MonteCarlo Biphasic Estimation of Fire Properties (McBEF): Part I, Algorithm formulation

Sub-pixel fire characterization is essential for quantifying wildfire energetics, combustion phase dynamics, and their atmospheric impacts from satellite observations. This study presents a series of Observing System Simulation Experiments (OSSEs) to systematically evaluate the influence of temperature phase complexity, channel selection, and intra-phase temperature heterogeneity on the performance of a Bayesian retrieval framework. Using synthetic observations derived from a multi-phase radiance model, we assess retrieval skill across key fire properties—including Fire Radiative Power (FRP), visible energy fraction (VEF), and flaming heat flux—under realistic sensor constraints. Results show that a bi-phasic fire model provides a strong balance between accuracy and observational feasibility, outperforming conventional approaches such as Wooster’s FRP regression, particularly in low-intensity fire scenarios. FRP is found to be the most resilient parameter to model and characterize observational uncertainties, while VEF and heat flux estimates are more sensitive to both fire phase partitioning and spectral coverage. Shortwave and mid-infrared bands, especially the day-night band (DNB) and 2.25 \mum m channels, are shown to offer the greatest constraint on retrieval accuracy. We further analyze the impact of intra-phase temperature variability and propose practical filtering strategies for identifying low-confidence cases, such as smoldering-dominated fires. This paper serves as the first part of a companion study on multi-channel fire retrieval. In the subsequent paper, we apply the bi-phasic Bayesian framework to real VIIRS nighttime observations and assess algorithm performance in an operational setting.