Teaching: Remote Sensing, GIS and
Google Earth Engine
To date, I have been proud to teach about GIS software and Google Earth Engine on multiple occasions. At the University of Salzburg I first tutored, then taught a Bachelor level course on "Remote Sensing Image Processing and Analysis".
During my time teaching the course, I focused on making it practical, applied to real-life scenarios and hands-on. Every week, students learnt about satellite image analysis through practicing the techniques themselves in ArcGIS or QGIS or Google Earth Engine! I also held a Google Earth Engine workshop at a special faculty visit by PhD and Master students from another university.
​
Some of my teaching material and impressions below:
Introduction to the Course
This is an animated StoryMap. I used it as a presentation during the first lecture of the course, but it can also be explored independently. Scroll down inside window below, or view in a new window here.
Lecture Material Samples
These samples include the slides I used for a few of the lectures for Orthorecification, Georeferencing, Image Calibration, Spectral Indices, Object Detection etc.
Orthorecification of WorldView-2 Images to correct Terrain-based Distortions:
Satellite Image Object Detection using Deep Learning Models:
Image Calibration (from Digital Number to At-Sensor to Top-of-Atmosphere):
Computing Spectral Indices and Understanding how Image Calibration matters:
An Introductory Session for Google Earth Engine
Tutoring: Remote Sensing & Image Analysis
3 Different Environments​
​
Historically the course I joined as a tutor was based on ArcGIS Pro. In my first semester as tutor, we worked on diversifying the software and processing environments to also include open source and cloud processing environments.
​
-
ArcGIS Pro: A commercial, off-the-shelf, software package.
-
QGIS: An open-source GIS software.
-
Google Earth Engine: A cloud-processing platform.
The hands-on sessions are designed to switch between these three environments to give students an introduction to each and to help them understand how to use each to perform image analyses.
​
An introduction to Google Earth Engine and band combinations
​
GEE can be daunting, so for a first introduction to the platform, I made sure to keep the coding to a minimum while still making sure everyone in class could get an exciting result to see how powerful cloud processing can be.
​
With just 60 neat lines of code, students could set their own point of interest and visualise 3 million km2 of Landsat 8 imagery in RGB, color infrared, of short-wave infrared.
Check out the PDF (opens in browser)
Spectral Profiles, Burn Ratio Index, the difference between Raw and Atmospherically Corrected Images in Google Earth Engine​
​
For an extended hands-on session, I prepared a real-life applied case study (the wildfires in Australia in 2019) to ​see how we can use satellite imagery in GEE to create spectral profiles for burnt and healthy vegetation and how to apply UN-SPIDER's normalised burn ratio index.
​
Pixel-Based SVM Image Classification ​
In preparation for a classification assignment, I created a simple hands-on exercise to classify a Sentinel-2 image into vegetation and non-vegetation classes. The accuracy assessment is then done against a prepared shapefile (created with the NDVI and a threshold of 0.2). By using only two classes, the focus was on the tools, techniques, the do's and the don't's of image classification.
Check out the PDF (opens in browser)
Searching for Satellite Images in Online Portals
​
Being able to find and access relevant satellite data is one of the most important skills that should be acquired. In the first assignment, I compiled four different real-life case studies from which students could choose to embark on a satellite image search for.
Check out the PDF (opens in browser)
