A comprehensive assessment of the FireCCILT11 global burned area product

• FireCCILT11 shows low accuracy and consistency, especially before 2001. • Annual burned area from FireCCILT11 often diverges from national fire records. • Per-pixel analysis reveals high omission and commission errors in FireCCILT11. • A pixelation effect reduces FireCCILT11′s effective spatial resolution to 0.25°. • FireCCILT11 is unsuitable for long-term fire regime studies in many regions. The FireCCILT11 global burned area (BA) product, developed under the ESA Climate Change Initiative, spans nearly four decades (1982–2018) and is the longest nearly continuous global fire record currently available. Despite its growing use in high-profile studies, its performance has not been comprehensively evaluated. Moreover, while two earlier studies identified severe instrumental artifacts in the FireCCILT11 dataset, the supporting evidence in those studies was primarily circumstantial. Here, we directly assess FireCCILT11 BA using a suite of national fire history datasets and independently derived satellite-based BA products across eight countries. Our analysis reveals substantial limitations in both the accuracy and consistency of FireCCILT11, particularly before 2001. FireCCILT11 frequently diverges from national records in terms of annual BA totals, exhibits poor spatial agreement at the pixel level, and suffers from high omission and commission errors. A pixelation effect further compromises its spatial resolution and usability. These findings suggest that FireCCILT11 is generally unsuitable for long-term fire regime analyses, especially in regions or periods not represented in its training data. We highlight the ongoing need for a robust, AVHRR-based global BA product and propose best practices for the development and evaluation of long-term BA datasets.