Climatic Correlates of Wildfire Burned Area in Six Turkish Provinces (2016–2025): Source-Verified Records and Distributional Sensitivity

Background: Wildfire burned area reflects weather, fuels, topography, and response conditions, but incident-level associations are sensitive to data linkage and temporal measurement support. Methods: We analysed administrative fire records from six Turkish provinces during 2016–2025, reconciled burned area to original event identifiers, audited monthly station provenance and spatial locations, and compared positive-area Gamma and zero-inclusive Tweedie/two-part models with geographic sensitivity analyses. Results: The fire ledger contained 364 events, of which 363 had climate covariates; 357 previously misassigned burned-area values were corrected against the source register. In 332 positive-area fires, monthly minimum relative humidity was positively associated with recorded area in a Gamma model (mean-area ratio 1.057, 95% CI 1.016–1.100; nominal p = 0.007; four-predictor FDR p = 0.026), although explained deviance was only 2.8%. A zero-inclusive Tweedie model (n = 363) yielded a smaller association (ratio 1.043; nominal p = 0.015) that was not significant after FDR adjustment (p = 0.058). Three hundred fifty-three wildfire positions passed preliminary screening, but original weather-station allocation and fire-day humidity remain unverified. Conclusions: Limited explanatory performance and monthly exposure aggregation preclude causal or operational fire-risk interpretation. Auditable incident-level source linkage is essential because outcome misalignment can reverse an apparent climate association.