Geographic space and location can be an important component of data sampling. Traditional data sampling methods do not cater well for phenomena in geographic space. Spatial sampling methods do consider the geographic location and distribution of the population, and this is reflected in the sampling strategy used to identify samples representative of the the population.

So what is the main difference between traditional and spatial data sampling? The main difference is that geographic location has a strong influence in the choice of the sample. In other words, it is important WHERE the sample is located.

Note that the "geographic" population (ie. spatial universe) to be sampled can be one of two types:

1. Discrete - comprises discrete phenomenon such as households, trees, mine sites, or roads, and
2. Continuous - a continuous region or phenomenon such as a park, soils, rainfall, or vegetation.

Both discrete and continuous space can be sampled. For example, we can choose point samples of (discrete) trees or point samples of (continuous) vegetation cover. A number of questions must be considered in the choice of a spatial sampling method:

• Is there sufficient coverage of the geographic area in question? In other words, are all locations sampled?
  
• Is there sufficient coverage of the relevant phenomena (discrete or continuous) within the study region? In other words, is the phenomena sufficiently sampled?

Spatial data sample types

The spatial data sample types can be:

• point samples, eg: soil samples, drill hole samples, dieback-infested tree samples
• areal samples, eg: local government area samples, vegetation coverage samples
• linear samples, eg: traffic flow along road, pollution content in a stream, etc.

To think about...

How do you sample a continuous population such as "rainfall"? Presumably you would use some form of point sampling.

Point sampling is a "discrete" form of sampling (ie. you get "discrete" results). Is is possible to have a "continuous" form of sampling?