All attributes in our Earth system are not discrete, however, so there is another option with respect to viewing the world: Continuous Field View. For instance, any data set that can be represented as a continuous series qualifies- temperature, gas concentration, elevation would be difficult to model accurately using a discrete object view. For such data sets, a raster data model may be more appropriate. In a raster data model data is represented using a set of evenly distributed and equally sized square grid cells. Each grid cell contains a single value that best approximates the data found at that grid location. Datasets such as land cover, soils, and elevation, all categories of Earth data that are characterized by gradual gradations from one location (grid cell) to the next grid cell are commonly represented using a raster data model.
Remote sensing data such as that collected through satellites orbiting our Earth are often best suited for a raster data model.
The image above shows a set of satellites that are coordinated to collect a variety of data about the Earth's atmosphere and surface.
Spatial Analysis of Raster Data
Map algebra is used to combine datasets together using simple mathematical operators. An example of the application of this spatial analysis approach is a site suitability determination: where useful or non useful locations are identified using numerical criteria. The mathematical formulae can be used to rank a site according to how well it fits predetermined criteria.
Examine the image to the right. Each pixel (picture cell) has a value. When you have two or more raster layers, you can perform map algebra, such as multiplying, dividing, adding and subtracting. You can also change the values within one layer. For instance, if you had a raster layer for elevation with values in feet, you might want to convert the units to meters. All you need to do is multiply by the value of 0.3048 using the raster calculator.