Raster Analysis
Travis Zalesky
As part of UA GIST 602A
Embedded Files
Figure 1. A digital image, made of pixels.
Introduction
Introduction
Raster analysis is a critical and often overlooked aspect of Geographic Information Sciences (GIS). A raster is a two-dimensional, data set that is made up of a matrix of cells, also known as pixels. Most people are familiar with the concept of pixels used in digital imagery, but they may not know that the same concept can be used to model space (Figure 1). Raster data is particularly useful when modeling phenomenon such as air temperature or elevation, which are (1) continuous, (2) vary smoothly across space, and (3) are widely distributed (or universally applicable). The applications of raster datasets and raster analysis are too numerous to count, but they require specific methods and modes of thinking which are often challenging or counterintuitive. As such, raster datasets are often ignored or underutilized in geospatial analysis.
In this exercise, I will demonstrate common raster analysis tools and workflows in three parts. Parts 1 and 2 will focus on a case study in the Olympic Peninsula, WA. First, I will demonstrate the creation of a Watershed Model for the Olympic Peninsula and SW Washington. Secondly, I will focus my analysis on the Skokomish River Watershed in the SE of the Olympic Peninsula, and will demonstrate how rasters can be used to map complex phenomena, like Flood Risk. Lastly, I will present two unrelated mini-projects which will demonstrate use of focal statistics to identify complex land forms, and generation of a cost-surface for path optimization.
Using a high resolution Digital Elevation Model (DEM) of the Olympic Peninsula I will demonstrate the creation of a hydrologically conditioned watershed model. I will break the workflow up into three logical steps, (1) preprocessing, (2) conditioning, and (3) watershed modeling. I will discuss the methods used and the importance of each step, as well as various use cases and limitations of watershed models.
Using the hydrologically conditioned watershed model created in Part 1, I will present a focused analysis of flood risk within the Skokomish River watershed, located in the SE of the Olympic Peninsula. I will demonstrate how known risk factors can be individually mapped using raster data. I will then reclassify risk factors to a common unitless scale, and show how raster math can be used to map complex phenomena. Furthermore, I will detail the methods used throughout and will discuss the results of my findings within the Skokomish River watershed, as well as the limitations of my risk model. I will also discuss additional use cases and the broader applicability of raster math generally.
In this section, I will present two mini-projects completed as part of my UA GIST 602A curriculum. These projects demonstrate use of several raster techniques not shown elsewhere in this project page. The first project demonstrates how focal statistics can be effectively utilized to identify specific land formations across a complex landscape. The second project illustrates how to effectively create a cost surface for path optimization.
Acknowledgements
Acknowledgements
I would like to thank Professors Dr. Matthew Marcus and Dr. Fernando Sanchez-Trigueros for their excellent curriculum. I would especially like to thank Professor Marcus for testing my knowledge and challenging me to do better. Thank you to all my peers, and everyone working in the Geographic Information Sciences Technology program at the University of Arizona School of Geography, for allowing me the opportunity to study raster analysis. Finally, thank you to anyone who reads or interacts with this webpage. I hope that you have learned something. Let me know if you have any questions or comments.
Page updated
Google Sites
Report abuse