Understanding raster data

The other popular method of storing geospatial data is through a process called rasterization. Raster data is a grid of cells organized into columns and rows called pixels. Each cell can hold a single color or value which can represent all kinds of information, from the color of light that are reflected off the surface, or an elevation above sea level, or how many inches of rain fell in January. And as you stack these cells together into a large gridded surface, they cover an area of the earth, just like a giant chessboard. The most common raster formats that you will work with is an areal photograph.

And they're exactly what you expect them to be. Photographs of the earth taken from a satellite or an aircraft. Aerial images go through a process called orthorectification, which eliminates most of the distortion that comes with the perspective and camera lenses and tags the image with location attributes, so that it correctly aligns on the GIS. The grid of cells each represents an area of the earth's surface, and takes a value that represents the average return for the covered area. The resolution of a raster image is described by the area that each pixel covers. For instance, you'll often hear of a raster image as being something like a ten meter image, which means that each pixel covers an area of the earth that is ten meters wide by ten meters tall.

So, the smaller the area, the more detail you have to work with in the image. Some orbiting satellites return images that are 30 meters or more, and some highly detailed photographs from airplanes come out in a sub meter resolution, measuring just a few inches per pixel. If you've ever worked with digital photographs, then you might be familiar with the red, green, and blue channels that make up the composite color picture. In the world of remotely sensed data, we work with that same kind of idea, where different wavelengths of energy are captured in different data bands. Single band data is represented as a black and white image.

For a natural color image, this would be the same red, green and blue wavelengths that your camera sees and requires a combination of data from three separately tuned sensors, one for each primary color. But aerial images typically include much more information by including data from sensors that look outside the visible spectrum of light, and are tuned to gather information in very specific wavelengths of energy. These get saved into the images as different databands or components of the image. Scenes or images from the new Landsat A satellite include 11 different bands of data, including three standard color bands and add a host of additional bands aimed at specific scientific analysis including infrared, which is useful for studying the health of forest and crops, and thermobands useful for looking and tracking wildfires and volcanic activity.

It even includes bands that are designed to just locate high altitude cirrus clouds and identify coastal aerosols in the atmosphere. Air photos represent just one kind of raster data though. Raster data can also included scanned paper maps or the direct digital descendants such as nautical charts and topographic maps, or other types of continuous surfaces such as elevation or rainfall data. In the case of continuous data, the same resolution designations apply. For instance in the case of a ten meter elevation raster, it means that every pixel still covers an area that is ten meters wide by ten meters tall, but instead of a color value in the case of a photograph, the pixels value represents the average height of that ten meter by ten meter area.

These types of data typically only include a single band of information and get drawn on the screen as a grayscale representation where black represents low values in the dataset, and white represents high values in the dataset. In arc map though, we can choose to symbolize these datasets in a variety of ways including applying a color gradient or using other classification or rendering techniques. Since raster data are organized as digital images, you might already be familiar with some of their common file extensions. They typically use the .tiff, .jpg, or .png file extensions, but there are some other specialized file formats that you'll come across.

For instance the U.S. geological survey elevation data comes in a .dem format which stands for digital elevation model. In addition to the files that store all of the pixel values, you might also see what are called Sidecar files, additional files that share the same name as the image but store location values or other information about the image and have extensions like .prg, for projection, or .ovr for overview. Just like with vector files, these file groups need to stay together in order to maintain the integrity of the data source. So, that's a brief overview of the raster formats you are likely to run across.

Let's take a look at a few of those in our catalog in the next movie.