A new openly available dataset for Türkiye, compiled by Bikem Ekberzade of Istanbul Technical University’s Eurasia Institute of Earth Sciences and published as a preprint under review in the journal Earth System Science Data, confronts that flaw head-on. The work presents a geographically validated wildfire inventory covering the years 2013 to 2025, built from official records maintained by Türkiye’s General Directorate of Forestry, known by its Turkish abbreviation OGM. Rather than accepting the archive at face value or quietly deleting its problematic entries, the study subjects every record to a reproducible five-stage validation workflow designed to identify, reconcile, and document spatial inconsistencies while preserving the statistical integrity of the underlying fire events.

The scale of the problem is easy to underestimate. Administrative fire records are typed by hand, often under time pressure during active fire seasons, and they accumulate the full range of human error: transposed digits in coordinates, misspelled district names, coordinates that fall in the wrong province, and mismatches between the administrative location recorded in text and the GPS position attached to the file. A single misplaced decimal point can shift a fire by tens or hundreds of kilometers, dropping a Mediterranean blaze into the sea or relocating an Aegean ignition to the Anatolian interior. When such records feed into fire regime analyses, climate impact assessments, or the validation of satellite products, the errors propagate silently, biasing regional statistics in ways that no amount of downstream modeling can repair.

The first stage of the new workflow tackles the textual layer of the archive through administrative harmonization. Place names recorded in the forestry archive are standardized and reconciled against official administrative geographies, so that a fire logged under a variant spelling or an outdated administrative designation can be matched to its correct district. This step matters because Türkiye’s administrative boundaries have been reorganized over time, and a record that references a historical district name cannot be spatially verified until it is mapped onto the current administrative framework. Once the text is harmonized, each fire gains a candidate location that can be tested against the real world.

The second stage performs spatial verification against official district boundaries. Here the workflow checks whether the recorded coordinates actually fall within the administrative unit named in the record. A fire attributed to a district in Antalya, for example, must plot inside that district’s polygon; if the coordinates land in a neighboring province or offshore, the inconsistency is flagged. This cross-check between the textual and spatial layers is the conceptual heart of the method: it treats agreement between independent sources of information as evidence of validity, and disagreement as a signal that at least one of the two is wrong. Crucially, the workflow does not assume which source is correct, but instead escalates the record to the next stage of reconciliation.

That next stage is where the dataset becomes genuinely distinctive. For records whose coordinates cannot be reconciled with their administrative labels, the workflow turns to satellite-assisted coordinate recovery using NASA’s VIIRS Active Fires and Thermal Anomalies detections. The VIIRS instrument, flying on the Suomi NPP and NOAA-20 satellites, detects hotspots at a resolution fine enough to pinpoint active burning on the day a fire was reported. By searching the satellite detections for thermal anomalies near the reported district and date, the workflow can recover a plausible fire location for records whose original coordinates were unusable. This is a powerful form of data rescue: events that would otherwise be discarded, or worse, mapped at meaningless positions, are restored to the landscape with an explicit indication of how their coordinates were obtained.

The fourth stage applies distance-based border assessment, a conservative test for records whose recovered or reported positions sit near administrative boundaries. Fires do not respect district lines, and a coordinate that falls just across a border may still be the correct location of a fire reported under the adjacent district’s name. The border assessment evaluates such near-boundary cases quantitatively, accepting positions that are plausibly consistent with the reported administrative location while rejecting those that are not. For the irrecoverable remainder, records with no defensible location of any kind, the workflow applies conservative on-land coordinate imputation: the fire is retained in the statistical record, but its coordinates are placed at a deliberately generic on-land position that prevents misleading spatial interpretation while preserving the event’s contribution to national fire counts and burned-area totals.

Transparency is the thread that runs through the entire methodology. Every wildfire record in the inventory carries a provenance flag that states explicitly whether its coordinates are original and administratively validated, satellite-corrected, border-accepted, or conservatively imputed. This design choice has profound implications for how the dataset can be used. A researcher studying the fine-scale geography of fire ignitions can filter to include only the highest-confidence coordinates; a modeler validating coarse-resolution Earth system outputs can retain the full statistical record; and an operational analyst can quantify exactly how much of the archive rests on recovered rather than original positions. Nothing is hidden, nothing is silently corrected, and no record is discarded merely because it was inconvenient.

The scientific payoff extends well beyond Türkiye’s borders. Validated administrative inventories are among the most valuable ground-truth resources available for evaluating remote sensing fire products, which are themselves used to calibrate the fire modules of Earth system models. Satellite burned-area and active-fire datasets must be checked against independent observations, and a thirteen-year, nationally consistent inventory with explicit quality flags provides exactly the kind of reference that such validation requires. The eastern Mediterranean, where Türkiye sits, is one of the world’s most fire-prone and climate-stressed regions, and reliable historical fire data there underpins everything from emissions accounting to ecosystem management to the design of suppression policy. An archive whose spatial errors have been systematically reconciled changes the evidentiary basis for all of that work.

The dataset also arrives at a moment when the scientific community is rethinking how historical records should be treated. The traditional approach, deleting or ignoring records that fail quality checks, trades statistical completeness for spatial accuracy, and often does so invisibly. The alternative demonstrated here, recovering what can be recovered and flagging what cannot, keeps both goals in view and forces users to make explicit, documented choices about which dimension of quality their analysis requires. As climate change intensifies fire regimes across the Mediterranean and beyond, the demand for trustworthy, transparent, and reproducible fire data will only grow. This inventory, openly available and fully documented, offers a template that other national forestry agencies and data providers could follow, turning decades of imperfect administrative bookkeeping into a scientific asset rather than a liability.

Subject of Research: Geographic validation of administrative wildfire records for Türkiye from 2013 to 2025

Article Title: A Geographically Validated Administrative Wildfire Inventory for Türkiye (2013–2025)

Article References: Ekberzade, B. (2026). A Geographically Validated Administrative Wildfire Inventory for Türkiye (2013–2025). https://doi.org/10.5194/essd-2026-701

Image Credits: AI Generated

DOI: 10.5194/essd-2026-701

Keywords: wildfire, Türkiye, dataset, satellite, VIIRS, geospatial validation, Earth System Science Data, fire regime, data rescue, remote sensing, forestry records, Mediterranean