This document will provide an overview of the workflow and best practices for managing multidimensional raster datasets and publishing them as image services, either on-premise or in the cloud.
There are three approaches to managing multidimensional data in ArcGIS.
Use native multidimensional formats (netCDF, GRiB, HDF)—For a single, relatively small multidimensional dataset on desktop, users can work with the multidimensional dataset in ArcGIS Pro in its native format (netcdf, grib, hdf) as a multidimensional raster layer. Using the multidimensional tab, users can explore different variables and dimensions of their data. These datasets can also be published and shared as image services, which can be visualized using the multidimensional filter in Map Viewer.
Convert files to CRF or transposed CRF—NetCDF, GRIB, HDF file formats are generally slow to read. Converting them to a transposed CRF (using the Copy Raster or Build Multidimensional Transpose geoprocessing tools) will drastically improve performance. For example, creating a temporal profile of a large dataset from a transposed CRF can take less than a second, compared to almost a minute with traditional multidimensional formats.
Use mosaic datasets—When working with collections of multidimensional datasets, or with datasets with multiple variables and slices which need additional management options, the data can be ingested into a mosaic dataset using multidimensional raster types.
If you plan to publish multidimensional data as an image service, there are a few options:
- Publish individual files directly from their native format
- Convert individual files to CRF or transpose CRF, then publish them
- Manage the native format files in a mosaic dataset, then publish the mosaic dataset
- Manage the native format files in a mosaic dataset, export the data as a CRF, then publish
Things to consider when choosing the best options for your workflow are detailed in considerations for managing multidimensional data.