GEOINT Capstone

Capstone Project — Automated Change Monitoring Pipeline

This is the final integration project. You will build a working, end-to-end pipeline that monitors a location for changes using satellite imagery, then produce an analysis report. This combines every skill from the course: data acquisition, raster processing, change detection, object detection, SAR analysis, and intelligence assessment.

---

Project Requirements

Input

• Location coordinates (lat, lon) + area of interest radius
• Date range (6+ months)
• Monitoring frequency (weekly or per-acquisition)

Pipeline Steps

1. Data Acquisition — Programmatically search and download Sentinel-2 (optical) and Sentinel-1 (SAR) imagery for the location and date range. Use the Copernicus Data Space API. (Tutorial - Acquiring Free Satellite Imagery)

2. Preprocessing

   • Cloud mask Sentinel-2 using SCL band (Tutorial - Working with Raster Data)
   • Speckle filter Sentinel-1 (SAR Fundamentals and Analysis)
   • Clip to area of interest
   • Compute NDVI and other relevant indices (Multispectral Analysis)

3. Baseline Establishment

   • Use first 2-3 cloud-free optical scenes to compute baseline NDVI (mean, std)
   • Use first 2-3 SAR scenes to compute baseline backscatter

4. Change Detection (Change Detection)

   • For each new optical scene: NDVI differencing against baseline
   • For each new SAR scene: amplitude ratio against baseline
   • Morphological cleanup, minimum area threshold
   • Flag significant changes

5. Object Detection (if changes detected) (Tutorial - Object Detection in Satellite Imagery)

   • On changed areas: run ship detection (SAR), or building/vehicle detection (optical)
   • Count and classify detected objects

6. Report Generation

   • Before/after visualization for each detected change
   • Change timeline (area changed per date)
   • Object count time series
   • Structured text report following Case Study - Monitoring Military Installations template

Output

• Change detection map (GeoTIFF with change labels)
• Before/after PNG visualizations
• Timeline plot
• Text report with findings and confidence assessment

---

Stretch Goals

• Maritime extension: Add AIS data correlation for port monitoring (Case Study - Maritime Domain Awareness)
• Airfield extension: Add ADS-B data for aircraft counting
• Terrain: Generate viewshed from key positions near the facility (Terrain Analysis and Geolocation)
• Alerting: Email or webhook notification when change exceeds threshold
• Multi-source fusion: Incorporate OSINT check (automated news search for the location) (Multi-Source Intelligence Fusion)

---

Assessment Gate

To pass the capstone, demonstrate:

1. Working pipeline — end-to-end, from coordinates to report, with real data
2. Analysis report — at least one real change detected and assessed, OR explanation of why no changes were detected and what that means
3. Confidence levels — state confidence and justify it (what sources support/contradict)
4. Limitations acknowledged — what the pipeline misses, resolution limits, false positive rate
5. Code quality — clean, runnable, documented (another analyst can run it with different coordinates)

---

Recommended Workflow

1. Start with a simple version: one location, optical only, NDVI change detection
2. Add SAR when optical works
3. Add object detection on change areas
4. Add reporting
5. Then try stretch goals

The goal is a pipeline that WORKS end-to-end, not a perfect one. A simple pipeline that runs and produces a report is worth more than an elaborate one that’s unfinished.

---

See also: GEOINT Home | Change Detection | Case Study - Monitoring Military Installations