How Townsville's Flood Archives Ended Up Full of Duplicates — and What It Took to Fix ItUpdated

Townsville City Council's digital records unit is midway through a structured duplicate-image replacement program that archivists say has already cleared more than 14,000 redundant files from the municipality's infrastructure and community photograph collections. The work, which began in earnest in March 2026, targets a problem that accumulated quietly over nearly a decade.

The timing matters because the council is simultaneously preparing its next capital-works submission to the Queensland Reconstruction Authority, a process that requires clean, traceable photographic evidence of pre- and post-flood conditions at dozens of sites across the local government area. Submitting duplicate or mislabelled images to that process has previously delayed reimbursement claims, according to the council's own project documentation circulated to relevant work units earlier this year.

How the Backlog Built Up

The root cause traces back to the February 2019 floods. When floodwater reached Ross River and inundated suburbs from Hermit Park through to Idalia, council field staff and contractors were capturing damage photographs under extreme pressure — often using personal phones, department cameras and hired equipment simultaneously. Files were uploaded to at least three separate systems: the council's legacy Civica Authority platform, a temporary QRA evidence portal, and individual departmental SharePoint folders.

Nobody had time during the emergency to deduplicate. That problem compounded across subsequent wet seasons. By the time the council migrated to its current records management system in mid-2023, staff discovered that single infrastructure sites — a retaining wall on Nathan Street, a stormwater drain near Riverway Drive — sometimes had 40 or more images that were functionally identical, saved under different file names and lodged in different folders. The Queensland State Archives Act 2002 requires councils to maintain accurate, non-redundant records of public assets, which gave the issue a compliance dimension beyond mere tidiness.

The 2024–25 wet season added another layer. Drone surveys commissioned through the council's Resilient North Queensland program generated high-resolution imagery of the Bohle River corridor and low-lying industrial land near the Port of Townsville. That footage, captured across multiple flights between December 2024 and March 2025, was ingested without a deduplication step, replicating the original 2019 problem on a larger scale and in higher resolution.

The Replacement Program in Practice

The current program uses a combination of perceptual hashing software and manual review by two full-time archivists based at Townsville City Council's customer service and records centre on Walker Street. Perceptual hashing detects images that are visually near-identical even if their file metadata differs — catching cases where the same photograph was resaved at a different compression level or cropped slightly before being uploaded a second time.

Where a duplicate cluster is identified, archivists designate one image as the canonical record, attach verified location metadata and a standardised asset reference number, and retire the remaining copies to a quarantine folder before deletion. The program set a target of processing 60,000 image files by 30 September 2026. As of the most recent internal progress report, roughly 38,000 had been assessed.

James Cook University's College of Information and Communications Technology has provided technical advice to the council on the hashing methodology, part of an existing memorandum of understanding between the two organisations around smart-city data governance. The university's Bebegu Yumba campus on Douglas, about 12 kilometres from the CBD, houses the team that helped specify the thresholds the software uses to flag a match.

For residents and contractors dealing with the council, the practical consequence is straightforward. Infrastructure grant applications lodged with the QRA after October 2026 will be expected to reference image IDs drawn from the cleaned archive, which the council says will make evidence packages faster to compile and easier for state assessors to cross-check. Contractors working on roads, drainage and community facility projects should register their site photographs through the council's asset-submission portal rather than emailing files directly to project managers — a practice that was, ironically, one of the main reasons duplicates proliferated after 2019 in the first place.