← All posts
🌊

Case Study: 400,000+ Drone Images for the 22 Coastal Towns Program, Bangladesh

11 October 2026 case-study drone photogrammetry pix4d bangladesh
Written by M. Mashrur Zaman • GIS & Photogrammetry Specialist

RTK drone photogrammetry for 15 municipalities in an ADB-funded water supply and sanitation program: 400,000+ images processed into survey-grade orthomosaics, DSMs and DTMs.

Context

The 22 Coastal Towns program is a feasibility study and detailed engineering design for water supply and sanitation in coastal Bangladesh, funded by the Asian Development Bank (ADB) for the Department of Public Health Engineering (DPHE). I delivered the drone mapping through Onushandhani Creeds Limited, covering 20 of the program's 22 municipalities: 15 delivered in full and 5 in part.

Map of the 20 municipalities in the 22 Coastal Towns program mapped by drone, across Bangladesh's southern coastal belt

The challenge

Engineering design needs accurate, consistent base mapping for every town. That meant planning flights, placing ground control and turning a very large volume of RTK drone imagery into survey-grade products, town after town, for an engineering team that would build directly on them.

Approach

  • Flight plans built to the program specification: at least 75% forward and 65% side overlap
  • RTK-enabled UAV acquisition with ground control points for QA/QC and accuracy validation (about 100 GCPs in a single town)
  • Each municipality split into processing zones with 50–70 m overlap between zones, so edges join cleanly and no seams appear in the final mosaics
  • Python scripts to sort every town's images into its zones
  • A distributed Pix4D and Agisoft Metashape workflow spread across multiple computers, built for very large UAV datasets
  • Checks with PDAL, GDAL, QGIS and Python

Results

  • 400,000+ drone images processed across 15 municipalities (zone overlaps included)
  • Survey-grade orthomosaics for every town
  • DSM and DTM production for each town, starting with Gournadi, Banaripara, Bakerganj and Nalchity
  • Contours at 50 cm and spot heights on a 1 m grid for engineering design
  • Built to the program accuracy specification for UAV products: within ±10 cm horizontal and ±15 cm vertical

A closer look: Gournadi

Gournadi shows the workflow end to end. The figures below are drawn from the project's own flight plans, photo positions, GCP survey and digitized layers.

Gournadi in numbers: 20 flight blocks, 27,389 geotagged photos, 100 ground control points, 13 processing zones, 32,484 photos processed, 4,247 ponds and 73 km of road centerlines digitized Map of Gournadi flight plan: 20 flight blocks, photo density and 100 ground control points Map of Gournadi processing zones Z1 to Z13 with the overlap strips shared by neighbouring zones Map of Gournadi roads and water bodies digitized from the orthomosaic

A closer look: Banaripara

Banaripara shows the full set of deliverables, from the orthomosaic to the terrain products the engineers design on.

Banaripara orthomosaic covering the whole municipality, with the detail area marked One 200 m block of Banaripara shown four ways: orthomosaic, DSM, DTM, and contours with spot heights Banaripara DSM with buildings and trees, compared with the DTM showing bare ground

The DTM matters most for water and sanitation design: with buildings and trees removed, it shows how the ground actually slopes. The 50 cm contours and 2.4 million spot heights come straight from it.

What the client said

"Processed over 400,000 UAV images into survey-grade Ortho mosaics, DSMs, and DTMs across 15 municipalities ... on schedule, and to a standard we could hand straight to our clients." — Engr. Md. Shahadat Hossain, Managing Director & CEO, Onushandhani Creeds Limited