A seamless workflow for wildfire modelers to validate dynamic global vegetation model outputs against remote sensing data

Abstract. Validation of wildfire simulations remains a persistent challenge in dynamic vegetation and Earth system modeling due to structural inconsistencies between model outputs and satellite-derived observation datasets. Researchers often rely on ad hoc, non-reproducible preprocessing workflows to bridge the gap between different projections and resolutions. This study presents a structured, open-source R-based workflow for processing dynamic vegetation model outputs and systematically comparing them with remote sensing burned area products. The workflow provides an end-to-end solution that (i) aggregates burned area from remote sensing data, (ii) harmonizes sinusoidal HDF files into WGS84-based model formats, and (iii) performs regional-scale comparisons across user-defined spatial units. While demonstrated using LPJ-GUESS and MODIS MCD64A1, the workflow employs a model- and dataset-agnostic "Master Grid" approach. This creates a static template to strictly harmonize observational data, thereby eliminating spatial mismatch errors while ensuring reproducibility.