Eroding Coastlines: A Geospatial & Computer Vision Analysis
summary
Uses Planet satellite imagery and runnable Colab notebooks to show a reproducible pipeline: reading multispectral data with rasterio and numpy, computing NDWI, applying morphological filters (OpenCV) to clean masks, then Canny edge detection to find the shoreline. Quantifies loss as about 13 km² over seven years and ~2.5 km coastline retreat (~356 m/yr). Clouds and shallow features cause errors but the notebooked workflow and thresholds produce clear, measurable recession and velocity estimates.
AI-written summary · about
chapters (21)
- 0:00Coastal erosion analysis intro
- 2:52Why coastal erosion matters
- 4:39Planet’s satellite imagery
- 8:13Geospatial concepts explained
- 11:46Resolution in satellite data
- 14:40Python geospatial tools
- 15:55Colab workshop setup
- 17:00Reading data with Rasterio
- 23:40Extracting satellite bands
- 24:58Image metadata and projections
- 31:54NDWI for water detection
- 35:27Land-water masks creation
- 41:30Pixel histograms analysis
- 46:13Q&A break
- 47:19Coastal erosion case study
- 51:57Multi-year data processing
- 1:00:06Measuring land recession
- 1:19:21Edge detection for coastlines
- 1:24:19Policy implications
- 1:28:52Further resources
- 1:33:31Q&A and contacts
related talks
19mDigital Earth Pacific DEP cloud native approaches to detect mineral extraction
31mFrom Radar Echoes To Real-Time Recognition: Deep Learning on Another Planet -An Open Source Workflow
24mNatural Disaster Damage Assessment from Satellite Imagery
17mFree satellite imagery and DEM analyses enabling NRM in the developing world
33mUsing EOLearn to build a machine learning pipeline
50mRestoring Coral Reefs at Scale: AI and the Next Generation of Conservation
45mFrom Radar Echoes To Real-Time Recognition: Deep Learning on Another Planet an Open Source Workflow
18mMapping reef islands in 3D