| Summary: | Current information about the timing, location and characteristics of African biomass burning is primarily derived from two satellite sensors: the geostationary Spinning Enhanced Visible and Infrared Imager (SEVIRI) and the polar-orbiting Moderate Resolution Imaging Spectroradiometer (MODIS). Due to opposing limitations in their sampling design, however, SEVIRI and MODIS cannot independently detect and characterize all active fires from ignition to extinction. Instead SEVIRI (at 15 min temporal resolution) fails to detect relatively smaller and/or lower intensity fires within its coarser 3 km pixels while MODIS (at 1 km spatial resolution) fails to detect fires between the 3hr and 9hr gaps in overpasses. As part of an ongoing validation exercise, this work compares the SEVIRI and MODIS active fire products across a strong gradient of fire activity in the Central African Republic. The evaluation of SEVIRI’s errors of omission and commission were found to depend on MODIS scan angle, and were strongly affected by the SEVIRI cloud detection strategy and uncertainties in the background thermal characterization. , Despite instantaneous and localized differences in the SEVIRI and MODIS active fire products, the agreement between SEVIRI and MODIS derived maps of fire occurrence and the fire season duration increased as active fire pixels were aggregated into coarser ispatiotemporal grids. Beyond direct comparisons, this thesis also explores the ability to enhance active i detection and characterization across Africa by synthesizing the SEVIRI and 10DIS datasets. The first method developed here adjusts SEVIRI observations at 5.0° rid cell resolution and 15-minute temporal resolution to account for the undetected fire liative power (FRP) associated with SEVIRI’s errors of omission. Conversely, the ond method adjusts MODIS observations at 0.5° grid cell resolution and monthly aral resolution to account for the aforementioned undetected FRP emitted between ses. After assuaging a MODIS imaging artefact referred to as the ’bow-tie :," estimates of the fire radiative energy (FRE) emitted from Africa derived from ! two methods herein agree within 4% of each other and are ~1.46x greater than idously published estimates of FRE derived from MODIS. Nevertheless, continental-lie estimates of fuel consumption based on measurements of FRP alone remain ~ 2.5x I than recent bottom-up modelling approaches.
|
|---|
