A transformation in a raster dataset defines how the pixels will be transformed when displayed or accessed as well as the output spatial reference of the raster dataset after the transformation is applied (rectified). Therefore, a raster dataset that contains a native extent and native spatial reference before the transformation is applied will have a different extent and spatial reference after the transformation. ArcGIS supports many geodata transforms (used to apply a transformation), including rational polynomial coefficient (RPC), frame camera, polynomial, spline, and so on. A geodata transform is the mathematical model that geometrically transforms the pixels.
Normally, you would permanently alter the spatial referencing information of a raster dataset by applying a transformation, which also alters the orientation of the pixels. However, there are some cases where you want to keep your data in its original condition and just display it using a new transformation.
To see the raster information before and after the transformation, view the Raster Dataset Properties dialog box.
A transformation can be stored with a raster dataset as a property of the dataset and will be used to transform the pixels on the fly. A geodatabase raster dataset stores the transformation information as a persisted stream internally with the raster dataset, and a file-based raster stores the transformation in an external .aux.xml file. When you georeference an image using the Georeferencing toolbar, the Rectify operation will apply the transformation to the input raster data and save it to a new raster dataset, while Update Georeferencing will write the transformation as a property of the raster dataset and persist it with the dataset. The actual pixels in the input raster dataset do not change when the georeferencing is updated.
Transformation and spatial reference of raster datasets
A transformation in a raster dataset defines how the pixels will be transformed during display as well as the output spatial reference of the raster dataset after the transformation has been applied. Therefore, a raster dataset that contains a native extent and native spatial reference before the transformation is applied will also have an extent and spatial reference after the transformation. You will be able to view the updated extent and spatial reference information on the Raster Dataset Properties dialog box.
Managing the persisted transformation with raster datasets
ArcGIS will manage the persisted transformations with the raster datasets, just as it manages other accompanying files. The data management functions, such as Copy and Paste in ArcCatalog, Export Raster Data in ArcMap, and the Copy Raster tool, can persist the transformation with the raster dataset. If the input raster dataset contains a persisted transformation, the output raster dataset will also contain a persisted transformation, whether it is a file-based raster dataset or a geodatabase raster dataset.
- If you use the geoprocessing tools (Raster To Geodatabase, Raster To Other Format, or Copy Raster) to load a TIFF that has RPCs into a geodatabase, the raw pixels will be loaded as a raster dataset and the RPC transformation will be stored with the raster dataset.
- If you export a raster dataset with a transformation using the geoprocessing tools (Raster To Other Format or Copy Raster), the raw pixels will be exported to a raster dataset and the transformation will be written to an .aux.xml file that is managed by the raster dataset.
A raster catalog in ArcGIS 9.2 can store raster datasets with their own spatial references, which can be different from the one defined in the raster column. A raster catalog can also store raster datasets with a persisted transformation. The transformations or spatial references of the raster datasets will be applied on the fly when the raster catalog is displayed in ArcGIS. For example:
- The images of a whole state, such as Texas, cover many Universal Transverse Mercator (UTM) zones, and those images have different spatial references. You can load all the images into one raster catalog and preserve the spatial references with the raster datasets without having to reproject them.
- QuickBird and Space Imaging deliver oriented images, which contain RPC information along with the images. You can create a raster catalog that stores the raw images, as well as the RPC transformation, with the raster datasets.
When you export a raster dataset that is within a raster catalog, it will retrieve the raster dataset transformation information both before and after the transformation is applied.
Similarly to raster catalogs, mosaic datasets can manage many raster datasets in various spatial references. There is a single spatial reference used by a mosaic dataset. Each raster will be reprojected on-the-fly to this same spatial reference when displaying the mosaicked image. Additionally, any footprints or other features will be generated using the spatial reference of the mosaic dataset, regardless of the source raster's spatial reference.