-enclosed
areas with a colour and displays the
features as polygons.

Source: Lo et al.
Separate GIS (2007)
files.
 Singleparts vs.Multiparts
Singlepart
One feature is represented
by one record.

Multipart
Multiple features are represented by one
record.
Applications for multipart features are for
example, multiple polygons representing
protective buffers around the same type of
features (e.g. eagle nests).
A downside of a multipart: only one
attribute is assigned to all multipart parts,
which, e.g. for areas, can be deceiving when
looking at individual parts.
 Vector Data Model Attributes

In GIS vector files attributes are, in a simple form, stored in tables (databases).

A table consists of records (rows) representing individual features, fields (columns)

representing a particular theme describing the feature, and attributes an intersection between
a record and a field.

In ArcGIS, FID and Shape fields, although shown in the Table of Attributes, are not actually part
of the attributes, but rather represent the spatial and index information (e.g. the .shp and .shx
content in shapefiles). Because of it FID and Shape fields cannot be deleted from the table,
unlike any other pure attribute field.

Field
Record
Attribute
 Typical Storages of Vector Data
All here presented happen to be created by ESRI.
Vector File Description
Format

- first vector file format developed by ESRI

- it was the main vector file format of, at the time, major ESRI software, ArcINFO
Coverage - composite
- has inherent topology
- discontinued but can still be found in older GIS databases
- developed as a simpler file format than coverages, at first meant only for
converting coverages into it and viewing it in a more user friendly GIS software
called ArcView
Shapefile - composite
- become wide spread and adopted by all newly developed GIS software,
including open source
- often less stable than coverages, doesn`t have inherent topology
Feature Class - successor to shapefile
(not a true file but - composite but residing in a geodatabase, in the form of bit tables
a geodatbase - can only exist within a geodatabase
component) - as part of a geodatabase can have, e.g., table joins and topology defined.
 Shapefile Vector File Format Possible composite file extensions:
*.dbf dBase table (database) file,
Shapefiles are vector composite files, made up
containing attributes.
of 3-13 separate files.
* Basic *.shp the file that stores the
In Windows Explorer all shapefile components shapefile feature geometry, i.e., x,y
are shown, in ArcCatalog entire shapefile is elements (a coordinates.
shown as one item. shapefile is *.shx the file that stores the
defunct index connecting .dbf and .shp
without files.
any of .prj projection file.
Shapefile them).
.shp.xml metadata file.
.sbn, .sbx spatial
index files sometimes present.
.ain and .aih attribute index file.
All components of a shapefile need to be present together
(important when transferring files!), otherwise the shapefile can be .atx new, ArcGIS,atribute index
defunct or incomplete. file.

The projection file is a beneficial addition to a shapefile, although .lxs, .mxs geocoding index files.
not a necessary oneshapefiles can be used without a projection file .cpg specifies character set code
but this is to be avoided. page.
All elements have the same filename (e.g., roads).
Shapefiles can be either a point or a line (arc) or a polygon file -- they
cannot contain more than one shape type!
 Raster Data Model
Raster data model is represented by square cells of same size organized in horizontal rows
and vertical columns (rows and columns are never slanted).
A rasters position (geographic location) is defined by
assigning x,y coordinates to one or two corners -- positions
of the cells themselves flow from this reference point and Spatial resolution of a raster is
the overall Cartesian coordinate system structure. defined by the cell size (1 m ,
20 m , 100 m, ).

Nominal or ordinal values can be used

Each cell contains a value. Values are in the raster when something is not
always numeric and can be either integer measured but classified into categories
or decimal numbers (or nodata in grids). (themes).
 An ArcMap layer representing a raster that
A raster dataset representing terrain. contains elevation values. The raster itself
doesnt contain colours colours are assigned
to raster cells by software for interpretation
purposes and are based on cell vales.

The raster converted into a simple text raster format,

ASCII.
 Rasters Spatial Resolution
Expresses the size of the earth surface represented
by one (square) cell.
A 10 m raster is a raster whose each cell represents
10 x 10 meters of the earth surface.
Because of the nature of rasters and their easy
computational resampling, raster spatial resolution
does not necessarily represent the spatial
variability on the ground.
The measure of represented spatial variability is
expressed with Ground Sampling Distance.
For example, the 80 m raster in the example on the
right can be taken and resampled into a 40 m
raster, but in that case all four new 40 m cells
within each original 80 m cell area would have the
same cell value the spatial resolution would
become 40 m, but the Ground Sampling Distance
would stay 80 m.
Chosen coarseness of
raster resolution has
implications on the feature
representation.
 Raster Types Based on Cell Values

In terms of the cell values, generally, rasters can be divided into image,
interpolated, and thematic rasters.
Image rasters have values that represent measurements of energy
(reflections) captured by a sensor (camera).
Image raster cell values are usually whole numbers (integers).
 Image raster files are often multiband

~ 610 660 nm Values representing

different electromagnetic
spectrum segments (i.e.,
~ 530 590 nm
primary colours, NIR, etc.),
are stored in separate bands
~ 430 490 nm (channels) spectral
resolution!

Source: ESRI, ArcGIS

Help File, 2011.

One byte (0-255) per pixel, per channel , if an 8 bit image;

three bytes per pixel (0-16,777,215) per channel , if a 24 bit image, etc.
 Raster Types Based on Cell Values
Thematic rasters have cell values that represent a theme, category, class. Usually these
are whole, integer numbers.
Cell values in interpolated rasters represent interpolated measurements such as terrain
elevations, precipitation levels, soil pH levels, etc. Usually the cell values are decimal
numbers (floats), although they can be integers as well.

Cell value 1 = steep slope

Cell value 100 = not steep slope

Thematic and interpolated rasters can be multi-band but are predominantly one-band.
 Raster Types Based on Cell Values

Image raster.

Interpolated raster. Classification Thematic

into classes
raster.
Regular pattern of sizes and distances Relation Analyses Through Overlapping
in the raster data model means that
topology is inherent to it, allowing for
a range of spatial, topological
analyses.

Neighbourhood Analyses

Cost-Path Analyses
 Raster Dataset Properties
Format: the type of file used to store the raster (for example, .jpg, .tiff).

Number of Bands: the number of layers in the raster file, representing the same area but
through different values.

Data Type: the type of data assigned to raster cells; integer (whole number), float (decimal
number), etc. The data type is assigned to the entire raster, all cells have the same data type
there cant be one cell having an integer value and the next cell a float number value.

Data Depth: also known as pixel or bit depth the per-cell binary range assigned to the
raster; e.g., an 8-bit integer raster can have integers (whole numbers) from 0 255 (8 bits =
255) assigned to a cell, only.

Statistics: include minimum and maximum value, mean, and standard deviation of all
values assigned to raster cells.

Extents: left, right, top and bottom coordinates of the raster dataset.

Projection: rasters coordinate system (map projection).

Size of the Raster: the number of rows and columns in the raster.
 Raster Data Model
In the raster data model, attributes are limited to the numeric values of the cells themselves,
and while it is possible to link additional attributes to the groups of cells having same values,
this is rarely done in practice for the reasons of low utilization value and cumbersome data
management.
Raster data models often take more memory space than the spatial component of the vector
data models (attributes attached to the vector data models can tip the balance the other way).
Rasters can also have NoData values, which indicates absence of data (not the same as
zero!).

Typical Raster Data Formats in ArcGIS

GRID a raster data format used by ESRI platforms; can contain both integers and floats;
usually one-band but can be stacked into so called grid stacks.
.img a raster file format created for ERDAS, a remote sensing software.; can contain both
integers (up to 32 bit) and floats; can be both one-band or multiband (including > 3 bands).
.tiff - can contain both integers (up to 32 bit) and floats; can be both one-band or multiband
(including > 3 bands).
.jpg - cannot contain > 8-bit (unsigned) values and cannot have > 3 bands.
ASCII uncompressed, simple raster data format, used to transfer information.
 Vector and Raster Data Models and Their Advantages

Vector Data Model Raster Data Model

Suitable for: Suitable for:
Recording discrete features with Representing continuous spatial
definable boundaries. features.
Attaching both textual and numerical Containing images (aerial, satellite) .
attributes to spatial features, and
Analyzing relations between
managing and analyzing these attributes
continuous phenomena taking place on
and through them the features
the same geography.
themselves.
Analyzing spatial relations between
Editing, update, and management of
features (cost-path, density,
the spatial features.
interspersion, edge contact, etc.).
Representing natural shapes.
Neighbourhood analyses.

In many respects, vector and raster data models complement each other and
get switched between along the processes of data collection, observation,
analyses, and presentation.
References:

Heywood, I., S. Cornelius, and S. Carver. 2006. Introduction to Geographical Information

Systems. Pearson Prentice Hall, New York.
Lo, C. P. and A. K. W. Yeung. 2007. Concepts and Techniques of Geographic Information
Systems. Pearson Education Inc. 532 pp.