Tutorials

CDTS covers a lot of ground — from classic pixel-based change detection to deep learning and cloud data acquisition. Pick the group below that matches what you're trying to do; each tutorial is self-contained and covers theory, data prep, a full code walkthrough, and validation against reference implementations where one exists.

• Trajectory & Change Detection

  ---

  Pixel-based statistical algorithms that model a time series' trajectory to flag disturbances, breakpoints, and land-cover change.

  • LandTrendr
  • CCDC
  • Tmask
  • BFAST
  • BFAST Monitor
  • BFAST Lite

• Time-Series Analysis

  ---

  Tools for comparing, clustering, and extracting statistical patterns from time series — independent of any particular change-detection algorithm.

  • TWDTW (Time-Weighted DTW)
  • Phenology Extraction
  • Mann-Kendall Trend Test
  • SNIC Segmentation
  • SOM (Self-Organizing Maps)

• AI & Deep Learning

  ---

  Native PyTorch architectures for per-pixel classification, spatio-temporal segmentation, and bi-temporal change detection, plus foundation-model transfer learning.

  • Overview
  • LTAE & LightTAE
  • UTAE
  • TempCNN
  • Siamese Change Detector
  • GeoFoundationViT

• Data Acquisition & Integration

  ---

  Streaming imagery straight from cloud catalogs into analysis-ready data cubes, without full downloads.

  • STAC & ARD Integration
  • Google Earth Engine (GEE)

• Infrastructure

  ---

  Scaling any of the above from a single machine to a distributed cluster.

  • Parallel & Cloud Processing

Not sure where to start?

• Want to detect disturbances or land-cover change over time? Start with LandTrendr or CCDC.
• Have labeled data and want to train a classifier? Start with AI & Deep Learning.
• Need to pull satellite imagery into a data cube first? Start with STAC & ARD Integration or Google Earth Engine.
• Your area is too large to process on one machine? See Parallel & Cloud Processing.