Guest GIS Contributor Justine Nofal writes about some of the basic types of spatial analysis using GIS.
A GIS (geographic or geospatial information system) allows you to record a base map with a geospatial referencing system such as longitude or latitude and then to add additional layers of other information. Importantly that information is identified using the same geospatial referencing.
The GIS then allows the individual layers, or themes as they are called to be linked. Analysis of the information can then be undertaken using the statistical and analytical tools that are provided as part of the GIS. By providing spatial analysis of suitably coded data it is possible to provide striking, visual representations of data. These representations can often reveal patterns and trends that might otherwise have gone unnoticed without the use of GIS techniques.
|• What is GIS?
• GIS Glossary
• Types of GIS Data
• Types of Error in GIS
• What is Metadata?
Visual representations of this type can also be extended further so that more considered and objective means of measurement are applied. Such approaches allow macro views to be taken in many fields. Epidemiologists are interested in the global spread of disease occurrence; criminologists are interested in local, national and international patters of criminal activity whilst geologists are interested in physical formations to predict earthquake activity and the availability of resources such as oil and gas. None of these disciplines would be able to work as they currently do without the ability to apply analytical tools to otherwise disparate sets of data in the way a Geographical Information System permits.
The Uses of GIS
Summarized below are some of the more common and basic uses of GIS.
The central function of a geographic information system is to provide a visual representation of data. It is estimated that 80% of the data we consider has a geospatial element of some form. GIS provides a means for that data to be stored in a database and then represented visually in a mapped format. Simply understanding where things are is a first step in understanding spatial patterns and relationships.
In the example below, simply mapping out the geographic features helps the viewer to understand where the wells are around a lake.
A proximity analysis is an analytical technique that is used to define the relationship between a specific location and other locations or points that are linked in some way. It is used by many commercial organisations to identify sites suitable for business outlets. The technique will consider different factors such as social and economic demographics and the presence of competitor outlets. For an accurate proximity analysis the various themes to be used must all use the same referencing system otherwise accuracy may suffer.
Proximity analysis can be used to answer several types of questions that include:
- How far is it between point a and point b? The simplest type of proximity analysis calculates distances between two vector points.
- On average how far is one point from a set of other points or conditions?
- What is the closest point in terms of time or cost taken to reach that point?
- What is the straight line distance between a single point and other selected points in that layer?
- How far are the points or edges of the nearest polygon?
A technique called buffering is commonly used with proximity analysis to indicate the sphere of influence of a given point. Buffering involves creating a zone around a given point, line, or polygon (area) of a specified distance. Buffering is useful for creating a zone around a given geographic feature for further analysis using the overlay method. For example, a 1000′ buffer could be generated around a school to then use overlay analysis to find out how many libraries are within 1000′ of that school.
Using multiple algorithms it is possible to select a group of unrelated points on a theme that match a set of criteria. A cluster could include members where distance between them is less than a specific amount or areas where there is density of points greater than a specific level. Typically a GIS will require multiple levels of iteration before the correct algorithms are identified.
Typical clustering models include:
- Connectivity models – the simplest that depend upon simple distance based relationships
- Centroid models – where inclusion in a cluster is determined by identifying the mean value of the cluster that is most appropriate to the point being considered
- Distribution models – where inclusion is determined by the application of a statistical distribution theory such as the normal probability
- Density models – using techniques specially identified for GIS work that link areas with specific densities of an event or population such as racial profiles in a given area
- Subspace models – this technique allows the element to be included into a cluster by considering specific attributes of that element
- Group models – those models where an algorithm cannot be established to demonstrate a shared link where they are in effect linked manually
A technique that can be used to measure the distances between a point and the edge of a specific element that has been defined as a polygon using vector points. Nearest neighbour algorithms have been the subject of intense research since the 1980s and new approaches were defined by academics such as Benezecri and Juan in 1982. The algorithm defined focuses on identifying points that are either maximal, minimal or median members of the data set.
What’s in an Area?
A basic analysis that allows a given area from one later to be overlaid onto data from other themes. A good example would be – what type of soil do we find in the school grounds or what type of industrial uses has this area been put to in the past 50 years.
There are two methods of performing this type of analysis:
- Feature overlay – a simple technique to drop single or multiple points into an area
- Raster overlay – best used when characteristics of multiple themes are required to be examined because each area is referenced and combined on a grid basis
The technique best used to identify a location for a new retail outlet. The technique has been developed from theoretical methods used to explain observed conditions to an algorithm for identifying optimal locations. The algorithms used tend to focus on either maximal, minimal or median members of a given dataset.