2.nature of Spatial Data - Submitted-1

You might also like

Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 20

CE 413

Introduction to

GEOGRAPHIC INFORMATION SYSTEMS


Dr. Ahmet Çizmeli
METU GGIT
Geodetic and Geographic Information
Technologies
Fall 2008

acizmeli@metu.edu.tr
The nature of spatial data

• Special language for spatial phenomena


• Various ways of defining spatial phenomena
• Importance of spatial scale
• Types of spatial objects / data models
• Measurements
• Information content of spatial data
The nature of spatial data
A spatial language for spatial phenomena
Now that we have a better idea why we need spatial data, we have
to :
Think more clearly about space & spatial phenomena;

Communicate spatial phenomena with others;



One needs to use a language specific to spatial phenomena and
become more efficient in communicating spatial ideas;
A regular use of a common spatial language will eventually change

the way we think and perceive spatial phenomena :


orthophoto continuous raster digital high scale/low resolution
random autocorrelated transect GIScience kriging
DEM cylindrical equidistant feature extraction
WMS/WFS postgreSQL Voronoi contour map SWOT
The nature of spatial data
Spatial phenomena (objects) described
digitally :
Spatial phenomena are the real-world objects and events that

we would like to describe digitally with GIS;

digital spatial data


Spatial phenomena represented in models
GIS using
The nature of spatial data
Spatial data models :
Spatial data models are the computer-level data organization

models that one uses to represent real world objects or


phenomena. There are 4 main types of elementary spatial
data models :

Point (0 dimensional);

Line (1 dimensional);

Raster (2 dimensional);

Surface (3 dimensional);

Others ...

The choice of the right spatial object depends not only on the
spatial phenomena to be described but also on the spatial
scale of the study and the specific needs of the project.
The nature of spatial data
Spatial phenomena (objects) described digitally :

The digital representation of a real-world phenomena is only an
approximation at best;

To make the digital representation of a given spatial
phenomena, one has to choose the appropriate data object at
the optimal spatial resolution;
 There is always a given amount of
information that is lost in the
representation process;
 This loss can be significant if care is
not taken during the GIS design stage;
 A thorough understanding and a
skillful mastering of the data models is
mandatory for a successful GIS study.
The nature of spatial data
Types of spatial phenomena :

There are various majors types of spatial phenomena. One
major way of categorizing them is with respect to their spatial
continuity :

Discrete phenomena;

Continuous phenomena.
Discrete phenomena are easier
to represent. Usually points are
used.

• Continuous phenomena are harder to model. One needs to makes


some assumptions/simplifications.
• Although some alternatives exist, the raster data model is generally
suitable for the representation of continuous phenomena;
The nature of spatial data
Importance of the spatial scale :

The scale has a primary effect on all spatial considerations;

Depending on the
scale, a user may
decide whether to
express a building as a
0-dimensional point or a
couple of raster pixels;

Political boundaries
are abstract (virtual)
concepts that do not
exist in the physical
world; Most of the time,
they are correctly
expressed as lines (no
width).
The nature of spatial data
Importance of A area object can be expressed using a point
object or an area object or else. There is no
the spatial magical formula. Its all about the art of spatial
scale : modeling!
In reality, a road is
a 3-dimensional
surface. It is
however possible to
display it as a 2-
dimensional area
(patch) or as two 1-
dimensional lines or
simply as a 1-
dimensional line.
The choice will
certainly depend on
the scale;
The nature of spatial data
Information content of spatial data :
There are various levels of coexisting information stored in a
given spatial dataset :
 Description of the way objects occupy space (simple location of

a dam, spatial extent of a garden);


 Identification of the objects (what type of tree? what brand of

car?);

The abundance of objects
(what concentration of a
substance in water/air?);

Other attributes (all sorts of
data... age, skin color,
language spoken ... )
The nature of spatial data
Spatial patterns :
Spatial patterns describe the way spatial objects occupy space.
Among the usual patterns we encounter in every day life, one can
count patterns which are :

Regular (systematic);

Random (with no underlying
spatial design);

Clustered;

Oriented;

High-density;
One can also note other patterns on

Dispersed;
the temporal dimension such as static

... and dynamic (steady, exponential)
patterns;
The nature of spatial data
Geographic data
can be obtained by :
 Taking measurements in
the field; Real Time
 Purchasing; Coordinate
Input
 Finding from existing

sources in digital form;


Global Ground
 Capturing from analog
Positioning Surveying
maps by GEOCODING : System (GPS)

Scanning hardcopy
maps; Existing
Digital Data
 Digitizing with a
digitizer;
Topo Property Road Census
Maps Maps Networks Data
The nature of spatial data
Data measurements :
 Data used in GIS needs to be
collected with the objective of serving
a spatial purpose.

A thorough understanding of the
spatial objects and their spatial
attributes (described best with the
geospatial language) is critical even
at the beginning of data collection
stage;

Geospatial data sampling should be
conducted in the correct
measurement/spatial/temporal scale.
Otherwise the quality and the
usability of the data is jeopardized;
The nature of spatial data
Data measurement strategies :
 The nature of the spatial phenomena to be studied should be taken
into account from at the early stages of data collection design;
One can define the following
data schemes :

Data collected on points,
over quadrats, along
transects over a regular grid,
etc.;

Probabilistic sampling;

Random sampling;

Systematic sampling;

Stratified sampling;

Homogeneous sampling;
The nature of spatial data
Measurements can be in :
 Nominal scale, in ordinal scale or as interval/ratio.
The nature of spatial data
Associations between patterns :
 The main role of the GIS is to illustrate, describe and quantify
spatial associations;
 Spatial autocorrelation is one of the most significant spatial

association that one studies using GIS. It helps quantify the


spatial heterogeneity;
 Because of spatial autocorrelation, points (phenomena) located

closer have a tendency to be more similar than points apart;


 Many other associations also do exist : spatial causality

(crime in one place causes more crime in the surrounding


areas), spatial interaction (persons committing crime interact
with their surrounding), spatial co-variance (cities are nearly
always located near water sources) etc....
The nature of spatial data
Spatial autocorrelation :
The nature of spatial data
Associations between patterns :
A GIS user regularly faces the challenge of being obliged to estimate
non-existing measurements using existing values :
 Interpolation is the process of estimating non-existing

measurements within bounding values ;


 Extrapolation is the process of estimating information outside

bounding values;
 The estimation can be 1-2-3-or 4 dimensional (or sometimes more).

When it is two-dimensional, it is called surface fitting;

There are two steps in the process :


 Fitting all existing values to an equation;


Solving the equation for each desired missing values.
Nearer values have more weight on missing values than points further
away : spatial autocorrelation;
The nature of spatial data
Associations between patterns :
 Extrapolation is always
associated with more
uncertainty than
interpolation : spatial
autocorrelation;
 Both procedures are

challenging and non-


exact. One needs to
make assumptions on
the nature of the spatial
data (and choose the
mathematical model
accordingly (linear,
exponential, weighted
etc...);

You might also like