Skip to content

Key findingsΒΆ

  1. Directional increase-only scoring sharply reduced raw false positives because construction is expected to raise radar response.
  2. Candidate/polygon workflows avoided the whole-AOI base-rate problem and produced a reviewable queue.
  3. Durability filtering helped remove snow/vegetation/moisture transients.
  4. Coherence fusion was the strongest optional recall lever in current validation.
  5. Sentinel-2 fusion helped recover vegetation-to-built changes but remains cloud/season dependent.

For small Norwegian buildings near the Sentinel-1 GRD pixel scale, intensity-only recall can plateau around 0.3-0.5. That is a sensor/signal limit as much as an algorithm limit.