There are three types of raster data that can be stored in a geodatabase: raster datasets, raster catalogs, and raster as attributes.

Raster dataset

Raster datasets are single images that are stored in the database. These images may be as simple as a single image imported from a file on disk to a large image that has been created by mosaicking or appending multiple images together into a single, large, and seamless image. Thus, inside ArcSDE, a raster dataset is a table with one row, referencing the raster dataset.

MrSIDs, GRIDs, TIFFs, and ERDAS Imagine files are all examples of raster datasets. There is no restriction on the maximum size of a dataset created in or accessible from any ESRI application.

A raster dataset can have only one color map and occupies only one row in a business table, since it is essentially one seamless image. It can also be used as an attribute of Features, such as snapshot of a home associated with a land parcel polygon or a scanned pump diagram associated with a well point.

A raster dataset should be used when

• Continuous analysis and clipping is to be carried out on one large raster
• Retaining all areas of overlap is not important
• Retaining metadata of each input is not important
• Fast viewing at any scale is desirable

Raster catalog

A raster catalog is a collection of raster datasets. These raster datasets are defined in a table format, where one record represents one raster dataset. A raster catalog is most often used to display adjacent or overlapping raster datasets without having to mosaic them into one large raster dataset.

Each raster dataset in a raster catalog maintains its own properties. For example, one raster dataset might have a different color map than another raster, or one might have a different number of bands than another. Raster Catalogs can accommodate a different color map for each raster dataset, but the projection needs to be the same for each dataset. If geoprocessing tools are used to load raster datasets into raster catalogs, each raster dataset will be automatically reprojected to the projection of the raster catalog.

A raster dataset inside a raster catalog behaves in the same way as a stand-alone raster dataset. Therefore, you can mosaic raster data into a raster dataset that resides in a raster catalog.

A raster catalog should be used when

• Overlapping areas of individual inputs are important
• Metadata of individual inputs is important
• Queries on attributes/metadata are needed (i.e. percent cloud cover)
You simply want to keep/store individual images

See Also: Mosaicked Raster Datasets vs Raster Catalogs

An attribute of type raster

Feature class or stand-alone tables can contain a column of type RASTER. A RASTER type column can contain any supported image, picture, or raster dataset, such as digital photos of a feature or a scanned floor plan. Only one RASTER column can be defined for a particular feature class or table.

When a raster is stored in a RASTER column, it is converted to an ArcSDE Raster format and stored in the geodatabase.