Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Open Access Battle Damage Detection via Pixel-Wise T-Test on Sentinel-1 Imagery

Ollie Ballinger

TL;DR

Open Access Battle Damage Detection via Pixel-Wise T-Test on Sentinel-1 Imagery presents PWTT, a lightweight, unsupervised method for estimating building damage over large areas using freely available SAR data. By performing per-pixel time-series t-tests across pre- and post-conflict periods and taking the maximum absolute t-value, PWTT yields a damage probability raster without training. Validation on 633,199 manually labeled footprints across 12 cities in Ukraine, Gaza, Syria, and Iraq shows strong discrimination (AUC up to 0.88 in Ukraine and 0.81 in Gaza) and competitive F1 scores relative to deep-learning baselines, with far lower data and computational requirements. The approach enables open Battle Damage Dashboards in Google Earth Engine that integrate population exposure and geolocated social-media corroboration for near-real-time humanitarian context.

Abstract

In the context of recent, highly destructive conflicts in Gaza and Ukraine, reliable estimates of building damage are essential for an informed public discourse, human rights monitoring, and humanitarian aid provision. Given the contentious nature of conflict damage assessment, these estimates must be fully reproducible, explainable, and derived from open access data. This paper introduces a new method for building damage detection-- the Pixel-Wise T-Test (PWTT)-- that satisfies these conditions. Using a combination of freely-available synthetic aperture radar imagery and statistical change detection, the PWTT generates accurate conflict damage estimates across a wide area at regular time intervals. Accuracy is assessed using an original dataset of over half a million labeled building footprints spanning 12 cities across Ukraine, Palestine, Syria, and Iraq. Despite being simple and lightweight, the algorithm achieves building-level accuracy statistics (AUC=0.88 across Ukraine, 0.81 in Gaza) rivalling state of the art methods that use deep learning and high resolution imagery. The workflow is open source and deployed entirely within the Google Earth Engine environment, allowing for the generation of interactive Battle Damage Dashboards for Ukraine and Gaza that update in near-real time, allowing the public and humanitarian practitioners to immediately get estimates of damaged buildings in a given area.

Open Access Battle Damage Detection via Pixel-Wise T-Test on Sentinel-1 Imagery

TL;DR

Open Access Battle Damage Detection via Pixel-Wise T-Test on Sentinel-1 Imagery presents PWTT, a lightweight, unsupervised method for estimating building damage over large areas using freely available SAR data. By performing per-pixel time-series t-tests across pre- and post-conflict periods and taking the maximum absolute t-value, PWTT yields a damage probability raster without training. Validation on 633,199 manually labeled footprints across 12 cities in Ukraine, Gaza, Syria, and Iraq shows strong discrimination (AUC up to 0.88 in Ukraine and 0.81 in Gaza) and competitive F1 scores relative to deep-learning baselines, with far lower data and computational requirements. The approach enables open Battle Damage Dashboards in Google Earth Engine that integrate population exposure and geolocated social-media corroboration for near-real-time humanitarian context.

Abstract

In the context of recent, highly destructive conflicts in Gaza and Ukraine, reliable estimates of building damage are essential for an informed public discourse, human rights monitoring, and humanitarian aid provision. Given the contentious nature of conflict damage assessment, these estimates must be fully reproducible, explainable, and derived from open access data. This paper introduces a new method for building damage detection-- the Pixel-Wise T-Test (PWTT)-- that satisfies these conditions. Using a combination of freely-available synthetic aperture radar imagery and statistical change detection, the PWTT generates accurate conflict damage estimates across a wide area at regular time intervals. Accuracy is assessed using an original dataset of over half a million labeled building footprints spanning 12 cities across Ukraine, Palestine, Syria, and Iraq. Despite being simple and lightweight, the algorithm achieves building-level accuracy statistics (AUC=0.88 across Ukraine, 0.81 in Gaza) rivalling state of the art methods that use deep learning and high resolution imagery. The workflow is open source and deployed entirely within the Google Earth Engine environment, allowing for the generation of interactive Battle Damage Dashboards for Ukraine and Gaza that update in near-real time, allowing the public and humanitarian practitioners to immediately get estimates of damaged buildings in a given area.
Paper Structure (19 sections, 8 equations, 13 figures, 5 tables)

This paper contains 19 sections, 8 equations, 13 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Building Damage Assessment using Deep Learning
  3. Building Damage Assessment in Synthetic Aperture Radar Imagery
  4. Methods
  5. Data
  6. Synthetic Aperture Radar Imagery
  7. Annotated Building Footprints
  8. The Pixel-Wise T-Test
  9. Damage Classification
  10. Results
  11. Building-Level Damage
  12. Damage Spillovers
  13. Damage Intensity
  14. Discussion
  15. Appendix
  16. ...and 4 more sections

Figures (13)

  • Figure 1: Time-series Sentinel-1 Backscatter Amplitude of a Damaged Building in Mariupol, Ukraine
  • Figure 2: Damage Prediction and Accuracy Assessment, Mosul
  • Figure 3: Country-level ROC and Precision-Recall curves
  • Figure 4: Predicted and Observed Damage, Rubizhne
  • Figure 5: Predicted and Observed Damage Intensity, Gaza
  • ...and 8 more figures