In this chapter, the recipes are presented in a step-by-step workflow that you may apply while working with a raster. This entails loading the raster, getting a basic understanding of the raster, processing and analyzing it, and delivering it to consumers. We intentionally add some detours to the workflow to reflect the reality that the raster, in its original form, may be confusing and not suitable for analysis. At the end of this chapter, you should be able to take the lessons learned from the recipes and confidently apply them to solve your raster problems.

Before going further, we should describe what a raster is, and what a raster is used for. At the simplest level, a raster is a photo or image with information describing where to place the raster on the Earth's surface. A photograph typically has three sets of values: one set for each primary color (red, green, and blue). A raster also has sets of values, often more than those found in a photograph. Each set of values is known as a band. So, a photograph typically has three bands, while a raster has at least one band. Like digital photographs, rasters come in a variety of file formats. Common raster formats you may come across include PNG, JPEG, GeoTIFF, HDF5, and NetCDF. Since rasters can have many bands and even more values, they can be used to store large quantities of data in an efficient manner. Due to their efficiency, rasters are used for satellite and aerial sensors and modeled surfaces, such as weather forecasts.

There are a few keywords used in this chapter and in the PostGIS ecosystem that need to be defined:

• Raster: This is the PostGIS data type for storing raster files in PostgreSQL.
• Tile: This is a small chunk of the original raster file to be stored in one column of a table's row. Each tile has its own set of spatial information, and thus is independent of all the other tiles in the same column of the same table, even if the other tiles are from the same original raster file.
• Coverage: This consists of all the tiles of a single raster column from one table.

We make heavy use of GDAL in this chapter. GDAL is generally considered the de facto Swiss Army knife for working with rasters. GDAL is not a single application, but is a raster-abstraction library with many useful utilities. Through GDAL, you can get the metadata of a raster, convert that raster to a different format, and warp that raster among many other capabilities. For our needs in this chapter, we will use three GDAL utilities: gdalinfo, gdalbuildvrt, and gdal_translate.