Chapter 4 - Raster Data Model

If you are going to use GIS you need to know there are fundamentally only 2 types of data. Vector data and raster data. Chapter 3 talked about vector data, and vector data is lines, points, and areas which are called polygons. Vector data represents objects that have start and end points like streets, ponds, cities, etc. Raster data on the other hand represents continuous data like elevation which needs to be recorded in a grid system where every part of the map can be represented with a different value.  

Raster is easier and better if you want to analyse your data which if you are learning about GIS that is what you want to be able to do.

So first you need to understand that a raster data map is a grid with rows and columns that make up cells. Cell one is located in the upper left. The types of numbers that get put into each cell are floating-points which is just a decimal digit, and integers which as you know means any whole number. Say you want to make map that you can analyse that has land dived up by categories. You can assign an integer to represent each category.

Raster cell's have a size which gives a map resolution. The higher the resolution the smaller the cells. A satellite image is actually a raster map. The pixels of the image represent raster cells. Raster graphs include actual pictures from satellites and even jpeg images.

An issue that exists with raster data cell's is what if in the cell there exists multiple spatial data. Like what if in the cell there was apple trees and there was wheat. For which attribute would you assign the cell? There are three solutions (they are more like compromises). First, majority rules = which ever attribute is in majority gains the cells. Second, attribute priority = if you care more about apple trees than if it is even a little in a cell that entire cell will represent apple trees. Third, boarder = just make ever cell that shares attributes a border area of no-man's land.

DOQ or digital orthophotoquad are very cool because they are aerial photos taken by a plane with a special camera on that takes so many pictures and complies them together producing a map viewing every area from a straight on 90 degrees.

DEMs or digital elevation models  are raster maps that display uniformly spaced elevation data. LiDAR is a light detection and ranging remote sensing technology that acts as a radar to see how far away things are producing a DEM.

DRGs or digital raster graphics are merely just scanned topographic maps.

Satellite images are raster maps.

The structure of raster data writing:

Raster data is written using cell-by-cell encoding. Which creates a written grid system. For multiple bands, multiple grid systems are written and compiled under band headings. Another way to write raster data is run-length encoding. This method only stores data for certain numbers in each row. By doing so files size is reduced. A final raster data structure is the quad tree. Imagine a tree representing the gird of a map. The stem stands alone representing the entire graph. From there divide the map into four. keep dividing each subgroup into 4 until you can not because all the data in a subgroup is the same. This is used for both vector and raster data.

Importing raster data:

In order to import raster data from say a DEM or satellite image you need information that tells your GIS program what you are working with. This information, found in the header file includes number of rows and columns and bands. Number of pixels. Basically information about the raster data structure, cell size, and values for no data.

Another aspect of importing is files size. Raster data files are relatively big but there are ways to reduce file size. Lossless compression allows the image to be precisely reconstructed where lossy compression allows smaller files to be made but the original image can not be regenerated. JPEG is an example of Lossy compression. JPEG's divide the image in 64 blocks. Now wavelet transformation is allowing lossless compression with lossy file sizes.

Rasterization converts vector data into raster data! This process first creates a raster grid that covers the entire vector graph. At this point all cell values are 0. The second step assigns numbers to all points, lines, and polygons. The third step fills polygons with their values.

Vectorization converts raster data into vector data! This process first creates line thinning. Second the process determines where lines begin and end. Third connects lines and shows where the errors are for correcting.