
This blog post is part two of a four-part series, “Machine Learning for Change Detection”.
Time sliders not yet supported on mobile. Click [Full Screen] to view on mobile devices.
In part one, we made a change detection map for an area of 6,750 sq. mi. using Planet’s satellite imagery.
For part two, let’s zoom in and use time sliders to visualize changes to specific areas in the Austin metro area.
A Closer Look at Austin, Texas
Change Detection Example #1: Georgetown and Buda
What do these cities have in common?
They are both near the edges of the greater Austin metro area, and they are both seeing rapid urbanization.
These time sliders allow us to quickly visualize what changed between 2017 and 2022.
Click the “Toggle Diff” button to turn on and off the change map.
Change Detection for Georgetown, Texas from 2017 to 2022. Satellite imagery: Planet Labs PBC. [Full Screen]
Change Detection for Buda, Texas from 2017 to 2022. Satellite imagery: Planet Labs PBC. [Full Screen]
Each green spot shows land cover changes from 2017 to 2022.
The main changes are:
- New subdivisions
- New infrastructure, such as roads and large buildings
- Cropland shifting locations
Takeaway 1: Cities in the greater Austin metro area are growing mostly in the form of new suburbs and functional infrastructure.
Takeaway 2: Since both rasters are from May, the shifting cropland suggests standard crop rotation rather than a loss of crop yield.
Change Detection Example #2: Colorado River
The water level at Lake Travis, a portion of the Colorado River west of Austin. Satellite imagery: Planet Labs PBC. [Full Screen]
The green highlighted sections of the bank have changed between 2017 and 2022.
Key changes include:
- “Sometimes Island”, in the middle of Lake Travis, is visible in 2022 and not in 2017.
- New water-access infrastructure, including docks, was built.
Takeaway 1: The overall water level has decreased, by several meters at some points, since 2017.
Takeaway 2: Water-access infrastructure continues to be built despite the decreasing water level.
Note: Changes in water level could be related to increased drought in the Austin area since 2017.
Change Detection Example #3: Pflugerville Neighborhood
This neighborhood in Pflugerville, Texas was built between 2017 and 2022.
A neighborhood in Pflugerville, Texas. Satellite imagery: Planet Labs PBC. [Full Screen]
In 5 years, the land cover transformed from bare land into a large-scale housing development.
A Machine Learning Approach to Time Sliders for Change Detection
Open Source Machine Learning tools and Open Source GIS tools helped us analyze our data:
Using the same change detection modeling workflow from part one and a single NVIDIA A100 GPU, we classified land cover and created change maps to display as time sliders.

The Open Source Machine Learning model classifying land cover changes.
What’s Next in this Series
In the next part of this series, we’ll predict the future of Greater Austin development with a cross-comparison of Planet’s satellite data and Regrid’s parcel data.
Join the Conversation
Which time slider had the most changes?
Let’s map more together.
Join the conversation on Twitter @makepathGIS, or email us at contact@makepath.com
Popular
Machine Learning for Change Detection: Part 1
GPU-Enhanced Geospatial Analysis
Open Source Machine Learning Tools (Updated for 2023)
Getting Started with Open Source (Updated for 2023)
The History of Open Source GIS: An Interactive Infographic (Updated for 2023)
Superpowered GIS: ESRI’s ArcGIS + Open Source Spatial Analysis Tools.
Seniors at Risk: Using Spatial Analysis to Identify Pharmacy Deserts
Open Source Spatial Analysis Tools for Python: A Quick Guide (Updated for 2022)