Designing and implementing a GIS using raster data is no different than it is for any other GIS, except that you use raster data instead of, or together with, vector feature data.
When working with raster data, you can use the workflow described below.
Identify the purpose or objective
You can use raster data for display and analysis purposes. Raster data for display is quite common, especially using orthoimagery as a background for a map. Raster data for analysis can be implemented in many ways, such as watershed or terrain analysis, updating topographic features in other datasets, or updating land-cover classes to assess the location of a new housing development.
Identify the data
To extract information from imagery, consider the resolution you require and whether you need one or more spectral bands. Consider whether the data comes from an aircraft or satellite. If you’re going to work with elevation data, consider the most appropriate methods for collection, such as lidar, contour lines, radar interferometry, or generated as an ortho mapping product. If you intend to create a collection of scanned maps, you must identify what those maps are, such as scanned documents, CAD drawings, or topographic maps.
Refine the requirements
Determine more detailed requirements based on the following:
- Cost—What are your budget limits? Can you afford the data you want? Is there an alternative within your budget?
- Availability—Does the data already exist? How often is the data updated? Will you receive updates as individual tiles or a single update with complete coverage? Can you receive this data in a timely manner?
- Licenses—Can you share or distribute this data? Can you use this data in multiple projects? Does the license have any limits? What can you do with the information or data derived from the original data? Can you serve this to the public using the internet?
- Resolution—Will the available level of detail provide the required information? Do the imagery bands provide required information such as an infrared band for vegetation analysis?
- Storage—What database or file formats will be used? How large is each file? Will you use pyramids or overviews? How much total disk space is needed? Do the files need to be on your local machine or stored in the cloud?
- Extent—Can you cover the area of interest with one raster image or will you need multiple raster datasets?
- Accuracy—Will the available data resolution provide you with the required spatial accuracy? What is the level of accuracy promoted by the data vendor? Are ground control points needed? How will the data accuracy be verified and validated?
- Accessibility and pricing—Is the data accessible locally or will it be accessible on a network? Will you charge fees for usage or downloads? Who will have access to the data? How will you control access and sales?
Acquire and review data
This can involve placing orders for the data with a company capable of providing it, scanning the maps you need, or acquiring the source data and building the corresponding raster datasets. It is important that you have a system for checking the quality of the data, whether created in-house or acquired from outside sources. You may need to check for missing data (such as dropped lines or pixels), for poorly represented data, or whether the data is georeferenced for your area of interest.
Prepare the data
Building the database can require the prior extraction or conversion from one data format to another, such as from lidar elevation points to a DEM. It can also involve preprocessing, such as image calibration, histogram enhancement, georeferencing, or rubber sheeting.
Design and build the database
This can involve one of the following choices:
- Build a large, seamless raster dataset (mosaic) from multiple images.
- Build a separate, distinct raster dataset from each source image. (Essentially, each dataset is accessed independently of the others.)
- Build a raster catalog containing all the imagery.
- Build a mosaic dataset that contains links to the separately stored image files, called items.
- Retain the data in separate image files.
Additional considerations include which compression method to use, whether to use a file geodatabase or a multiuser geodatabase management system, and what your data dissemination will include. For example, if you will be serving your imagery, consider a mosaic dataset since it is optimized for this type of dissemination.
You will need to create some level of metadata, depending on your intended distribution and access to the data. For example, what types of queries should users expect to perform to find your raster data over the web? If you're using raster catalogs or mosaic datasets, consider additional catalog fields to allow more extensive querying capabilities.
Deploy and maintain the geodatabase
One of the main reasons for using this loading process is to allow many people to use the data for various purposes and projects. This requires administration and management.
In most situations, you will plan on reusing your dataset or database. You must plan for updates, modifications, and the ability to build on your initial implementation.
Database fragmentation and frequent data manipulation may increase the size of your mosaic dataset dramatically. If your database size is inflated due to constant transactions, you should run the Compact tool.