Go back

Urbanization in Far Northwest San Antonio, 2013-2022
AKA What Can Go Wrong in RS Analysis

~ / Portfolio / 2022 /
An clip of this project's PowerPoint presentation, showing the transition from the original project slide to the retrospective slide.

Overview

This project served as my term project for the Principles of Remote Sensing course at UTSA (GEO 4093). Within this course, I was surrounded by classmates in the earth sciences—I was the lone computer scientist in the room. As such, when we were presented with the opportunity to put together a large project, I found it difficult to pinpoint a study subject. However, drawing from my years of residing in San Antonio, I decided to perform an analysis of urbanization in far northwest San Antonio—using the RS techniques from the course.

I came up with a few analysis methods that should have shown changes in the extent of built-up areas (visual analysis, NDVI, NDBI, surface temperature, band transformations); and after combing through the dataset of Landsat 8 images and defining the boundaries of my study area in ArcGIS, I began to put together some data to analyze. Through that process, I faced a number of issues—from the need to manually adjust road lines to put together the study area shape file, to filtering out the invalid values found in the USGS-processed satellite images.

But after processing everything, I faced the worst outcome that any researcher can face—an inconclusive result. Hours of work went towards a result that, on its face, made no sense. Numbers and changes over time seemed almost random, with no discernible pattern.

Faced with this dilemma, I chose to make the best of it.

I began to put together a second slide: What Can Go Wrong in RS Analysis. I pointed out a range of issues that I found in retrospect, from the environmental variables affecting data accuracy, to the limitations of my small dataset, to some fundamental issues regarding the measuring of urbanization. And since this was my final project for this class, I ended it on an optimistic note, pointing out that RS analysis is “really hard”—but with a careful and thoughtful approach, it's possible to put together some solid research.

Methods used

Skills

  • Database retrieval skills, in order to gather Landsat images from USGS; gather TIGER/Line road path files to draw the study area; and perform basic research of external studies to determine my study methods.
  • GIS techniques, in order to create a shapefile for the study area from a dataset distributed by the U.S. Census Bureau.
  • Data and statistical analysis techniques, in order to determine the lack of a consistent pattern across all study years, as well as to determine possible root causes for the inconclusive result.
  • Remote sensing knowledge and techniques, in order to determine possible study methods (e.g., NDBI to measure changes in built-up areas), and to understand how to possibly account for environmental variables.
  • Graphic and layout design techniques, in order to create a presentable, attractive poster to present to the class; while also incorporating a cool transition animation to perform my bait-and-switch.

Software, tools, and libraries

ENVI program icon
ENVI 5.6

Geospatial imagery analysis software

Website
Microsoft PowerPoint program icon
Microsoft PowerPoint

Presentation software

Website
ArcGIS Pro program icon
ArcGIS Pro 3.0

GIS software

Website

Project files

Download term project poster (.pdf) View project files (OneDrive)

Media credits

Article thumbnail was created by me and was derived from TIGER/Line shapefiles published by the U.S. Census Bureau, projected on top of the Dark Gray Canvas basemap published by Esri. Presentation poster and ENVI icon created by me. Microsoft PowerPoint icon created by Icons8. ArcGIS Pro icon is created by me and is derived from "Earth 1", from the Streamline Bold icon pack designed by Streamline.